1 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
3 * machmode.h (mode_for_int_vector): New function.
4 * stor-layout.c (mode_for_int_vector): Likewise.
5 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
6 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
7 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
9 (s390_expand_vcond): Likewise.
11 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
13 * machmode.h (opt_machine_mode): New type.
14 (opt_mode<T>): Allow construction from anything that can be
16 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
17 (mode_for_size): Return an opt_machine_mode.
18 * stor-layout.c (mode_for_size): Likewise.
19 (mode_for_size_tree): Update call accordingly.
20 (bitwise_mode_for_mode): Likewise.
21 (make_fract_type): Likewise.
22 (make_accum_type): Likewise.
23 * caller-save.c (replace_reg_with_saved_mem): Update call
25 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
26 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
27 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
28 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
29 * expmed.c (extract_bit_field_1): Likewise.
30 * reload.c (get_secondary_mem): Likewise.
31 * varasm.c (assemble_integer): Likewise.
32 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
35 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
37 * machmode.h (decimal_float_mode_for_size): New function.
38 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
39 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
40 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
41 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
42 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
43 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
45 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
47 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
48 (get_builtin_sync_mode): Likewise.
49 (expand_ifn_atomic_compare_exchange): Likewise.
50 (expand_builtin_atomic_clear): Likewise.
51 (expand_builtin_atomic_test_and_set): Likewise.
52 (fold_builtin_atomic_always_lock_free): Likewise.
53 * calls.c (compute_argument_addresses): Likewise.
54 (emit_library_call_value_1): Likewise.
55 (store_one_arg): Likewise.
56 * combine.c (combine_instructions): Likewise.
57 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
58 * config/arm/arm.c (arm_function_value): Likewise.
59 (aapcs_allocate_return_reg): Likewise.
60 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
61 * config/i386/i386.c (construct_container): Likewise.
62 (ix86_gimplify_va_arg): Likewise.
63 (ix86_expand_sse_cmp): Likewise.
64 (emit_memmov): Likewise.
65 (emit_memset): Likewise.
66 (expand_small_movmem_or_setmem): Likewise.
67 (ix86_expand_pextr): Likewise.
68 (ix86_expand_pinsr): Likewise.
69 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
70 * config/microblaze/microblaze.c (microblaze_block_move_straight):
72 * config/mips/mips.c (mips_function_value_1) Likewise.
73 (mips_block_move_straight): Likewise.
74 (mips_expand_ins_as_unaligned_store): Likewise.
75 * config/powerpcspe/powerpcspe.c
76 (rs6000_darwin64_record_arg_advance_flush): Likewise.
77 (rs6000_darwin64_record_arg_flush): Likewise.
78 * config/rs6000/rs6000.c
79 (rs6000_darwin64_record_arg_advance_flush): Likewise.
80 (rs6000_darwin64_record_arg_flush): Likewise.
81 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
82 (sparc_function_value_1): Likewise.
83 * config/spu/spu.c (adjust_operand): Likewise.
84 (spu_emit_branch_or_set): Likewise.
85 (arith_immediate_p): Likewise.
86 * emit-rtl.c (gen_lowpart_common): Likewise.
87 * expr.c (expand_expr_real_1): Likewise.
88 * function.c (assign_parm_setup_block): Likewise.
89 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
90 * reload1.c (alter_reg): Likewise.
91 * stor-layout.c (mode_for_vector): Likewise.
92 (layout_type): Likewise.
94 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
96 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
97 (spu_convert_move): Likewise.
98 * lower-subreg.c (resolve_simple_move): Likewise.
100 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
103 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
104 define_insn to a define_expand.
105 (altivec_vsum2sws_direct): New define_insn.
106 (altivec_vsumsws): Convert from a define_insn to a define_expand.
108 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
110 * config/arm/arm.c (arm_option_params_internal): Improve setting of
113 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
118 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
119 realign stack if stack alignment needed is less than incoming
122 2017-09-05 Marek Polacek <polacek@redhat.com>
125 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
128 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
130 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
132 2017-09-05 Richard Biener <rguenther@suse.de>
134 PR tree-optimization/82084
135 * fold-const.c (can_native_encode_string_p): Handle wide characters.
137 2017-09-05 Richard Biener <rguenther@suse.de>
139 PR tree-optimization/82102
140 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
142 2017-09-05 Martin Liska <mliska@suse.cz>
144 PR tree-optimization/82032
145 * tree-cfg.c (generate_range_test): New function.
146 * tree-cfg.h (generate_range_test): Declared here.
147 * tree-cfgcleanup.c (convert_single_case_switch): New function.
148 (cleanup_control_expr_graph): Use it.
149 * tree-switch-conversion.c (try_switch_expansion): Remove
151 (emit_case_nodes): Use generate_range_test.
153 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
156 * config/i386/i386.md (*<btsc><mode>_mask): Add
157 TARGET_USE_BT to insn constraint.
158 (*btr<mode>_mask): Ditto.
160 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
162 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
163 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
165 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
168 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
169 TARGET_NEON and TARGET_IWMMXT.
170 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
171 TARGET_NEON and TARGET_IWMMXT.
172 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
174 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
176 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
177 (ix86_rewrite_tls_address): Ditto.
178 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
179 (ix86_rewrite_tls_address_1): Ditto.
180 (ix86_rewrite_tls_address): Ditto.
181 * config/i386/predicates.md (tls_address_pattern): New predicate.
182 * config/i386/i386.md (TLS address splitter): New splitter.
184 2017-09-04 Richard Biener <rguenther@suse.de>
186 PR tree-optimization/82084
187 * fold-const.h (can_native_encode_string_p): Declare.
188 * fold-const.c (can_native_encode_string_p): Factor out from ...
189 (native_encode_string): ... here.
190 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
191 vectorizing stores from constants we later cannot handle.
193 2017-09-04 Marek Polacek <polacek@redhat.com>
196 * doc/invoke.texi: Update -Wtautological-compare documentation.
198 2017-09-04 Jeff Law <law@redhat.com>
200 PR tree-optimization/64910
201 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
202 swap the first and last operand if the last is a constant.
204 2017-09-04 Marek Polacek <polacek@redhat.com>
207 * convert.c (do_narrow): When sanitizing signed integer overflows,
208 bail out for signed types.
209 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
211 2017-09-04 Richard Biener <rguenther@suse.de>
213 PR tree-optimization/82060
214 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
215 Move devirtualization after stmt folding and before EH/AB/noreturn
216 cleanup to get the stmt refs canonicalized. Use a bool instead
217 of gimple_modified_p since that doesn't work for NOPs. Schedule
218 NOPs generated by folding for removal.
220 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
221 Alan Hayward <alan.hayward@arm.com>
222 David Sherwood <david.sherwood@arm.com>
224 * coretypes.h (pad_direction): New enum.
225 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
226 (FUNCTION_ARG_PADDING): Likewise.
227 * target.def (function_arg_padding): New hook.
228 * targhooks.h (default_function_arg_padding): Declare.
229 * targhooks.c (default_function_arg_padding): New function.
230 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
231 (TARGET_FUNCTION_ARG_PADDING): ...this.
232 * doc/tm.texi: Regenerate.
233 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
234 instead of direction.
235 (compute_argument_addresses): Likewise.
236 (load_register_parameters): Likewise.
237 (emit_library_call_value_1): Likewise.
238 (store_one_arg): Use targetm.calls.function_arg_padding instead
239 of FUNCTION_ARG_PADDING.
240 (must_pass_in_stack_var_size_or_pad): Likewise.
241 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
242 (emit_group_store): Likewise.
243 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
244 instead of FUNCTION_ARG_PADDING.
245 (emit_push_insn): Likewise, and propagate enum change throughout
247 * function.h (direction): Delete.
248 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
250 * function.c (assign_parm_find_stack_rtl): Likewise.
251 (assign_parm_setup_block_p): Likewise.
252 (assign_parm_setup_block): Likewise.
253 (gimplify_parameters): Likewise.
254 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
255 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
257 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
258 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
259 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
260 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
261 (aarch64_function_arg_padding): ...this new function.
262 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
263 (TARGET_FUNCTION_ARG_PADDING): Redefine.
264 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
265 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
266 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
267 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
268 (arm_pad_arg_upward): Replace with...
269 (arm_function_arg_padding): ...this new function.
270 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
272 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
273 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
274 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
275 (ia64_hpux_function_arg_padding): Replace with...
276 (ia64_function_arg_padding): ...this new function. Use pad_direction
277 instead of direction. Check for TARGET_HPUX.
278 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
279 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
280 (iq2000_function_arg_padding): New function.
281 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
282 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
283 (mips_function_arg_padding): ...this new function.
284 (mips_pad_reg_upward): Update accordingly.
285 (TARGET_FUNCTION_ARG_PADDING): Redefine.
286 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
287 targetm.calls.function_arg_padding.
288 (FUNCTION_ARG_PADDING): Delete.
289 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
290 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
291 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
292 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
293 (nios2_block_reg_padding): Return pad_direction instead of direction.
294 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
295 instead of direction.
296 (nios2_function_arg_padding): Likewise. Make static.
297 (TARGET_FUNCTION_ARG_PADDING): Redefine.
298 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
299 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
300 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
301 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
302 (pa_function_arg_padding): Make static. Return pad_direction instead
304 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
305 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
306 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
307 instead of direction. Use targetm.calls.function_arg_padding.
308 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
309 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
310 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
311 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
312 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
314 (function_arg_padding): Rename to...
315 (rs6000_function_arg_padding): ...this. Make static. Return
316 pad_direction instead of direction.
317 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
318 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
319 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
320 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
321 instead of direction. Use targetm.calls.function_arg_padding.
322 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
323 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
324 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
325 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
326 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
327 (function_arg_padding): Rename to...
328 (rs6000_function_arg_padding): ...this. Make static. Return
329 pad_direction instead of direction.
330 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
331 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
332 * config/s390/s390.c (s390_function_arg_padding): New function.
333 (TARGET_FUNCTION_ARG_PADDING): Redefine.
334 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
335 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
336 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
337 (function_arg_padding): Rename to...
338 (sparc_function_arg_padding): ...this. Make static. Return
339 pad_direction instead of direction.
340 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
341 * config/spu/spu.c (spu_function_arg_padding): New function.
342 (TARGET_FUNCTION_ARG_PADDING): Redefine.
343 * system.h (FUNCTION_ARG_PADDING): Poison.
345 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
346 Alan Hayward <alan.hayward@arm.com>
347 David Sherwood <david.sherwood@arm.com>
349 * target.def (modes_tieable_p): New hook.
350 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
351 (TARGET_MODES_TIEABLE_P): ...this.
352 * doc/tm.texi.in: Regenerate.
353 * hooks.h (hook_bool_mode_mode_true): Declare.
354 * hooks.c (hook_bool_mode_mode_true): New function.
355 * combine.c (subst): Use targetm.modes_tieable_p instead of
357 * dse.c (find_shift_sequence): Likewise.
358 * expmed.c (extract_low_bits): Likewise.
359 * lower-subreg.c: Include target.h.
360 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
362 * rtlanal.c (rtx_cost): Likewise.
363 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
364 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
365 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
366 (TARGET_MODES_TIEABLE_P): Redefine.
367 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
368 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
369 (TARGET_MODES_TIEABLE_P): Redefine.
370 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
371 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
372 (arc_modes_tieable_p): New function.
373 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
374 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
375 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
376 (arm_modes_tieable_p): Make static.
377 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
378 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
379 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
380 (TARGET_MODES_TIEABLE_P): Redefine.
381 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
382 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
383 (TARGET_MODES_TIEABLE_P): Redefine.
384 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
385 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
386 (cr16_modes_tieable_p): New function.
387 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
388 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
389 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
390 (TRULY_NOOP_TRUNCATION): Update comment.
391 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
392 (TRULY_NOOP_TRUNCATION): Update comment.
393 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
394 (frv_modes_tieable_p): New function.
395 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
396 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
397 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
398 (TARGET_MODES_TIEABLE_P): Redefine.
399 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
400 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
401 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
402 (TARGET_MODES_TIEABLE_P): Redefine.
403 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
404 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
405 (ia64_modes_tieable_p): New function.
406 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
407 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
408 (iq2000_modes_tieable_p): New function.
409 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
410 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
411 (lm32_modes_tieable_p): New function.
412 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
413 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
414 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
415 (TARGET_MODES_TIEABLE_P): Redefine.
416 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
417 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
418 (m32r_modes_tieable_p): New function.
419 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
420 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
421 (m68k_modes_tieable_p): New function.
422 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
423 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
424 (mcore_modes_tieable_p): New function.
425 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
426 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
428 (TARGET_MODES_TIEABLE_P): Redefine.
429 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
430 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
431 * config/mips/mips.c (mips_modes_tieable_p): Make static.
432 (TARGET_MODES_TIEABLE_P): Redefine.
433 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
434 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
435 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
436 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
437 (mn10300_modes_tieable_p): ...this and make static.
438 (TARGET_MODES_TIEABLE_P): Redefine.
439 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
440 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
441 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
442 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
443 (msp430_modes_tieable_p): Make static.
444 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
445 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
446 (TARGET_MODES_TIEABLE_P): Redefine.
447 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
448 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
449 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
450 (TARGET_MODES_TIEABLE_P): Redefine.
451 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
452 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
453 * config/pa/pa.c (pa_modes_tieable_p): Make static.
454 (TARGET_MODES_TIEABLE_P): Redefine.
455 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
456 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
457 (pdp11_modes_tieable_p): New function.
458 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
459 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
460 (rs6000_modes_tieable_p): New function.
461 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
462 * config/powerpcspe/powerpcspe.md: Update comment.
463 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
464 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
465 (TARGET_MODES_TIEABLE_P): Redefine.
466 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
467 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
468 (rl78_modes_tieable_p): New function.
469 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
470 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
471 (rs6000_modes_tieable_p): New function.
472 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
473 * config/rs6000/rs6000.md: Update comment.
474 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
475 * config/rx/rx.c (rx_modes_tieable_p): New function.
476 (TARGET_MODES_TIEABLE_P): Redefine.
477 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
478 * config/s390/s390.c (s390_modes_tieable_p): New function.
479 (TARGET_MODES_TIEABLE_P): Redefine.
480 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
481 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
482 (sh_modes_tieable_p): New function.
483 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
484 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
485 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
486 (sparc_modes_tieable_p): Make static.
487 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
488 * config/spu/spu.c (spu_modes_tieable_p): New function.
489 (TARGET_MODES_TIEABLE_P): Redefine.
490 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
491 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
492 (TARGET_MODES_TIEABLE_P): Redefine.
493 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
494 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
495 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
496 * config/v850/v850.c (v850_modes_tieable_p): New function.
497 (TARGET_MODES_TIEABLE_P): Redefine.
498 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
499 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
500 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
501 (visium_modes_tieable_p): New function.
502 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
503 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
504 (xtensa_modes_tieable_p): New function.
505 * system.h (MODES_TIEABLE_P): Poison.
507 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
508 Alan Hayward <alan.hayward@arm.com>
509 David Sherwood <david.sherwood@arm.com>
511 * target.def (hard_regno_mode_ok): New hook.
512 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
513 (TARGET_HARD_REGNO_MODE_OK): ...this.
514 * doc/tm.texi.in: Regenerate.
515 * hooks.h (hook_bool_uint_mode_true): Declare.
516 * hooks.c (hook_bool_uint_mode_true): New function.
517 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
519 * genpreds.c (write_insn_preds_c): Add an include of target.h.
520 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
521 instead of HARD_REGNO_MODE_OK.
522 * caller-save.c: Include target.h.
523 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
525 * combine.c (can_combine_p): Likewise.
526 (combinable_i3pat): Likewise.
527 (can_change_dest_mode): Likewise.
528 * expr.c (init_expr_target): Likewise.
529 (convert_move): Likewise.
530 (convert_modes): Likewise.
531 * ira.c (setup_prohibited_class_mode_regs): Likewise.
532 (setup_prohibited_mode_move_regs): Likewise.
533 * ira.h (target_ira): Likewise.
534 * lra-assigns.c (find_hard_regno_for_1): Likewise.
535 * lra-constraints.c (process_alt_operands): Likewise.
536 (split_reg): Likewise.
537 * recog.c (peep2_find_free_register): Likewise.
538 * ree.c (combine_reaching_defs): Likewise.
539 * regcprop.c (maybe_mode_change): Likewise.
540 * reginfo.c (init_reg_sets_1): Likewise.
541 (choose_hard_reg_mode): Likewise.
542 (simplifiable_subregs): Likewise.
543 * regrename.c (check_new_reg_p): Likewise.
544 * reload.c (find_valid_class): Likewise.
545 (find_valid_class_1): Likewise.
546 (reload_inner_reg_of_subreg): Likewise.
547 (push_reload): Likewise.
548 (combine_reloads): Likewise.
549 (find_dummy_reload): Likewise.
550 (find_reloads): Likewise.
551 * reload1.c (find_reg): Likewise.
552 (set_reload_reg): Likewise.
553 (allocate_reload_reg): Likewise.
554 (choose_reload_regs): Likewise.
555 (reload_adjust_reg_for_temp): Likewise.
556 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
557 (simplify_subreg_regno): Likewise.
558 * sel-sched.c (init_regs_for_mode): Likewise.
559 * varasm.c (make_decl_rtl): Likewise.
560 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
561 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
563 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
564 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
565 (TARGET_HARD_REGNO_MODE_OK): Redefine.
566 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
567 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
568 (TARGET_HARD_REGNO_MODE_OK): Redefine.
569 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
570 (arc_mode_class): Delete.
571 (HARD_REGNO_MODE_OK): Delete.
572 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
573 (arc_hard_regno_mode_ok): Rename old array to...
574 (arc_hard_regno_mode_ok_modes): ...this.
575 (arc_conditional_register_usage): Update accordingly.
576 (arc_mode_class): Make static.
577 (arc_hard_regno_mode_ok): New function.
578 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
579 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
580 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
581 (arm_hard_regno_mode_ok): Make static.
582 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
584 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
585 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
586 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
588 (TARGET_HARD_REGNO_MODE_OK): Redefine.
589 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
590 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
591 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
592 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
593 (TARGET_HARD_REGNO_MODE_OK): Redefine.
594 * config/bfin/predicates.md (valid_reg_operand): Use
595 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
596 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
597 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
598 (TARGET_HARD_REGNO_MODE_OK): Redefine.
599 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
600 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
601 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
602 (cr16_hard_regno_mode_ok): Make static and return a bool.
603 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
604 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
605 (cris_hard_regno_mode_ok): New function.
606 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
607 (epiphany_mode_class): Delete.
608 (HARD_REGNO_MODE_OK): Delete.
609 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
610 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
611 (hard_regno_mode_ok): Rename to...
612 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
613 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
614 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
616 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
617 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
618 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
619 (frv_hard_regno_mode_ok): Make static and return a bool.
620 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
622 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
623 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
624 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
625 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
627 (TARGET_HARD_REGNO_MODE_OK): Redefine.
628 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
629 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
630 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
632 (TARGET_HARD_REGNO_MODE_OK): Redefine.
633 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
634 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
635 (ia64_hard_regno_mode_ok): New function.
636 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
637 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
638 (iq2000_hard_regno_mode_ok): New function.
639 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
640 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
641 (lm32_hard_regno_mode_ok): New function.
642 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
643 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
644 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
645 instead of HARD_REGNO_MODE_OK.
646 (m32c_hard_regno_ok): Rename to...
647 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
648 (m32c_cannot_change_mode_class): Update accordingly.
649 (TARGET_HARD_REGNO_MODE_OK): Redefine.
650 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
651 (m32r_mode_class): Delete.
652 (HARD_REGNO_MODE_OK): Delete.
653 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
654 (m32r_hard_regno_mode_ok): Rename to...
655 (m32r_hard_regno_modes): ...this.
656 (m32r_mode_class): Make static.
657 (m32r_hard_regno_mode_ok): New function.
658 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
659 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
660 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
661 (m68k_hard_regno_mode_ok): Make static.
662 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
663 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
664 (mcore_hard_regno_mode_ok): New function.
665 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
666 (HARD_REGNO_MODE_OK): Delete.
667 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
669 (microblaze_hard_regno_mode_ok_p): ...this and make static.
670 (microblaze_hard_regno_mode_ok): New function.
671 (TARGET_HARD_REGNO_MODE_OK): Redefine.
672 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
673 (mips_hard_regno_mode_ok): Delete.
674 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
675 (mips_hard_regno_mode_ok_p): ...this and make static.
676 (mips_hard_regno_mode_ok_p): Rename to...
677 (mips_hard_regno_mode_ok_uncached): ...this.
678 (mips_hard_regno_mode_ok): New function.
679 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
680 of HARD_REGNO_MODE_OK.
681 (mips_option_override): Update after above name changes.
682 (TARGET_HARD_REGNO_MODE_OK): Redefine.
683 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
684 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
685 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
686 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
687 (TARGET_HARD_REGNO_MODE_OK): Redefine.
688 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
689 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
690 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
691 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
692 (msp430_hard_regno_mode_ok): Make static and return a bool.
693 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
694 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
695 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
697 (TARGET_HARD_REGNO_MODE_OK): Redefine.
698 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
699 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
700 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
701 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
702 (PA_HARD_REGNO_MODE_OK): ...this
703 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
704 (PA_HARD_REGNO_MODE_OK): ...this.
705 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
706 (pa_hard_regno_mode_ok): New function.
707 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
708 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
709 (pdp11_hard_regno_mode_ok): New function.
710 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
711 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
713 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
715 (TARGET_HARD_REGNO_MODE_OK): Redefine.
716 (rs6000_hard_regno_mode_ok): Rename to...
717 (rs6000_hard_regno_mode_ok_uncached): ...this.
718 (rs6000_init_hard_regno_mode_ok): Update accordingly.
719 (rs6000_hard_regno_mode_ok): New function.
720 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
721 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
722 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
723 (riscv_hard_regno_mode_ok): ...this and make static.
724 (TARGET_HARD_REGNO_MODE_OK): Redefine.
725 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
726 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
727 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
728 (rl78_hard_regno_mode_ok): Make static and return bool.
729 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
730 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
732 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
733 (TARGET_HARD_REGNO_MODE_OK): Redefine.
734 (rs6000_hard_regno_mode_ok): Rename to...
735 (rs6000_hard_regno_mode_ok_uncached): ...this.
736 (rs6000_init_hard_regno_mode_ok): Update accordingly.
737 (rs6000_hard_regno_mode_ok): New function.
738 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
739 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
740 (TARGET_HARD_REGNO_MODE_OK): Redefine.
741 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
742 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
743 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
744 (TARGET_HARD_REGNO_MODE_OK): Redefine.
745 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
746 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
747 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
748 (sh_hard_regno_mode_ok): Make static.
749 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
750 instead of HARD_REGNO_MODE_OK.
751 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
752 (sparc_mode_class): Delete.
753 (HARD_REGNO_MODE_OK): Delete.
754 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
755 (hard_regno_mode_classes): Make static.
756 (sparc_mode_class): Likewise.
757 (sparc_hard_regno_mode_ok): New function.
758 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
759 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
760 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
762 (TARGET_HARD_REGNO_MODE_OK): Redefine.
763 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
764 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
765 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
766 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
767 (TARGET_HARD_REGNO_MODE_OK): Redefine.
768 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
769 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
770 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
771 (visium_hard_regno_mode_ok): New function.
772 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
773 instead of HARD_REGNO_MODE_OK.
774 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
775 (HARD_REGNO_MODE_OK): Delete.
776 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
777 (xtensa_hard_regno_mode_ok_p): ...this and make static.
778 (xtensa_option_override): Update accordingly.
779 (TARGET_HARD_REGNO_MODE_OK): Redefine.
780 (xtensa_hard_regno_mode_ok): New function.
781 * system.h (HARD_REGNO_MODE_OK): Poison.
783 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
784 Alan Hayward <alan.hayward@arm.com>
785 David Sherwood <david.sherwood@arm.com>
787 * target.def (hard_regno_call_part_clobbered): New hook.
788 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
789 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
790 * doc/tm.texi: Regenerate.
791 * hooks.h (hook_bool_uint_mode_false): Declare.
792 * hooks.c (hook_bool_uint_mode_false): New function.
793 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
794 * cselib.c (cselib_process_insn): Use
795 targetm.hard_regno_call_part_clobbered instead of
796 HARD_REGNO_CALL_PART_CLOBBERED.
797 * ira-conflicts.c (ira_build_conflicts): Likewise.
798 * ira-costs.c (ira_tune_allocno_costs): Likewise.
799 * lra-constraints.c (need_for_call_save_p): Likewise.
800 * lra-lives.c: Include target.h.
801 (check_pseudos_live_through_calls): Use
802 targetm.hard_regno_call_part_clobbered instead of
803 HARD_REGNO_CALL_PART_CLOBBERED.
804 * regcprop.c: Include target.h.
805 (copyprop_hardreg_forward_1): Use
806 targetm.hard_regno_call_part_clobbered instead of
807 HARD_REGNO_CALL_PART_CLOBBERED.
808 * reginfo.c (choose_hard_reg_mode): Likewise.
809 * regrename.c (check_new_reg_p): Likewise.
810 * reload.c (find_equiv_reg): Likewise.
811 * reload1.c (emit_reload_insns): Likewise.
812 * sched-deps.c (deps_analyze_insn): Likewise.
813 * sel-sched.c (init_regs_for_mode): Likewise.
814 (mark_unavailable_hard_regs): Likewise.
815 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
816 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
817 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
819 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
820 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
821 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
823 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
825 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
826 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
827 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
829 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
830 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
831 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
833 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
834 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
836 * config/powerpcspe/powerpcspe.c
837 (rs6000_hard_regno_call_part_clobbered): New function.
838 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
839 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
840 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
842 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
843 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
844 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
846 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
847 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
848 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
850 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
851 Alan Hayward <alan.hayward@arm.com>
852 David Sherwood <david.sherwood@arm.com>
854 * rtl.h (subreg_memory_offset): Declare.
855 * emit-rtl.c (subreg_memory_offset): New function.
856 * expmed.c (store_bit_field_1): Use it.
857 * expr.c (undefined_operand_subword_p): Likewise.
858 * simplify-rtx.c (simplify_subreg): Likewise.
860 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
862 PR rtl-optimization/57448
865 * optabs.c (expand_atomic_load): Place compiler memory barriers if
866 using atomic_load pattern.
867 (expand_atomic_store): Likewise.
869 2017-09-04 Jakub Jelinek <jakub@redhat.com>
872 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
873 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
874 handling. Use replace_call_with_value with NULL instead of
875 gsi_replace, unlink_stmt_vdef and release_defs.
877 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
880 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
882 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
885 * rtl.h (emit_library_call_value_1): Declare.
886 (emit_library_call): Replace declaration with a series of overloads.
887 Remove the parameter count argument.
888 (emit_library_call_value): Likewise.
889 * calls.c (emit_library_call_value_1): Make global. Replace varargs
890 with an "rtx_mode_t *".
891 (emit_library_call_value): Delete.
892 (emit_library_call): Likewise.
893 * asan.c (asan_emit_stack_protection): Update calls accordingly.
894 (asan_emit_allocas_unpoison): Likewise.
895 * builtins.c (expand_builtin_powi): Likewise.
896 (expand_asan_emit_allocas_unpoison): Likewise.
897 * cfgexpand.c (expand_main_function): Likewise.
898 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
899 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
900 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
901 * config/arm/arm.c (arm_trampoline_init): Likewise.
902 (arm_call_tls_get_addr): Likewise.
903 (arm_expand_divmod_libfunc): Likewise.
904 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
905 (smulsi3_highpart): Likewise.
906 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
907 (c6x_expand_compare): Likewise.
908 (c6x_expand_movmem): Likewise.
909 * config/frv/frv.c (frv_trampoline_init): Likewise.
910 * config/i386/i386.c (ix86_trampoline_init): Likewise.
911 (ix86_expand_divmod_libfunc): Likewise.
912 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
913 (ia64_expand_compare): Likewise.
914 (ia64_profile_hook): Likewise.
915 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
916 (nonlocal_goto): Likewise.
917 (restore_stack_nonlocal): Likewise.
918 * config/m32r/m32r.c (block_move_call): Likewise.
919 (m32r_trampoline_init): Likewise.
920 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
921 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
922 (m68k_call_m68k_read_tp): Likewise.
923 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
924 (microblaze_expand_divide): Likewise.
925 * config/mips/mips.h (mips_args): Likewise.
926 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
927 (MIPS_ICACHE_SYNC): Likewise.
928 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
929 (nios2_trampoline_init): Likewise.
930 * config/pa/pa.c (hppa_tls_call): Likewise.
931 (pa_trampoline_init): Likewise.
932 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
933 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
934 (expand_strn_compare): Likewise.
935 (rs6000_generate_compare): Likewise.
936 (rs6000_expand_float128_convert): Likewise.
937 (output_profile_hook): Likewise.
938 (rs6000_trampoline_init): Likewise.
939 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
940 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
941 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
942 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
943 (rs6000_generate_compare): Likewise.
944 (rs6000_expand_float128_convert): Likewise.
945 (output_profile_hook): Likewise.
946 (rs6000_trampoline_init): Likewise.
947 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
948 * config/sh/sh.c (sh_trampoline_init): Likewise.
949 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
950 (sparc_emit_float_lib_cmp): Likewise.
951 (sparc32_initialize_trampoline): Likewise.
952 (sparc64_initialize_trampoline): Likewise.
953 (sparc_profile_hook): Likewise.
954 * config/spu/spu.c (ea_load_store): Likewise.
955 * config/spu/spu.md (floatunssidf2): Likewise.
956 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
957 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
958 * config/visium/visium.c (expand_block_move_4): Likewise.
959 (expand_block_move_2): Likewise.
960 (expand_block_move_1): Likewise.
961 (expand_block_set_4): Likewise.
962 (expand_block_set_2): Likewise.
963 (expand_block_set_1): Likewise.
964 (visium_trampoline_init): Likewise.
965 (visium_profile_hook): Likewise.
966 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
967 (xtensa_setup_frame_addresses): Likewise.
968 (xtensa_trampoline_init): Likewise.
969 * except.c (sjlj_emit_function_enter): Likewise.
970 (sjlj_emit_function_exit): Likewise.
971 * explow.c (allocate_dynamic_stack_space): Likewise.
972 (probe_stack_range): Likewise.
973 * expr.c (convert_mode_scalar): Likewise.
974 * optabs.c (expand_binop): Likewise.
975 (expand_twoval_binop_libfunc): Likewise.
976 (expand_unop): Likewise.
977 (prepare_cmp_insn): Likewise.
978 (prepare_float_lib_cmp): Likewise.
979 (expand_float): Likewise.
980 (expand_fix): Likewise.
981 (expand_fixed_convert): Likewise.
982 (maybe_emit_sync_lock_test_and_set): Likewise.
983 (expand_atomic_compare_and_swap): Likewise.
984 (expand_mem_thread_fence): Likewise.
985 (expand_atomic_fetch_op): Likewise.
987 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
989 * doc/generic.texi (OpenACC): Adjust URL.
990 * doc/invoke.texi (C Dialect Options): Ditto.
992 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
994 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
995 predicate for operand 1. Add (m,<S>) constraint.
996 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
997 Prevent memory operand 1 with register operand 2.
999 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
1001 PR rtl-optimization/82024
1002 * combine.c (try_combine): If the combination result is a PARALLEL,
1003 and we only need to retain the SET in there that would be placed
1004 at I2, check that we can place that at I3 instead, before doing so.
1006 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1009 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
1011 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
1012 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
1015 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
1016 Jeff Law <law@redhat.com>
1018 * varasm.c (bss_initializer_p): Do not put constants into .bss
1019 (categorize_decl_for_section): Handle bss_initializer_p returning
1020 false when DECL_INITIAL is NULL.
1022 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1025 * config/s390/s390.c (s390_can_inline_p): New function.
1027 2017-09-01 Jeff Law <law@redhat.com>
1029 PR tree-optimization/82052
1030 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
1031 Always initialize the returned slot after a hash table miss
1032 when INSERT is true.
1034 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
1036 * config/s390/s390.md (mem_signal_fence): Remove.
1037 * doc/md.texi (mem_signal_fence): Remove.
1038 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
1040 * target-insns.def (mem_signal_fence): Remove.
1042 2017-09-01 Jakub Jelinek <jakub@redhat.com>
1045 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
1048 * asan.c (create_odr_indicator): Strip name encoding from assembler
1049 name before appending it after __odr_asan_.
1051 2017-09-01 Martin Liska <mliska@suse.cz>
1053 PR tree-optimization/82059
1054 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
1055 frequency only when an edge is redirected.
1057 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1059 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
1060 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
1061 (arc_conditional_register_usage): Remove ARC600 lp_count
1063 (arc_file_start): Emit Tag_ARC_CPU_variation.
1064 (arc_can_use_doloop_p): New conditions to use ZOLs.
1065 (hwloop_fail): New function.
1066 (hwloop_optimize): Likewise.
1067 (hwloop_pattern_reg): Likewise.
1068 (arc_doloop_hooks): New struct, to be used with reorg_loops.
1069 (arc_reorg_loops): New function, calls reorg_loops.
1070 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
1071 (arc600_corereg_hazard): Remove ZOL checking, case handled by
1073 (arc_loop_hazard): Remove function, functionality moved into
1075 (arc_hazard): Remove arc_loop_hazard call.
1076 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
1077 into hwloop_optimize.
1078 (arc_label_align): Remove ZOL handling.
1079 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
1080 * config/arc/arc.md (doloop_begin): Remove pattern.
1081 (doloop_begin_i): Likewise.
1082 (doloop_end_i): Likewise.
1083 (doloop_fallback): Likewise.
1084 (doloop_fallback_m): Likewise.
1085 (doloop_end): Reimplement expand.
1086 (arc_lp): New pattern for LP instruction.
1087 (loop_end): New pattern.
1088 (loop_fail): Likewise.
1089 (decrement_and_branch_until_zero): Likewise.
1090 * config/arc/arc.opt (mlpc-width): New option.
1091 * doc/invoke.texi (mlpc-width): Document option.
1093 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1095 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
1096 (arc_ccfsm_advance): Fix checking for delay slots.
1097 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
1099 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1101 * config/arc/arc.md (movqi_insn): Add stores to save constant long
1103 (movhi_insn): Update store instruction constraint which are saving
1104 6-bit short immediates.
1105 (movsi_insn): Consider also short scaled load operations.
1106 (zero_extendhisi2_i): Use Usd constraint instead of T.
1107 (extendhisi2_i): Add q constraint.
1108 (arc_clzsi2): Add type and length attributes.
1109 (arc_ctzsi2): Likewise.
1110 * config/arc/constraints.md (Usc): Update constraint, the
1111 assembler can parse two relocations for a single instruction.
1113 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
1115 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
1116 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
1118 2017-08-31 Olivier Hainque <hainque@adacore.com>
1120 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
1121 match as powerpc-wrs-vxworks*.
1123 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
1125 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
1126 register constraint for by-element operand.
1127 (aarch64_mls_elt_merge<mode>): Likewise.
1129 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1131 * config/arc/arc.c (arc_can_follow_jump): Check for short
1134 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1136 * config.gcc: Use g.opt for arc.
1137 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
1138 functionality moved to ...
1139 (legitimate_scaled_address_p): New function, ...here.
1140 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
1141 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
1142 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
1144 (arc_override_options): Handle G option.
1145 (arc_output_pic_addr_const): Correct function definition.
1146 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
1147 (arc_decl_anon_ns_mem_p): Delete.
1148 (arc_in_small_data_p): Overhaul this function to take into
1149 consideration the value given via G option.
1150 (arc_rewrite_small_data_1): Renamed and corrected old
1151 arc_rewrite_small_data function.
1152 (arc_rewrite_small_data): New function.
1153 (small_data_pattern): Don't use pic_offset_table_rtx.
1154 * config/arc/arc.h (CC1_SPEC): Recognize G option.
1155 * config/arc/simdext.md (movmisalignv2hi): Use
1156 prepare_move_operands function.
1158 (movmisalign*): Likewise.
1159 * doc/invoke.texi (ARC options): Document -G option.
1161 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
1163 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
1165 * config/arc/arc.c (arc_print_operand): Output scalled address for
1166 sdata whenever is possible.
1167 (arc_in_small_data_p): Allow sdata for 64bit datum when double
1168 load/stores are available.
1169 (compact_sda_memory_operand): Check for the alignment required by
1170 code density instructions.
1171 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
1173 * config/arc/constraints.md (Usd): Update constraint.
1174 (Us0): New constraint.
1175 (Usc): Update constraint.
1177 2017-08-31 Richard Biener <rguenther@suse.de>
1180 * dwarf2out.c (dwarf2out_early_global_decl): Process each
1183 2017-08-31 Tamar Christina <tamar.christina@arm.com>
1185 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
1186 Resize type_signature.
1188 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1189 Alan Hayward <alan.hayward@arm.com>
1190 David Sherwood <david.sherwood@arm.com>
1192 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
1193 subregs whose inner modes can be stored in GPRs.
1194 (aarch64_classify_index): Likewise.
1196 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1197 Alan Hayward <alan.hayward@arm.com>
1198 David Sherwood <david.sherwood@arm.com>
1200 * config/aarch64/iterators.md (V_cmp_result): Rename to...
1201 (V_INT_EQUIV): ...this.
1202 (v_cmp_result): Rename to...
1203 (v_int_equiv): ...this.
1204 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
1205 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
1206 (copysign<mode>3): Likewise.
1207 (aarch64_simd_bsl<mode>_internal): Likewise.
1208 (aarch64_simd_bsl<mode>): Likewise.
1209 (vec_cmp<mode><mode>): Likewise.
1210 (vcond<mode><mode>): Likewise.
1211 (vcond<v_cmp_mixed><mode>): Likewise.
1212 (vcondu<mode><v_cmp_mixed>): Likewise.
1213 (aarch64_cm<optab><mode>): Likewise.
1214 (aarch64_cmtst<mode>): Likewise.
1215 (aarch64_fac<optab><mode>): Likewise.
1216 (vec_perm_const<mode>): Likewise.
1217 (vcond_mask_<mode><v_cmp_result>): Rename to...
1218 (vcond_mask_<mode><v_int_equiv>): ...this.
1219 (vec_cmp<mode><v_cmp_result>): Rename to...
1220 (vec_cmp<mode><v_int_equiv>): ...this.
1222 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
1223 Alan Hayward <alan.hayward@arm.com>
1224 David Sherwood <david.sherwood@arm.com>
1226 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
1228 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
1229 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
1230 (UNSPEC_LD4_DREG): New unspecs.
1231 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
1232 (aarch64_ld2<mode>_dreg_be): Replace with...
1233 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
1235 (aarch64_ld3<mode>_dreg_le)
1236 (aarch64_ld3<mode>_dreg_be): Replace with...
1237 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
1239 (aarch64_ld4<mode>_dreg_le)
1240 (aarch64_ld4<mode>_dreg_be): Replace with...
1241 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
1244 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1246 PR tree-optimization/81987
1247 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
1248 insert an initializer in a location not dominated by the stride
1251 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1253 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
1254 on the entire header of the finally block in the fallthru case.
1256 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
1258 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
1260 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
1262 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
1263 rs6000_emit_move_from_cr and call renamed function.
1264 (rs6000_emit_prologue): Call renamed functions.
1265 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
1266 movesi_from_cr, remove volatile CRs.
1268 2017-08-30 Jon Beniston <jon@beniston.com>
1269 Richard Biener <rguenther@suse.de>
1271 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
1272 of VECTOR_MODE_P check.
1273 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
1274 element vector types.
1276 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1278 * df.h (df_read_modify_subreg_p): Remove in favor of...
1279 * rtl.h (read_modify_subreg_p): ...this new function. Take a
1280 const_rtx instead of an rtx.
1281 * cprop.c (local_cprop_find_used_regs): Update accordingly.
1282 * df-problems.c (df_word_lr_mark_ref): Likewise.
1283 * ira-lives.c (mark_pseudo_reg_live): Likewise.
1284 (mark_pseudo_reg_dead): Likewise.
1285 (mark_ref_dead): Likewise.
1286 * reginfo.c (init_subregs_of_mode): Likewise.
1287 * sched-deps.c (sched_analyze_1): Likewise.
1288 * df-scan.c (df_def_record_1): Likewise.
1289 (df_uses_record): Likewise.
1290 (df_read_modify_subreg_p): Remove in favor of...
1291 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
1292 const_rtx instead of an rtx.
1294 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1295 Alan Hayward <alan.hayward@arm.com>
1296 David Sherwood <david.sherwood@arm.com>
1298 * rtl.h (partial_subreg_p): New function.
1299 * caller-save.c (save_call_clobbered_regs): Use it.
1300 * calls.c (expand_call): Likewise.
1301 * combine.c (combinable_i3pat): Likewise.
1302 (simplify_set): Likewise.
1303 (make_extraction): Likewise.
1304 (make_compound_operation_int): Likewise.
1305 (gen_lowpart_or_truncate): Likewise.
1306 (force_to_mode): Likewise.
1307 (make_field_assignment): Likewise.
1308 (reg_truncated_to_mode): Likewise.
1309 (record_truncated_value): Likewise.
1310 (move_deaths): Likewise.
1311 * cse.c (record_jump_cond): Likewise.
1312 (cse_insn): Likewise.
1313 * cselib.c (cselib_lookup_1): Likewise.
1314 * expmed.c (extract_bit_field_using_extv): Likewise.
1315 * function.c (assign_parm_setup_reg): Likewise.
1316 * ifcvt.c (noce_convert_multiple_sets): Likewise.
1317 * ira-build.c (create_insn_allocnos): Likewise.
1318 * lra-coalesce.c (merge_pseudos): Likewise.
1319 * lra-constraints.c (match_reload): Likewise.
1320 (simplify_operand_subreg): Likewise.
1321 (curr_insn_transform): Likewise.
1322 * lra-lives.c (process_bb_lives): Likewise.
1323 * lra.c (new_insn_reg): Likewise.
1324 (lra_substitute_pseudo): Likewise.
1325 * regcprop.c (mode_change_ok): Likewise.
1326 (maybe_mode_change): Likewise.
1327 (copyprop_hardreg_forward_1): Likewise.
1328 * reload.c (push_reload): Likewise.
1329 (find_reloads): Likewise.
1330 (find_reloads_subreg_address): Likewise.
1331 * reload1.c (alter_reg): Likewise.
1332 (eliminate_regs_1): Likewise.
1333 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1335 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
1337 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
1338 back to if statements, including unpack.
1340 2017-08-30 Martin Liska <mliska@suse.cz>
1343 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
1345 (func_checker::compare_asm_inputs_outputs): ... this function.
1346 (func_checker::compare_gimple_asm): Use the function to compare
1347 also ASM constrains.
1348 * ipa-icf-gimple.h: Rename the function.
1350 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1351 Alan Hayward <alan.hayward@arm.com>
1352 David Sherwood <david.sherwood@arm.com>
1354 * coretypes.h (complex_mode): New type.
1355 * gdbhooks.py (build_pretty_printer): Handle it.
1356 * machmode.h (complex_mode): New class.
1357 (complex_mode::includes_p): New function.
1358 (is_complex_int_mode): Likewise.
1359 (is_complex_float_mode): Likewise.
1360 * genmodes.c (get_mode_class): Handle complex mode classes.
1361 * function.c (expand_function_end): Use is_complex_int_mode.
1363 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1364 Alan Hayward <alan.hayward@arm.com>
1365 David Sherwood <david.sherwood@arm.com>
1367 * coretypes.h (scalar_mode_pod): New typedef.
1368 * gdbhooks.py (build_pretty_printer): Handle it.
1369 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
1370 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
1371 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
1372 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
1375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1376 Alan Hayward <alan.hayward@arm.com>
1377 David Sherwood <david.sherwood@arm.com>
1379 * machmode.h (mode_for_vector): Take a scalar_mode instead
1381 * stor-layout.c (mode_for_vector): Likewise.
1382 * explow.c (promote_mode): Use as_a <scalar_mode>.
1383 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
1385 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1386 Alan Hayward <alan.hayward@arm.com>
1387 David Sherwood <david.sherwood@arm.com>
1389 * target.def (preferred_simd_mode): Take a scalar_mode
1390 instead of a machine_mode.
1391 * targhooks.h (default_preferred_simd_mode): Likewise.
1392 * targhooks.c (default_preferred_simd_mode): Likewise.
1393 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
1394 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
1395 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
1396 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
1397 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
1398 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
1399 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
1400 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
1402 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
1403 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
1404 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
1405 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
1406 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
1407 * doc/tm.texi: Regenerate.
1408 * optabs-query.c (can_vec_mask_load_store_p): Return false for
1411 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1412 Alan Hayward <alan.hayward@arm.com>
1413 David Sherwood <david.sherwood@arm.com>
1415 * target.def (scalar_mode_supported_p): Take a scalar_mode
1416 instead of a machine_mode.
1417 * targhooks.h (default_scalar_mode_supported_p): Likewise.
1418 * targhooks.c (default_scalar_mode_supported_p): Likewise.
1419 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
1420 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1421 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
1422 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
1423 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
1424 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
1425 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
1426 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
1427 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
1428 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
1429 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
1430 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
1432 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
1433 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
1434 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1435 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1436 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
1438 * doc/tm.texi: Regenerate.
1440 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1441 Alan Hayward <alan.hayward@arm.com>
1442 David Sherwood <david.sherwood@arm.com>
1444 * coretypes.h (opt_scalar_mode): New typedef.
1445 * gdbhooks.py (build_pretty_printers): Handle it.
1446 * machmode.h (mode_iterator::get_2xwider): Add overload for
1448 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
1450 * expr.c (convert_mode_scalar): Likewise.
1451 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1452 * optabs.c (expand_float): Likewise.
1453 (expand_fix): Likewise.
1454 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1456 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1457 Alan Hayward <alan.hayward@arm.com>
1458 David Sherwood <david.sherwood@arm.com>
1460 * optabs.c (expand_float): Explicitly check for scalars before
1461 using a branching expansion.
1462 (expand_fix): Likewise.
1464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1465 Alan Hayward <alan.hayward@arm.com>
1466 David Sherwood <david.sherwood@arm.com>
1468 * expr.c (convert_mode): Split scalar handling out into...
1469 (convert_mode_scalar): ...this new function. Treat the modes
1472 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1473 Alan Hayward <alan.hayward@arm.com>
1474 David Sherwood <david.sherwood@arm.com>
1476 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
1478 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
1480 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1481 Alan Hayward <alan.hayward@arm.com>
1482 David Sherwood <david.sherwood@arm.com>
1484 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
1485 rather than a machine_mode.
1486 (fixed_from_string): Likewise.
1487 (fixed_convert): Likewise.
1488 (fixed_convert_from_int): Likewise.
1489 (fixed_convert_from_real): Likewise.
1490 (real_convert_from_fixed): Likewise.
1491 * fixed-value.c (fixed_from_double_int): Likewise.
1492 (fixed_from_string): Likewise.
1493 (fixed_convert): Likewise.
1494 (fixed_convert_from_int): Likewise.
1495 (fixed_convert_from_real): Likewise.
1496 (real_convert_from_fixed): Likewise.
1497 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
1499 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1500 Alan Hayward <alan.hayward@arm.com>
1501 David Sherwood <david.sherwood@arm.com>
1503 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
1504 of separate mode class checks. Do not allow vector modes here.
1505 (immed_wide_int_const): Use as_a <scalar_mode>.
1506 * explow.c (trunc_int_for_mode): Likewise.
1507 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
1508 (wi::shwi): Likewise.
1509 (wi::min_value): Likewise.
1510 (wi::max_value): Likewise.
1511 * dwarf2out.c (loc_descriptor): Likewise.
1512 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
1515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1516 Alan Hayward <alan.hayward@arm.com>
1517 David Sherwood <david.sherwood@arm.com>
1519 * tree.h (SCALAR_TYPE_MODE): New macro.
1520 * expr.c (expand_expr_addr_expr_1): Use it.
1521 (expand_expr_real_2): Likewise.
1522 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
1523 (fold_convert_const_fixed_from_int): Likewise.
1524 (fold_convert_const_fixed_from_real): Likewise.
1525 (native_encode_fixed): Likewise
1526 (native_encode_complex): Likewise
1527 (native_encode_vector): Likewise.
1528 (native_interpret_fixed): Likewise.
1529 (native_interpret_real): Likewise.
1530 (native_interpret_complex): Likewise.
1531 (native_interpret_vector): Likewise.
1532 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
1533 (simd_clone_adjust_argument_types): Likewise.
1534 (simd_clone_init_simd_arrays): Likewise.
1535 (simd_clone_adjust): Likewise.
1536 * stor-layout.c (layout_type): Likewise.
1537 * tree.c (build_minus_one_cst): Likewise.
1538 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
1539 * tree-inline.c (estimate_move_cost): Likewise.
1540 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
1541 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
1542 (vectorizable_reduction): Likewise.
1543 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
1544 (vect_recog_mixed_size_cond_pattern): Likewise.
1545 (check_bool_pattern): Likewise.
1546 (adjust_bool_pattern): Likewise.
1547 (search_type_for_mask_1): Likewise.
1548 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
1549 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1550 (vectorizable_load): Likewise.
1551 (vectorizable_store): Likewise.
1552 * ubsan.c (ubsan_encode_value): Likewise.
1553 * varasm.c (output_constant): Likewise.
1555 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1556 Alan Hayward <alan.hayward@arm.com>
1557 David Sherwood <david.sherwood@arm.com>
1559 * coretypes.h (scalar_mode): New class.
1560 * machmode.h (scalar_mode): Likewise.
1561 (scalar_mode::includes_p): New function.
1562 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
1563 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
1564 * genmodes.c (get_mode_class): Handle remaining scalar modes.
1565 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
1566 * expmed.c (store_bit_field_1): Likewise.
1567 (extract_bit_field_1): Likewise.
1568 * expr.c (write_complex_part): Likewise.
1569 (read_complex_part): Likewise.
1570 (emit_move_complex_push): Likewise.
1571 (expand_expr_real_2): Likewise.
1572 * function.c (assign_parm_setup_reg): Likewise.
1573 (assign_parms_unsplit_complex): Likewise.
1574 * optabs.c (expand_binop): Likewise.
1575 * rtlanal.c (subreg_get_info): Likewise.
1576 * simplify-rtx.c (simplify_immed_subreg): Likewise.
1577 * varasm.c (output_constant_pool_2): Likewise.
1579 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1580 Alan Hayward <alan.hayward@arm.com>
1581 David Sherwood <david.sherwood@arm.com>
1583 * expmed.c (extract_high_half): Use scalar_int_mode and remove
1585 (expmed_mult_highpart_optab): Likewise.
1586 (expmed_mult_highpart): Likewise.
1588 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1589 Alan Hayward <alan.hayward@arm.com>
1590 David Sherwood <david.sherwood@arm.com>
1592 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
1593 instead of a machine_mode.
1594 (builtin_memset_read_str): Likewise.
1595 * builtins.c (c_readstr): Likewise.
1596 (builtin_memcpy_read_str): Likewise.
1597 (builtin_strncpy_read_str): Likewise.
1598 (builtin_memset_read_str): Likewise.
1599 (builtin_memset_gen_str): Likewise.
1600 (expand_builtin_signbit): Use scalar_int_mode for local variables.
1601 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
1602 instead of a machine_mode.
1603 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
1605 (make_extraction): Likewise.
1606 (try_widen_shift_mode): Take and return scalar_int_modes instead
1608 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
1609 a scalar_int_mode instead of a machine_mode.
1610 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
1611 (avr_addr_space_pointer_mode): Likewise.
1612 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
1613 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
1614 (msp430_unwind_word_mode): Likewise.
1615 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
1616 (spu_addr_space_pointer_mode): Likewise.
1617 (spu_addr_space_address_mode): Likewise.
1618 (spu_libgcc_cmp_return_mode): Likewise.
1619 (spu_libgcc_shift_count_mode): Likewise.
1620 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
1621 (rl78_addr_space_pointer_mode): Likewise.
1622 (fl78_unwind_word_mode): Likewise.
1623 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
1625 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
1626 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
1627 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
1628 (mips_valid_pointer_mode): Likewise.
1629 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
1630 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
1631 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
1633 (ft32_addr_space_address_mode): Likewise.
1634 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
1635 scalar_int_mode instead of a machine_mode.
1636 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
1638 (m32c_addr_space_address_mode): Likewise.
1639 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
1640 (rs6000_eh_return_filter_mode): Likewise.
1641 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
1642 (rs6000_eh_return_filter_mode): Likewise.
1643 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
1644 (s390_libgcc_shift_count_mode): Likewise.
1645 (s390_unwind_word_mode): Likewise.
1646 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
1648 * target.def (mode_rep_extended): Likewise.
1649 (valid_pointer_mode): Likewise.
1650 (addr_space.valid_pointer_mode): Likewise.
1651 (eh_return_filter_mode): Return a scalar_int_mode rather than
1653 (libgcc_cmp_return_mode): Likewise.
1654 (libgcc_shift_count_mode): Likewise.
1655 (unwind_word_mode): Likewise.
1656 (addr_space.pointer_mode): Likewise.
1657 (addr_space.address_mode): Likewise.
1658 * doc/tm.texi: Regenerate.
1659 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
1661 (do_jump): Use scalar_int_mode for local variables.
1662 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
1663 rather than a machine_mode.
1664 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
1665 (scompare_loc_descriptor_wide): Likewise.
1666 (scompare_loc_descriptor_narrow): Likewise.
1667 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
1669 * except.c (sjlj_emit_dispatch_table): Likewise.
1670 (expand_builtin_eh_copy_values): Likewise.
1671 * explow.c (convert_memory_address_addr_space_1): Likewise.
1672 Take a scalar_int_mode rather than a machine_mode.
1673 (convert_memory_address_addr_space): Take a scalar_int_mode rather
1674 than a machine_mode.
1675 (memory_address_addr_space): Use scalar_int_mode for local variables.
1676 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
1677 rather than a machine_mode.
1678 * expmed.c (mask_rtx): Likewise.
1679 (init_expmed_one_conv): Likewise.
1680 (expand_mult_highpart_adjust): Likewise.
1681 (extract_high_half): Likewise.
1682 (expmed_mult_highpart_optab): Likewise.
1683 (expmed_mult_highpart): Likewise.
1684 (expand_smod_pow2): Likewise.
1685 (expand_sdiv_pow2): Likewise.
1686 (emit_store_flag_int): Likewise.
1687 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
1689 (extract_low_bits): Likewise.
1690 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
1692 * expr.c (pieces_addr::adjust): Likewise.
1693 (can_store_by_pieces): Likewise.
1694 (store_by_pieces): Likewise.
1695 (clear_by_pieces_1): Likewise.
1696 (expand_expr_addr_expr_1): Likewise.
1697 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
1698 (expand_expr_real_1): Likewise.
1699 (try_casesi): Likewise.
1700 * final.c (shorten_branches): Likewise.
1701 * fold-const.c (fold_convert_const_int_from_fixed): Change the
1702 type of "mode" to machine_mode.
1703 * internal-fn.c (expand_arith_overflow_result_store): Take a
1704 scalar_int_mode rather than a machine_mode.
1705 (expand_mul_overflow): Use scalar_int_mode for local variables.
1706 * loop-doloop.c (doloop_modify): Likewise.
1707 (doloop_optimize): Likewise.
1708 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
1709 than a machine_mode.
1710 (expand_doubleword_shift_condmove): Likewise.
1711 (expand_doubleword_shift): Likewise.
1712 (expand_doubleword_clz): Likewise.
1713 (expand_doubleword_popcount): Likewise.
1714 (expand_doubleword_parity): Likewise.
1715 (expand_absneg_bit): Use scalar_int_mode for local variables.
1716 (prepare_float_lib_cmp): Likewise.
1717 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
1718 rather than a machine_mode.
1719 (convert_memory_address_addr_space): Likewise.
1720 (get_mode_bounds): Likewise.
1721 (get_address_mode): Return a scalar_int_mode rather than a
1723 * rtlanal.c (get_address_mode): Likewise.
1724 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
1725 than a machine_mode.
1726 * targhooks.c (default_mode_rep_extended): Likewise.
1727 (default_valid_pointer_mode): Likewise.
1728 (default_addr_space_valid_pointer_mode): Likewise.
1729 (default_eh_return_filter_mode): Return a scalar_int_mode rather
1730 than a machine_mode.
1731 (default_libgcc_cmp_return_mode): Likewise.
1732 (default_libgcc_shift_count_mode): Likewise.
1733 (default_unwind_word_mode): Likewise.
1734 (default_addr_space_pointer_mode): Likewise.
1735 (default_addr_space_address_mode): Likewise.
1736 * targhooks.h (default_eh_return_filter_mode): Likewise.
1737 (default_libgcc_cmp_return_mode): Likewise.
1738 (default_libgcc_shift_count_mode): Likewise.
1739 (default_unwind_word_mode): Likewise.
1740 (default_addr_space_pointer_mode): Likewise.
1741 (default_addr_space_address_mode): Likewise.
1742 (default_mode_rep_extended): Take a scalar_int_mode rather than
1744 (default_valid_pointer_mode): Likewise.
1745 (default_addr_space_valid_pointer_mode): Likewise.
1746 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
1748 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
1749 rather than a machine_mode.
1750 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
1751 for local variables.
1752 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1753 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
1754 than a machine_mode.
1756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1757 Alan Hayward <alan.hayward@arm.com>
1758 David Sherwood <david.sherwood@arm.com>
1760 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
1761 the mode argument to scalar_int_mode.
1762 (do_jump_by_parts_zero_rtx): Likewise.
1763 (do_jump_by_parts_equality_rtx): Likewise.
1764 (do_jump_by_parts_greater): Take a mode argument.
1765 (do_jump_by_parts_equality): Likewise.
1766 (do_jump_1): Update calls accordingly.
1768 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1769 Alan Hayward <alan.hayward@arm.com>
1770 David Sherwood <david.sherwood@arm.com>
1772 * is-a.h (safe_dyn_cast): New function.
1773 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
1774 (jump_table_for_label): Likewise.
1775 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
1776 instead of an rtx_insn *.
1777 (shorten_branches): Use dyn_cast instead of LABEL_P and
1778 JUMP_TABLE_DATA_P. Use jump_table_for_label and
1779 rtx_jump_table_data::get_data_mode.
1780 (final_scan_insn): Likewise.
1782 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1783 Alan Hayward <alan.hayward@arm.com>
1784 David Sherwood <david.sherwood@arm.com>
1786 * combine.c (try_combine): Use is_a <scalar_int_mode> when
1787 trying to combine a full-register integer set with a subreg
1790 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1791 Alan Hayward <alan.hayward@arm.com>
1792 David Sherwood <david.sherwood@arm.com>
1794 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
1795 that is always either address_mode or pointer_mode.
1797 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1798 Alan Hayward <alan.hayward@arm.com>
1799 David Sherwood <david.sherwood@arm.com>
1801 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
1802 when the two are known to be equal.
1804 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1805 Alan Hayward <alan.hayward@arm.com>
1806 David Sherwood <david.sherwood@arm.com>
1808 * simplify-rtx.c (simplify_const_unary_operation): Use
1809 is_a <scalar_int_mode> instead of checking for a nonzero
1810 precision. Forcibly convert op_mode to a scalar_int_mode
1811 in that case. More clearly differentiate the operand and
1812 result modes and use the former when deciding what the value
1813 of a count-bits operation should be. Use is_int_mode instead
1814 of checking for a MODE_INT. Remove redundant check for whether
1815 this mode has a zero precision.
1817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1818 Alan Hayward <alan.hayward@arm.com>
1819 David Sherwood <david.sherwood@arm.com>
1821 * optabs.c (widen_leading): Change the type of the mode argument
1822 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
1823 (widen_bswap): Likewise.
1824 (expand_parity): Likewise.
1825 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
1826 (expand_ffs): Likewise.
1827 (epand_unop): Check for scalar integer modes before calling the
1830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1831 Alan Hayward <alan.hayward@arm.com>
1832 David Sherwood <david.sherwood@arm.com>
1834 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
1836 (expand_expr_real_1): Update call accordingly.
1838 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1839 Alan Hayward <alan.hayward@arm.com>
1840 David Sherwood <david.sherwood@arm.com>
1842 * expmed.c (store_bit_field_using_insv): Add op0_mode and
1843 value_mode arguments. Use scalar_int_mode internally.
1844 (store_bit_field_1): Rename the new integer mode from imode
1845 to op0_mode and use it instead of GET_MODE (op0). Update calls
1846 to store_split_bit_field, store_bit_field_using_insv and
1847 store_fixed_bit_field.
1848 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
1849 Use scalar_int_mode internally. Use a bit count rather than a mode
1850 when calculating the largest bit size for get_best_mode.
1851 Update calls to store_split_bit_field and store_fixed_bit_field_1.
1852 (store_fixed_bit_field_1): Add mode and value_mode arguments.
1853 Remove assertion that OP0 has a scalar integer mode.
1854 (store_split_bit_field): Add op0_mode and value_mode arguments.
1855 Update calls to extract_fixed_bit_field.
1856 (extract_bit_field_using_extv): Add an op0_mode argument.
1857 Use scalar_int_mode internally.
1858 (extract_bit_field_1): Rename the new integer mode from imode to
1859 op0_mode and use it instead of GET_MODE (op0). Update calls to
1860 extract_split_bit_field, extract_bit_field_using_extv and
1861 extract_fixed_bit_field.
1862 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
1863 to extract_split_bit_field and extract_fixed_bit_field_1.
1864 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
1865 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
1867 (extract_split_bit_field): Add an op0_mode argument. Update call
1868 to extract_fixed_bit_field.
1870 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1871 Alan Hayward <alan.hayward@arm.com>
1872 David Sherwood <david.sherwood@arm.com>
1874 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
1875 * explow.c (hard_function_value): Likewise.
1876 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
1877 convert_to_mode call outside the loop.
1878 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
1879 for the mode iterator. Require the mode specified by max_pieces
1881 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
1883 (copy_blkmode_to_reg): Likewise.
1884 (set_storage_via_setmem): Likewise.
1885 * optabs.c (prepare_cmp_insn): Likewise.
1886 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1887 * stor-layout.c (finish_bitfield_representative): Likewise.
1889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1890 Alan Hayward <alan.hayward@arm.com>
1891 David Sherwood <david.sherwood@arm.com>
1893 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
1894 * expr.c (convert_move): Use them.
1895 (convert_modes): Likewise.
1896 (store_expr_with_bounds): Likewise.
1898 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1899 Alan Hayward <alan.hayward@arm.com>
1900 David Sherwood <david.sherwood@arm.com>
1902 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
1903 parameter for the mode of "x". Remove the "known_x", "known_mode"
1904 and "known_ret" arguments. Change the type of the mode argument
1906 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
1907 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
1908 (reg_num_sign_bit_copies_for_combine): Likewise.
1909 * rtlanal.c (nonzero_bits1): Likewise.
1910 (num_sign_bit_copies1): Likewise.
1911 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
1912 (reg_num_sign_bit_copies_general): Likewise.
1913 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
1914 (reg_nonzero_bits_general): Likewise.
1916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1917 Alan Hayward <alan.hayward@arm.com>
1918 David Sherwood <david.sherwood@arm.com>
1920 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
1921 than in subroutines. Return 1 for non-integer modes.
1922 (cached_num_sign_bit_copies): Change the type of the mode parameter
1924 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
1925 classes. Handle CONST_INT_P first and then check whether X also
1926 has a scalar integer mode. Check the same thing for inner registers
1927 of a SUBREG and for values that are being extended or truncated.
1929 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1930 Alan Hayward <alan.hayward@arm.com>
1931 David Sherwood <david.sherwood@arm.com>
1933 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
1934 in subroutines. Return the mode mask for non-integer modes.
1935 (cached_nonzero_bits): Change the type of the mode parameter
1937 (nonzero_bits1): Likewise. Remove early exit for other mode
1938 classes. Handle CONST_INT_P first and then check whether X
1939 also has a scalar integer mode.
1941 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1942 Alan Hayward <alan.hayward@arm.com>
1943 David Sherwood <david.sherwood@arm.com>
1945 * expr.c (widest_int_mode_for_size): Make the comment match the code.
1946 Return a scalar_int_mode and assert that the size is greater than
1948 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
1949 (op_by_pieces_d::op_by_pieces_d): Likewise.
1950 (op_by_pieces_d::run): Likewise.
1951 (can_store_by_pieces): Likewise.
1953 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1954 Alan Hayward <alan.hayward@arm.com>
1955 David Sherwood <david.sherwood@arm.com>
1957 * combine.c (extract_left_shift): Add a mode argument and update
1959 (make_compound_operation_int): Change the type of the mode parameter
1960 to scalar_int_mode and update the call to extract_left_shift.
1962 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1963 Alan Hayward <alan.hayward@arm.com>
1964 David Sherwood <david.sherwood@arm.com>
1966 * combine.c (simplify_and_const_int): Change the type of the mode
1967 parameter to scalar_int_mode.
1968 (simplify_and_const_int_1): Likewise. Update recursive call.
1970 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1971 Alan Hayward <alan.hayward@arm.com>
1972 David Sherwood <david.sherwood@arm.com>
1974 * combine.c (simplify_compare_const): Check that the mode is a
1975 scalar_int_mode (rather than VOIDmode) before testing its
1977 (simplify_comparison): Move COMPARISON_P handling out of the
1978 loop and restrict the latter part of the loop to scalar_int_modes.
1979 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
1980 and when considering SUBREG_REGs. Use is_int_mode instead of
1981 checking GET_MODE_CLASS against MODE_INT.
1983 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1984 Alan Hayward <alan.hayward@arm.com>
1985 David Sherwood <david.sherwood@arm.com>
1987 * combine.c (try_widen_shift_mode): Move check for equal modes to...
1988 (simplify_shift_const_1): ...here. Use scalar_int_mode for
1989 shift_unit_mode and for modes involved in scalar shifts.
1991 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1992 Alan Hayward <alan.hayward@arm.com>
1993 David Sherwood <david.sherwood@arm.com>
1995 * combine.c (force_int_to_mode): New function, split out from...
1996 (force_to_mode): ...here. Keep xmode up-to-date and use it
1997 instead of GET_MODE (x).
1999 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2000 Alan Hayward <alan.hayward@arm.com>
2001 David Sherwood <david.sherwood@arm.com>
2003 * optabs-query.h (extraction_insn::struct_mode): Change type to
2004 opt_scalar_int_mode and update comment.
2005 (extraction_insn::field_mode): Change type to scalar_int_mode.
2006 (extraction_insn::pos_mode): Likewise.
2007 * combine.c (make_extraction): Update accordingly.
2008 * optabs-query.c (get_traditional_extraction_insn): Likewise.
2009 (get_optab_extraction_insn): Likewise.
2010 * recog.c (simplify_while_replacing): Likewise.
2011 * expmed.c (narrow_bit_field_mem): Change the type of the mode
2012 parameter to opt_scalar_int_mode.
2014 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2015 Alan Hayward <alan.hayward@arm.com>
2016 David Sherwood <david.sherwood@arm.com>
2018 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
2019 to a scalar_int_mode instead of a machine_mode.
2020 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
2021 (get_best_mode): Return a boolean and use a pointer argument to store
2022 the selected mode. Replace the limit mode parameter with a bit limit.
2023 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
2024 for the values returned by bit_field_mode_iterator::next_mode.
2025 (store_bit_field): Update call to get_best_mode.
2026 (store_fixed_bit_field): Likewise.
2027 (extract_fixed_bit_field): Likewise.
2028 * expr.c (optimize_bitfield_assignment_op): Likewise.
2029 * fold-const.c (optimize_bit_field_compare): Likewise.
2030 (fold_truth_andor_1): Likewise.
2031 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
2032 Update for new type of m_mode.
2033 (get_best_mode): As above.
2035 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2036 Alan Hayward <alan.hayward@arm.com>
2037 David Sherwood <david.sherwood@arm.com>
2039 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
2040 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
2041 (store_bit_field): Check is_a <scalar_int_mode> before calling
2042 strict_volatile_bitfield_p.
2043 (extract_bit_field): Likewise.
2045 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2046 Alan Hayward <alan.hayward@arm.com>
2047 David Sherwood <david.sherwood@arm.com>
2049 * target.def (cstore_mode): Return a scalar_int_mode.
2050 * doc/tm.texi: Regenerate.
2051 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
2052 * targhooks.h (default_cstore_mode): Likewise.
2053 * targhooks.c (default_cstore_mode): Likewise, using a forced
2055 * expmed.c (emit_cstore): Expect the target of the cstore to be
2058 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2059 Alan Hayward <alan.hayward@arm.com>
2060 David Sherwood <david.sherwood@arm.com>
2062 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
2064 (niter_desc): Likewise mode.
2065 (iv_analyze): Add a mode parameter.
2067 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
2068 and change its type to scalar_int_mode.
2069 * loop-iv.c: Update commentary at head of file.
2070 (iv_constant): Pass the mode paraeter before the rtx it describes
2071 and change its type to scalar_int_mode. Remove VOIDmode handling.
2072 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
2073 (iv_extend): Likewise.
2074 (shorten_into_mode): Likewise.
2075 (iv_add): Use scalar_int_mode.
2076 (iv_mult): Likewise.
2077 (iv_shift): Likewise.
2078 (canonicalize_iv_subregs): Likewise.
2079 (get_biv_step_1): Pass the outer_mode parameter before the rtx
2080 it describes and change its mode to scalar_int_mode. Also change
2081 the type of the returned inner_mode to scalar_int_mode.
2082 (get_biv_step): Likewise, turning outer_mode from a pointer
2083 into a direct parameter. Update call to get_biv_step_1.
2084 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
2085 iv_constant and get_biv_step.
2086 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
2087 and change its type to scalar_int_mode. Don't initialise iv->mode
2088 to VOIDmode and remove later checks for its still being VOIDmode.
2089 Update calls to iv_analyze_op and iv_analyze_expr. Check
2090 is_a <scalar_int_mode> when changing the mode under consideration.
2091 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
2092 Update call to iv_analyze_expr.
2093 (iv_analyze_op): Add a mode parameter. Reject subregs whose
2094 inner register is not also a scalar_int_mode. Update call to
2096 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
2097 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
2098 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
2099 separate mode class checks. Update calls to iv_analyze. Remove
2100 fix-up of VOIDmodes after iv_analyze_biv.
2101 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
2102 don't have a scalar_int_mode. Update call to biv_p.
2104 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2105 Alan Hayward <alan.hayward@arm.com>
2106 David Sherwood <david.sherwood@arm.com>
2108 * cfgexpand.c (convert_debug_memory_address): Use
2109 as_a <scalar_int_mode>.
2110 * combine.c (expand_compound_operation): Likewise.
2111 (make_extraction): Likewise.
2112 (change_zero_ext): Likewise.
2113 (simplify_comparison): Likewise.
2114 * cse.c (cse_insn): Likewise.
2115 * dwarf2out.c (minmax_loc_descriptor): Likewise.
2116 (mem_loc_descriptor): Likewise.
2117 (loc_descriptor): Likewise.
2118 * expmed.c (init_expmed_one_mode): Likewise.
2119 (synth_mult): Likewise.
2120 (emit_store_flag_1): Likewise.
2121 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
2122 of a comparison with size.
2123 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
2124 (reduce_to_bit_field_precision): Likewise.
2125 * function.c (expand_function_end): Likewise.
2126 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
2127 * loop-doloop.c (doloop_modify): Likewise.
2128 * optabs.c (expand_binop): Likewise.
2129 (expand_unop): Likewise.
2130 (expand_copysign_absneg): Likewise.
2131 (prepare_cmp_insn): Likewise.
2132 (maybe_legitimize_operand): Likewise.
2133 * recog.c (const_scalar_int_operand): Likewise.
2134 * rtlanal.c (get_address_mode): Likewise.
2135 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2136 (simplify_cond_clz_ctz): Likewise.
2137 * tree-nested.c (get_nl_goto_field): Likewise.
2138 * tree.c (build_vector_type_for_mode): Likewise.
2139 * var-tracking.c (use_narrower_mode): Likewise.
2141 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2142 Alan Hayward <alan.hayward@arm.com>
2143 David Sherwood <david.sherwood@arm.com>
2145 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
2146 * builtins.c (expand_builtin_signbit): Use it.
2147 * cfgexpand.c (expand_debug_expr): Likewise.
2148 * dojump.c (do_jump): Likewise.
2149 (do_compare_and_jump): Likewise.
2150 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
2151 * expmed.c (make_tree): Likewise.
2152 * expr.c (expand_expr_real_2): Likewise.
2153 (expand_expr_real_1): Likewise.
2154 (try_casesi): Likewise.
2155 * fold-const-call.c (fold_const_call_ss): Likewise.
2156 * fold-const.c (unextend): Likewise.
2157 (extract_muldiv_1): Likewise.
2158 (fold_single_bit_test): Likewise.
2159 (native_encode_int): Likewise.
2160 (native_encode_string): Likewise.
2161 (native_interpret_int): Likewise.
2162 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
2163 * internal-fn.c (expand_addsub_overflow): Likewise.
2164 (expand_neg_overflow): Likewise.
2165 (expand_mul_overflow): Likewise.
2166 (expand_arith_overflow): Likewise.
2167 * match.pd: Likewise.
2168 * stor-layout.c (layout_type): Likewise.
2169 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
2170 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
2171 * tree-ssanames.c (get_range_info): Likewise.
2172 * tree-switch-conversion.c (array_value_type) Likewise.
2173 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
2174 (vect_recog_divmod_pattern): Likewise.
2175 (vect_recog_mixed_size_cond_pattern): Likewise.
2176 * tree-vrp.c (extract_range_basic): Likewise.
2177 (simplify_float_conversion_using_ranges): Likewise.
2178 * tree.c (int_fits_type_p): Likewise.
2179 * ubsan.c (instrument_bool_enum_load): Likewise.
2180 * varasm.c (mergeable_string_section): Likewise.
2181 (narrowing_initializer_constant_valid_p): Likewise.
2182 (output_constant): Likewise.
2184 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2185 Alan Hayward <alan.hayward@arm.com>
2186 David Sherwood <david.sherwood@arm.com>
2188 * machmode.h (NARROWEST_INT_MODE): New macro.
2189 * expr.c (alignment_for_piecewise_move): Use it instead of
2190 GET_CLASS_NARROWEST_MODE (MODE_INT).
2191 (push_block): Likewise.
2192 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
2194 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2196 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2197 Alan Hayward <alan.hayward@arm.com>
2198 David Sherwood <david.sherwood@arm.com>
2200 * postreload.c (move2add_valid_value_p): Change the type of the
2201 mode parameter to scalar_int_mode.
2202 (move2add_use_add2_insn): Add a mode parameter and use it instead
2204 (move2add_use_add3_insn): Likewise.
2205 (reload_cse_move2add): Update accordingly.
2207 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2208 Alan Hayward <alan.hayward@arm.com>
2209 David Sherwood <david.sherwood@arm.com>
2211 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
2213 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
2214 * optabs.c (expand_unop): Likewise.
2216 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2217 Alan Hayward <alan.hayward@arm.com>
2218 David Sherwood <david.sherwood@arm.com>
2220 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
2221 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
2222 (popcount_loc_descriptor): Likewise.
2223 (bswap_loc_descriptor): Likewise.
2224 (rotate_loc_descriptor): Likewise.
2225 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
2226 calling the functions above.
2228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2229 Alan Hayward <alan.hayward@arm.com>
2230 David Sherwood <david.sherwood@arm.com>
2232 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
2234 (try_combine): Likewise.
2235 (simplify_if_then_else): Likewise.
2236 * cse.c (cse_insn): Likewise.
2237 * dwarf2out.c (mem_loc_descriptor): Likewise.
2238 * emit-rtl.c (gen_lowpart_common): Likewise.
2239 * simplify-rtx.c (simplify_truncation): Likewise.
2240 (simplify_binary_operation_1): Likewise.
2241 (simplify_const_relational_operation): Likewise.
2242 (simplify_ternary_operation): Likewise.
2243 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
2245 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2246 Alan Hayward <alan.hayward@arm.com>
2247 David Sherwood <david.sherwood@arm.com>
2249 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
2250 * reload.c (push_reload): Likewise.
2251 (find_reloads): Likewise.
2253 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2254 Alan Hayward <alan.hayward@arm.com>
2255 David Sherwood <david.sherwood@arm.com>
2257 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
2258 (make_compound_operation_int): Likewise.
2259 (change_zero_ext): Likewise.
2260 * expr.c (convert_move): Likewise.
2261 (convert_modes): Likewise.
2262 * fwprop.c (forward_propagate_subreg): Likewise.
2263 * loop-iv.c (get_biv_step_1): Likewise.
2264 * optabs.c (widen_operand): Likewise.
2265 * postreload.c (move2add_valid_value_p): Likewise.
2266 * recog.c (simplify_while_replacing): Likewise.
2267 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2268 (simplify_binary_operation_1): Likewise. Remove redundant
2269 mode equality check.
2271 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2272 Alan Hayward <alan.hayward@arm.com>
2273 David Sherwood <david.sherwood@arm.com>
2275 * combine.c (combine_simplify_rtx): Add checks for
2276 is_a <scalar_int_mode>.
2277 (simplify_if_then_else): Likewise.
2278 (make_field_assignment): Likewise.
2279 (simplify_comparison): Likewise.
2280 * ifcvt.c (noce_try_bitop): Likewise.
2281 * loop-invariant.c (canonicalize_address_mult): Likewise.
2282 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2285 Alan Hayward <alan.hayward@arm.com>
2286 David Sherwood <david.sherwood@arm.com>
2288 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
2289 is_a <scalar_int_mode> instead of != BLKmode.
2291 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2292 Alan Hayward <alan.hayward@arm.com>
2293 David Sherwood <david.sherwood@arm.com>
2295 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
2296 instead of != VOIDmode.
2297 * combine.c (if_then_else_cond): Likewise.
2298 (change_zero_ext): Likewise.
2299 * dwarf2out.c (mem_loc_descriptor): Likewise.
2300 (loc_descriptor): Likewise.
2301 * rtlanal.c (canonicalize_condition): Likewise.
2302 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
2304 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2305 Alan Hayward <alan.hayward@arm.com>
2306 David Sherwood <david.sherwood@arm.com>
2308 * simplify-rtx.c (simplify_binary_operation_1): Use
2309 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
2311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2312 Alan Hayward <alan.hayward@arm.com>
2313 David Sherwood <david.sherwood@arm.com>
2315 * wide-int.h (int_traits<unsigned char>) New class.
2316 (int_traits<unsigned short>) Likewise.
2317 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
2318 Use GET_MODE_UNIT_PRECISION and remove redundant test for
2320 * combine.c (set_nonzero_bits_and_sign_copies): Use
2321 is_a <scalar_int_mode>.
2322 (find_split_point): Likewise.
2323 (combine_simplify_rtx): Likewise.
2324 (simplify_logical): Likewise.
2325 (expand_compound_operation): Likewise.
2326 (expand_field_assignment): Likewise.
2327 (make_compound_operation): Likewise.
2328 (extended_count): Likewise.
2329 (change_zero_ext): Likewise.
2330 (simplify_comparison): Likewise.
2331 * dwarf2out.c (scompare_loc_descriptor): Likewise.
2332 (ucompare_loc_descriptor): Likewise.
2333 (minmax_loc_descriptor): Likewise.
2334 (mem_loc_descriptor): Likewise.
2335 (loc_descriptor): Likewise.
2336 * expmed.c (init_expmed_one_mode): Likewise.
2337 * lra-constraints.c (lra_constraint_offset): Likewise.
2338 * optabs.c (prepare_libcall_arg): Likewise.
2339 * postreload.c (move2add_note_store): Likewise.
2340 * reload.c (operands_match_p): Likewise.
2341 * rtl.h (load_extend_op): Likewise.
2342 * rtlhooks.c (gen_lowpart_general): Likewise.
2343 * simplify-rtx.c (simplify_truncation): Likewise.
2344 (simplify_unary_operation_1): Likewise.
2345 (simplify_binary_operation_1): Likewise.
2346 (simplify_const_binary_operation): Likewise.
2347 (simplify_const_relational_operation): Likewise.
2348 (simplify_subreg): Likewise.
2349 * stor-layout.c (bitwise_mode_for_mode): Likewise.
2350 * var-tracking.c (adjust_mems): Likewise.
2351 (prepare_call_arguments): Likewise.
2353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2354 Alan Hayward <alan.hayward@arm.com>
2355 David Sherwood <david.sherwood@arm.com>
2357 * machmode.h (is_int_mode): New fuction.
2358 * combine.c (find_split_point): Use it.
2359 (combine_simplify_rtx): Likewise.
2360 (simplify_if_then_else): Likewise.
2361 (simplify_set): Likewise.
2362 (simplify_shift_const_1): Likewise.
2363 (simplify_comparison): Likewise.
2364 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
2365 * cse.c (notreg_cost): Likewise.
2366 (cse_insn): Likewise.
2367 * cselib.c (cselib_lookup_1): Likewise.
2368 * dojump.c (do_jump_1): Likewise.
2369 (do_compare_rtx_and_jump): Likewise.
2370 * dse.c (get_call_args): Likewise.
2371 * dwarf2out.c (rtl_for_decl_init): Likewise.
2372 (native_encode_initializer): Likewise.
2373 * expmed.c (emit_store_flag_1): Likewise.
2374 (emit_store_flag): Likewise.
2375 * expr.c (convert_modes): Likewise.
2376 (store_field): Likewise.
2377 (expand_expr_real_1): Likewise.
2378 * fold-const.c (fold_read_from_constant_string): Likewise.
2379 * gimple-ssa-sprintf.c (get_format_string): Likewise.
2380 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
2381 * optabs.c (expand_binop): Likewise.
2382 (expand_unop): Likewise.
2383 (expand_abs_nojump): Likewise.
2384 (expand_one_cmpl_abs_nojump): Likewise.
2385 * simplify-rtx.c (mode_signbit_p): Likewise.
2386 (val_signbit_p): Likewise.
2387 (val_signbit_known_set_p): Likewise.
2388 (val_signbit_known_clear_p): Likewise.
2389 (simplify_relational_operation_1): Likewise.
2390 * tree.c (vector_type_mode): Likewise.
2392 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2393 Alan Hayward <alan.hayward@arm.com>
2394 David Sherwood <david.sherwood@arm.com>
2396 * machmode.h (smallest_mode_for_size): Fix formatting.
2397 (smallest_int_mode_for_size): New function.
2398 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
2399 instead of smallest_mode_for_size.
2400 * combine.c (make_extraction): Likewise.
2401 * config/arc/arc.c (arc_expand_movmem): Likewise.
2402 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
2403 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
2404 * config/s390/s390.c (s390_expand_insv): Likewise.
2405 * config/sparc/sparc.c (assign_int_registers): Likewise.
2406 * config/spu/spu.c (spu_function_value): Likewise.
2407 (spu_function_arg): Likewise.
2408 * coverage.c (get_gcov_type): Likewise.
2409 (get_gcov_unsigned_t): Likewise.
2410 * dse.c (find_shift_sequence): Likewise.
2411 * expmed.c (store_bit_field_1): Likewise.
2412 * expr.c (convert_move): Likewise.
2413 (store_field): Likewise.
2414 * internal-fn.c (expand_arith_overflow): Likewise.
2415 * optabs-query.c (get_best_extraction_insn): Likewise.
2416 * optabs.c (expand_twoval_binop_libfunc): Likewise.
2417 * stor-layout.c (layout_type): Likewise.
2418 (initialize_sizetypes): Likewise.
2419 * targhooks.c (default_get_mask_mode): Likewise.
2420 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
2422 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2423 Alan Hayward <alan.hayward@arm.com>
2424 David Sherwood <david.sherwood@arm.com>
2426 * machmode.h (opt_mode::else_blk): New function.
2427 (int_mode_for_mode): Declare.
2428 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
2429 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
2431 * cfgexpand.c (expand_debug_expr): Likewise.
2432 * combine.c (gen_lowpart_or_truncate): Likewise.
2433 (gen_lowpart_for_combine): Likewise.
2434 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
2435 * config/avr/avr.c (avr_to_int_mode): Likewise.
2436 (avr_out_plus_1): Likewise.
2437 (avr_out_plus): Likewise.
2438 (avr_out_round): Likewise.
2439 * config/i386/i386.c (ix86_split_to_parts): Likewise.
2440 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
2441 (s390_expand_vcond): Likewise.
2442 * config/spu/spu.c (spu_split_immediate): Likewise.
2443 (spu_expand_mov): Likewise.
2444 * dse.c (get_stored_val): Likewise.
2445 * expmed.c (store_bit_field_1): Likewise.
2446 (convert_extracted_bit_field): Use int_mode_for_mode instead of
2448 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
2449 (extract_low_bits): Likewise.
2450 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
2451 handling rather than repeating the check.
2452 (emit_group_store): Likewise.
2453 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
2454 * optabs.c (expand_absneg_bit): Likewise.
2455 (expand_copysign_absneg): Likewise.
2456 (expand_copysign_bit): Likewise.
2457 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
2458 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
2459 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
2460 * var-tracking.c (prepare_call_arguments): Likewise.
2461 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
2462 int_mode_for_mode instead of mode_for_size.
2463 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
2465 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2466 Alan Hayward <alan.hayward@arm.com>
2467 David Sherwood <david.sherwood@arm.com>
2469 * machmode.h (int_mode_for_size): New function.
2470 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
2471 instead of mode_for_size.
2472 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
2474 * combine.c (expand_field_assignment): Use int_mode_for_size
2475 instead of mode_for_size.
2476 (make_extraction): Likewise.
2477 (simplify_shift_const_1): Likewise.
2478 (simplify_comparison): Likewise.
2479 * dojump.c (do_jump): Likewise.
2480 * dwarf2out.c (mem_loc_descriptor): Likewise.
2481 * emit-rtl.c (init_derived_machine_modes): Likewise.
2482 * expmed.c (flip_storage_order): Likewise.
2483 (convert_extracted_bit_field): Likewise.
2484 * expr.c (copy_blkmode_from_reg): Likewise.
2485 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
2486 * internal-fn.c (expand_mul_overflow): Likewise.
2487 * lower-subreg.c (simple_move): Likewise.
2488 * optabs-libfuncs.c (init_optabs): Likewise.
2489 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2490 * tree.c (vector_type_mode): Likewise.
2491 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
2492 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
2493 * tree-vect-generic.c (expand_vector_parallel): Likewise.
2494 * tree-vect-stmts.c (vectorizable_load): Likewise.
2495 (vectorizable_store): Likewise.
2497 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2498 Alan Hayward <alan.hayward@arm.com>
2499 David Sherwood <david.sherwood@arm.com>
2501 * coretypes.h (pod_mode): New type.
2502 (scalar_int_mode_pod): New typedef.
2503 * machmode.h (pod_mode): New class.
2504 (int_n_data_t::m): Change type to scalar_int_mode_pod.
2505 * genmodes.c (emit_mode_int_n): Update accordingly.
2506 * lower-subreg.h (target_lower_subreg): Change type to
2507 scalar_int_mode_pod.
2508 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
2509 scalar_int_mode_pod.
2511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2512 Alan Hayward <alan.hayward@arm.com>
2513 David Sherwood <david.sherwood@arm.com>
2515 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
2516 machine_mode to scalar_int_mode.
2517 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
2518 (rs6000_option_override_internal): Remove cast to int.
2519 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
2520 machine_mode to scalar_int_mode.
2521 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
2522 (rs6000_option_override_internal): Remove cast to int.
2523 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
2524 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
2526 * config/s390/s390.c (s390_expand_builtin): Likewise.
2527 * coretypes.h (scalar_int_mode): New type.
2528 (opt_scalar_int_mode): New typedef.
2529 * machmode.h (scalar_int_mode): New class.
2530 (scalar_int_mode::includes_p): New function.
2531 (byte_mode): Change type to scalar_int_mode.
2532 (word_mode): Likewise.
2533 (ptr_mode): Likewise.
2534 * emit-rtl.c (byte_mode): Likewise.
2535 (word_mode): Likewise.
2536 (ptr_mode): Likewise.
2537 (init_derived_machine_modes): Update accordingly.
2538 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
2539 and MODE_PARTIAL_INT.
2540 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
2541 opt_scalar_int_mode.
2543 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2544 Alan Hayward <alan.hayward@arm.com>
2545 David Sherwood <david.sherwood@arm.com>
2547 * target.def (libgcc_floating_mode_supported_p): Take a
2549 * doc/tm.texi: Regenerate.
2550 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
2552 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
2553 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
2556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2557 Alan Hayward <alan.hayward@arm.com>
2558 David Sherwood <david.sherwood@arm.com>
2560 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
2561 * doc/tm.texi: Regenerate.
2562 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
2563 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
2564 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
2565 * targhooks.h (default_floatn_mode): Likewise.
2566 * targhooks.c (default_floatn_mode): Likewise.
2567 * tree.c (build_common_tree_nodes): Update accordingly.
2569 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2570 Alan Hayward <alan.hayward@arm.com>
2571 David Sherwood <david.sherwood@arm.com>
2573 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
2574 (mode_iterator::iterate_p): Likewise.
2575 (mode_iterator::get_wider): Likewise.
2576 * expr.c (init_expr_target): Use opt_scalar_float_mode.
2578 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2579 Alan Hayward <alan.hayward@arm.com>
2580 David Sherwood <david.sherwood@arm.com>
2582 * coretypes.h (opt_scalar_float_mode): New typedef.
2583 * machmode.h (float_mode_for_size): New function.
2584 * emit-rtl.c (double_mode): Delete.
2585 (init_emit_once): Use float_mode_for_size.
2586 * stor-layout.c (layout_type): Likewise.
2587 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
2589 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2590 Alan Hayward <alan.hayward@arm.com>
2591 David Sherwood <david.sherwood@arm.com>
2593 * output.h (assemble_real): Take a scalar_float_mode.
2594 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
2595 * config/arm/arm.md (consttable_4): Likewise.
2596 (consttable_8): Likewise.
2597 (consttable_16): Likewise.
2598 * config/mips/mips.md (consttable_float): Likewise.
2599 * config/s390/s390.c (s390_output_pool_entry): Likewise.
2600 * varasm.c (assemble_real): Take a scalar_float_mode.
2601 (output_constant_pool_2): Update accordingly.
2602 (output_constant): Likewise.
2604 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2605 Alan Hayward <alan.hayward@arm.com>
2606 David Sherwood <david.sherwood@arm.com>
2608 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
2609 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
2610 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
2611 (native_encode_real): Likewise.
2612 (native_interpret_real): Likewise.
2613 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
2614 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2616 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2617 Alan Hayward <alan.hayward@arm.com>
2618 David Sherwood <david.sherwood@arm.com>
2620 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
2621 <scalar_float_mode>. Simplify.
2622 (gen_extend_conv_libfunc): Likewise.
2624 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2625 Alan Hayward <alan.hayward@arm.com>
2626 David Sherwood <david.sherwood@arm.com>
2628 * coretypes.h (scalar_float_mode): New type.
2629 * machmode.h (mode_traits::from_int): Use machine_mode if
2630 USE_ENUM_MODES is defined.
2631 (is_a): New function.
2633 (dyn_cast): Likewise.
2634 (scalar_float_mode): New class.
2635 (scalar_float_mode::includes_p): New function.
2636 (is_float_mode): Likewise.
2637 * gdbhooks.py (MachineModePrinter): New class.
2638 (build_pretty_printer): Use it for scalar_float_mode.
2639 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
2640 (format_helper::format_helper): Turn into a template.
2641 * genmodes.c (get_mode_class): New function.
2642 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
2643 or machine_mode if none.
2644 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
2645 as_a <scalar_float_mode>.
2646 * dwarf2out.c (mem_loc_descriptor): Likewise.
2647 (insert_float): Likewise.
2648 (add_const_value_attribute): Likewise.
2649 * simplify-rtx.c (simplify_immed_subreg): Likewise.
2650 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
2651 (expand_unop): Update accordingly.
2652 (expand_abs_nojump): Likewise.
2653 (expand_copysign_absneg): Take a scalar_float_mode.
2654 (expand_copysign_bit): Likewise.
2655 (expand_copysign): Update accordingly.
2657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2658 Alan Hayward <alan.hayward@arm.com>
2659 David Sherwood <david.sherwood@arm.com>
2661 * coretypes.h (opt_mode): New class.
2662 * machmode.h (opt_mode): Likewise.
2663 (opt_mode::else_void): New function.
2664 (opt_mode::require): Likewise.
2665 (opt_mode::exists): Likewise.
2666 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
2667 (GET_MODE_2XWIDER_MODE): Likewise.
2668 (mode_iterator::get_wider): Update accordingly.
2669 (mode_iterator::get_2xwider): Likewise.
2670 (mode_iterator::get_known_wider): Likewise, turning into a template.
2671 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
2672 forcing a wider mode to exist.
2673 * config/cr16/cr16.h (LONG_REG_P): Likewise.
2674 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2675 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
2676 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
2677 * lower-subreg.c (init_lower_subreg): Likewise.
2678 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
2679 on the final iteration.
2680 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
2681 a wider mode exists before asking for a move pattern.
2682 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
2683 forcing a wider mode to exist.
2684 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
2685 returning false if no such mode exists.
2686 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
2687 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
2688 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
2689 Avoid checking for a MODE_INT if we already know the mode is not a
2691 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
2692 forcing a wider mode to exist.
2693 (expmed_mult_highpart_optab): Likewise.
2694 (expmed_mult_highpart): Likewise.
2695 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
2697 * lto-streamer-in.c (lto_input_mode_table): Likewise.
2698 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
2699 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
2700 * internal-fn.c (expand_mul_overflow): Update use of
2701 GET_MODE_2XWIDER_MODE.
2702 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2703 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
2704 GET_MODE_WIDER_MODE.
2705 (convert_plusminus_to_widen): Likewise.
2706 * tree-switch-conversion.c (array_value_type): Likewise.
2707 * var-tracking.c (emit_note_insn_var_location): Likewise.
2708 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
2709 Return false inside rather than outside the loop if no wider mode
2711 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
2712 and GET_MODE_2XWIDER_MODE
2713 (can_compare_p): Use else_void.
2714 * gdbhooks.py (OptMachineModePrinter): New class.
2715 (build_pretty_printer): Use it for opt_mode.
2717 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2718 Alan Hayward <alan.hayward@arm.com>
2719 David Sherwood <david.sherwood@arm.com>
2721 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
2722 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
2724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2725 Alan Hayward <alan.hayward@arm.com>
2726 David Sherwood <david.sherwood@arm.com>
2728 * machmode.h (mode_traits): New structure.
2729 (get_narrowest_mode): New function.
2730 (mode_iterator::start): Likewise.
2731 (mode_iterator::iterate_p): Likewise.
2732 (mode_iterator::get_wider): Likewise.
2733 (mode_iterator::get_known_wider): Likewise.
2734 (mode_iterator::get_2xwider): Likewise.
2735 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
2736 (FOR_EACH_MODE): Likewise.
2737 (FOR_EACH_MODE_FROM): Likewise.
2738 (FOR_EACH_MODE_UNTIL): Likewise.
2739 (FOR_EACH_WIDER_MODE): Likewise.
2740 (FOR_EACH_2XWIDER_MODE): Likewise.
2741 * builtins.c (expand_builtin_strlen): Use new mode iterators.
2742 * combine.c (simplify_comparison): Likewise
2743 * config/i386/i386.c (type_natural_mode): Likewise.
2744 * cse.c (cse_insn): Likewise.
2745 * dse.c (find_shift_sequence): Likewise.
2746 * emit-rtl.c (init_derived_machine_modes): Likewise.
2747 (init_emit_once): Likewise.
2748 * explow.c (hard_function_value): Likewise.
2749 * expmed.c (extract_fixed_bit_field_1): Likewise.
2750 (extract_bit_field_1): Likewise.
2751 (expand_divmod): Likewise.
2752 (emit_store_flag_1): Likewise.
2753 * expr.c (init_expr_target): Likewise.
2754 (convert_move): Likewise.
2755 (alignment_for_piecewise_move): Likewise.
2756 (widest_int_mode_for_size): Likewise.
2757 (emit_block_move_via_movmem): Likewise.
2758 (copy_blkmode_to_reg): Likewise.
2759 (set_storage_via_setmem): Likewise.
2760 (compress_float_constant): Likewise.
2761 * omp-low.c (omp_clause_aligned_alignment): Likewise.
2762 * optabs-query.c (get_best_extraction_insn): Likewise.
2763 * optabs.c (expand_binop): Likewise.
2764 (expand_twoval_unop): Likewise.
2765 (expand_twoval_binop): Likewise.
2766 (widen_leading): Likewise.
2767 (widen_bswap): Likewise.
2768 (expand_parity): Likewise.
2769 (expand_unop): Likewise.
2770 (prepare_cmp_insn): Likewise.
2771 (prepare_float_lib_cmp): Likewise.
2772 (expand_float): Likewise.
2773 (expand_fix): Likewise.
2774 (expand_sfix_optab): Likewise.
2775 * postreload.c (move2add_use_add2_insn): Likewise.
2776 * reg-stack.c (reg_to_stack): Likewise.
2777 * reginfo.c (choose_hard_reg_mode): Likewise.
2778 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
2779 * stor-layout.c (mode_for_size): Likewise.
2780 (smallest_mode_for_size): Likewise.
2781 (mode_for_vector): Likewise.
2782 (finish_bitfield_representative): Likewise.
2783 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
2784 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
2785 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2786 * var-tracking.c (prepare_call_arguments): Likewise.
2788 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2789 Alan Hayward <alan.hayward@arm.com>
2790 David Sherwood <david.sherwood@arm.com>
2792 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
2793 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
2794 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
2795 * machmode.h (mode_size): Move earlier in file.
2796 (mode_precision): Likewise.
2797 (mode_inner): Likewise.
2798 (mode_nunits): Likewise.
2799 (mode_unit_size): Likewise.
2800 (unit_unit_precision): Likewise.
2801 (mode_wider): Likewise.
2802 (mode_2xwider): Likewise.
2803 (machine_mode): New class.
2804 (mode_to_bytes): New function.
2805 (mode_to_bits): Likewise.
2806 (mode_to_precision): Likewise.
2807 (mode_to_inner): Likewise.
2808 (mode_to_unit_size): Likewise.
2809 (mode_to_unit_precision): Likewise.
2810 (mode_to_nunits): Likewise.
2811 (GET_MODE_SIZE): Use mode_to_bytes.
2812 (GET_MODE_BITSIZE): Use mode_to_bits.
2813 (GET_MODE_PRECISION): Use mode_to_precision.
2814 (GET_MODE_INNER): Use mode_to_inner.
2815 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
2816 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
2817 (GET_MODE_NUNITS): Use mode_to_nunits.
2818 * system.h (ALWAYS_INLINE): New macro.
2819 * config/powerpcspe/powerpcspe-c.c
2820 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
2821 int for arg1_mode and arg2_mode.
2823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2824 Alan Hayward <alan.hayward@arm.com>
2825 David Sherwood <david.sherwood@arm.com>
2827 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
2828 Prefix mode names with E_ in case statements.
2829 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2830 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
2831 (aarch64_split_simd_move): Likewise.
2832 (aarch64_gen_storewb_pair): Likewise.
2833 (aarch64_gen_loadwb_pair): Likewise.
2834 (aarch64_gen_store_pair): Likewise.
2835 (aarch64_gen_load_pair): Likewise.
2836 (aarch64_get_condition_code_1): Likewise.
2837 (aarch64_constant_pool_reload_icode): Likewise.
2838 (get_rsqrte_type): Likewise.
2839 (get_rsqrts_type): Likewise.
2840 (get_recpe_type): Likewise.
2841 (get_recps_type): Likewise.
2842 (aarch64_gimplify_va_arg_expr): Likewise.
2843 (aarch64_simd_container_mode): Likewise.
2844 (aarch64_emit_load_exclusive): Likewise.
2845 (aarch64_emit_store_exclusive): Likewise.
2846 (aarch64_expand_compare_and_swap): Likewise.
2847 (aarch64_gen_atomic_cas): Likewise.
2848 (aarch64_emit_bic): Likewise.
2849 (aarch64_emit_atomic_swap): Likewise.
2850 (aarch64_emit_atomic_load_op): Likewise.
2851 (aarch64_evpc_trn): Likewise.
2852 (aarch64_evpc_uzp): Likewise.
2853 (aarch64_evpc_zip): Likewise.
2854 (aarch64_evpc_ext): Likewise.
2855 (aarch64_evpc_rev): Likewise.
2856 (aarch64_evpc_dup): Likewise.
2857 (aarch64_gen_ccmp_first): Likewise.
2858 (aarch64_gen_ccmp_next): Likewise.
2859 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
2860 (alpha_emit_xfloating_libcall): Likewise.
2861 (emit_insxl): Likewise.
2862 (alpha_arg_type): Likewise.
2863 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
2864 (arc_preferred_simd_mode): Likewise.
2865 (arc_secondary_reload): Likewise.
2866 (get_arc_condition_code): Likewise.
2867 (arc_print_operand): Likewise.
2868 (arc_legitimate_constant_p): Likewise.
2869 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2870 * config/arc/arc.md (casesi_load): Likewise.
2871 (casesi_compact_jump): Likewise.
2872 * config/arc/predicates.md (proper_comparison_operator): Likewise.
2873 (cc_use_register): Likewise.
2874 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
2875 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
2876 (arm_init_iwmmxt_builtins): Likewise.
2877 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
2878 (neon_expand_vector_init): Likewise.
2879 (arm_attr_length_move_neon): Likewise.
2880 (maybe_get_arm_condition_code): Likewise.
2881 (arm_emit_vector_const): Likewise.
2882 (arm_preferred_simd_mode): Likewise.
2883 (arm_output_iwmmxt_tinsr): Likewise.
2884 (thumb1_output_casesi): Likewise.
2885 (thumb2_output_casesi): Likewise.
2886 (arm_emit_load_exclusive): Likewise.
2887 (arm_emit_store_exclusive): Likewise.
2888 (arm_expand_compare_and_swap): Likewise.
2889 (arm_evpc_neon_vuzp): Likewise.
2890 (arm_evpc_neon_vzip): Likewise.
2891 (arm_evpc_neon_vrev): Likewise.
2892 (arm_evpc_neon_vtrn): Likewise.
2893 (arm_evpc_neon_vext): Likewise.
2894 (arm_validize_comparison): Likewise.
2895 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
2896 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
2897 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
2898 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
2899 (c6x_preferred_simd_mode): Likewise.
2900 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
2901 (epiphany_rtx_costs): Likewise.
2902 * config/epiphany/predicates.md (proper_comparison_operator):
2904 * config/frv/frv.c (condexec_memory_operand): Likewise.
2905 (frv_emit_move): Likewise.
2906 (output_move_single): Likewise.
2907 (output_condmove_single): Likewise.
2908 (frv_hard_regno_mode_ok): Likewise.
2909 (frv_matching_accg_mode): Likewise.
2910 * config/h8300/h8300.c (split_adds_subs): Likewise.
2911 (h8300_rtx_costs): Likewise.
2912 (h8300_print_operand): Likewise.
2913 (compute_mov_length): Likewise.
2914 (output_logical_op): Likewise.
2915 (compute_logical_op_length): Likewise.
2916 (compute_logical_op_cc): Likewise.
2917 (h8300_shift_needs_scratch_p): Likewise.
2918 (output_a_shift): Likewise.
2919 (compute_a_shift_length): Likewise.
2920 (compute_a_shift_cc): Likewise.
2921 (expand_a_rotate): Likewise.
2922 (output_a_rotate): Likewise.
2923 * config/i386/i386.c (classify_argument): Likewise.
2924 (function_arg_advance_32): Likewise.
2925 (function_arg_32): Likewise.
2926 (function_arg_64): Likewise.
2927 (function_value_64): Likewise.
2928 (ix86_gimplify_va_arg): Likewise.
2929 (ix86_legitimate_constant_p): Likewise.
2930 (put_condition_code): Likewise.
2931 (split_double_mode): Likewise.
2932 (ix86_avx256_split_vector_move_misalign): Likewise.
2933 (ix86_expand_vector_logical_operator): Likewise.
2934 (ix86_split_idivmod): Likewise.
2935 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
2936 (ix86_build_const_vector): Likewise.
2937 (ix86_build_signbit_mask): Likewise.
2938 (ix86_match_ccmode): Likewise.
2939 (ix86_cc_modes_compatible): Likewise.
2940 (ix86_expand_branch): Likewise.
2941 (ix86_expand_sse_cmp): Likewise.
2942 (ix86_expand_sse_movcc): Likewise.
2943 (ix86_expand_int_sse_cmp): Likewise.
2944 (ix86_expand_vec_perm_vpermi2): Likewise.
2945 (ix86_expand_vec_perm): Likewise.
2946 (ix86_expand_sse_unpack): Likewise.
2947 (ix86_expand_int_addcc): Likewise.
2948 (ix86_split_to_parts): Likewise.
2949 (ix86_vectorize_builtin_gather): Likewise.
2950 (ix86_vectorize_builtin_scatter): Likewise.
2951 (avx_vpermilp_parallel): Likewise.
2952 (inline_memory_move_cost): Likewise.
2953 (ix86_tieable_integer_mode_p): Likewise.
2954 (x86_maybe_negate_const_int): Likewise.
2955 (ix86_expand_vector_init_duplicate): Likewise.
2956 (ix86_expand_vector_init_one_nonzero): Likewise.
2957 (ix86_expand_vector_init_one_var): Likewise.
2958 (ix86_expand_vector_init_concat): Likewise.
2959 (ix86_expand_vector_init_interleave): Likewise.
2960 (ix86_expand_vector_init_general): Likewise.
2961 (ix86_expand_vector_set): Likewise.
2962 (ix86_expand_vector_extract): Likewise.
2963 (emit_reduc_half): Likewise.
2964 (ix86_emit_i387_round): Likewise.
2965 (ix86_mangle_type): Likewise.
2966 (ix86_expand_round_sse4): Likewise.
2967 (expand_vec_perm_blend): Likewise.
2968 (canonicalize_vector_int_perm): Likewise.
2969 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
2970 (expand_vec_perm_1): Likewise.
2971 (expand_vec_perm_interleave3): Likewise.
2972 (expand_vec_perm_even_odd_pack): Likewise.
2973 (expand_vec_perm_even_odd_1): Likewise.
2974 (expand_vec_perm_broadcast_1): Likewise.
2975 (ix86_vectorize_vec_perm_const_ok): Likewise.
2976 (ix86_expand_vecop_qihi): Likewise.
2977 (ix86_expand_mul_widen_hilo): Likewise.
2978 (ix86_expand_sse2_abs): Likewise.
2979 (ix86_expand_pextr): Likewise.
2980 (ix86_expand_pinsr): Likewise.
2981 (ix86_preferred_simd_mode): Likewise.
2982 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
2983 * config/i386/sse.md (*andnot<mode>3): Likewise.
2984 (<mask_codefor><code><mode>3<mask_name>): Likewise.
2985 (*<code><mode>3): Likewise.
2986 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
2987 (ia64_expand_atomic_op): Likewise.
2988 (ia64_arg_type): Likewise.
2989 (ia64_mode_to_int): Likewise.
2990 (ia64_scalar_mode_supported_p): Likewise.
2991 (ia64_vector_mode_supported_p): Likewise.
2992 (expand_vec_perm_broadcast): Likewise.
2993 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
2994 (iq2000_function_arg_advance): Likewise.
2995 (iq2000_function_arg): Likewise.
2996 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
2997 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
2998 (m68k_libcall_value): Likewise.
2999 (m68k_function_value): Likewise.
3000 (sched_attr_op_type): Likewise.
3001 * config/mcore/mcore.c (mcore_output_move): Likewise.
3002 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
3004 (microblaze_function_arg): Likewise.
3005 * config/mips/mips.c (mips16_build_call_stub): Likewise.
3006 (mips_print_operand): Likewise.
3007 (mips_mode_ok_for_mov_fmt_p): Likewise.
3008 (mips_vector_mode_supported_p): Likewise.
3009 (mips_preferred_simd_mode): Likewise.
3010 (mips_expand_vpc_loongson_even_odd): Likewise.
3011 (mips_expand_vec_unpack): Likewise.
3012 (mips_expand_vi_broadcast): Likewise.
3013 (mips_expand_vector_init): Likewise.
3014 (mips_expand_vec_reduc): Likewise.
3015 (mips_expand_msa_cmp): Likewise.
3016 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
3017 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
3018 (cc_flags_for_mode): Likewise.
3019 * config/msp430/msp430.c (msp430_print_operand): Likewise.
3020 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
3021 (nds32_output_casesi_pc_relative): Likewise.
3022 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3023 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
3024 (nvptx_gen_unpack): Likewise.
3025 (nvptx_gen_pack): Likewise.
3026 (nvptx_gen_shuffle): Likewise.
3027 (nvptx_gen_wcast): Likewise.
3028 (nvptx_preferred_simd_mode): Likewise.
3029 * config/pa/pa.c (pa_secondary_reload): Likewise.
3030 * config/pa/predicates.md (base14_operand): Likewise.
3031 * config/powerpcspe/powerpcspe-c.c
3032 (altivec_resolve_overloaded_builtin): Likewise.
3033 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
3035 (rs6000_preferred_simd_mode): Likewise.
3036 (output_vec_const_move): Likewise.
3037 (rs6000_expand_vector_extract): Likewise.
3038 (rs6000_split_vec_extract_var): Likewise.
3039 (reg_offset_addressing_ok_p): Likewise.
3040 (rs6000_legitimate_offset_address_p): Likewise.
3041 (rs6000_legitimize_address): Likewise.
3042 (rs6000_emit_set_const): Likewise.
3043 (rs6000_const_vec): Likewise.
3044 (rs6000_emit_move): Likewise.
3045 (spe_build_register_parallel): Likewise.
3046 (rs6000_darwin64_record_arg_recurse): Likewise.
3047 (swap_selector_for_mode): Likewise.
3048 (spe_init_builtins): Likewise.
3049 (paired_init_builtins): Likewise.
3050 (altivec_init_builtins): Likewise.
3051 (do_load_for_compare): Likewise.
3052 (rs6000_generate_compare): Likewise.
3053 (rs6000_expand_float128_convert): Likewise.
3054 (emit_load_locked): Likewise.
3055 (emit_store_conditional): Likewise.
3056 (rs6000_output_function_epilogue): Likewise.
3057 (rs6000_handle_altivec_attribute): Likewise.
3058 (rs6000_function_value): Likewise.
3059 (emit_fusion_gpr_load): Likewise.
3060 (emit_fusion_p9_load): Likewise.
3061 (emit_fusion_p9_store): Likewise.
3062 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
3063 (fusion_gpr_mem_load): Likewise.
3064 (fusion_addis_mem_combo_load): Likewise.
3065 (fusion_addis_mem_combo_store): Likewise.
3066 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
3067 (fusion_gpr_mem_load): Likewise.
3068 (fusion_addis_mem_combo_load): Likewise.
3069 (fusion_addis_mem_combo_store): Likewise.
3070 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3072 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
3073 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
3074 (rs6000_preferred_simd_mode): Likewise.
3075 (output_vec_const_move): Likewise.
3076 (rs6000_expand_vector_extract): Likewise.
3077 (rs6000_split_vec_extract_var): Likewise.
3078 (reg_offset_addressing_ok_p): Likewise.
3079 (rs6000_legitimate_offset_address_p): Likewise.
3080 (rs6000_legitimize_address): Likewise.
3081 (rs6000_emit_set_const): Likewise.
3082 (rs6000_const_vec): Likewise.
3083 (rs6000_emit_move): Likewise.
3084 (rs6000_darwin64_record_arg_recurse): Likewise.
3085 (swap_selector_for_mode): Likewise.
3086 (paired_init_builtins): Likewise.
3087 (altivec_init_builtins): Likewise.
3088 (rs6000_expand_float128_convert): Likewise.
3089 (emit_load_locked): Likewise.
3090 (emit_store_conditional): Likewise.
3091 (rs6000_output_function_epilogue): Likewise.
3092 (rs6000_handle_altivec_attribute): Likewise.
3093 (rs6000_function_value): Likewise.
3094 (emit_fusion_gpr_load): Likewise.
3095 (emit_fusion_p9_load): Likewise.
3096 (emit_fusion_p9_store): Likewise.
3097 * config/rx/rx.c (rx_gen_move_template): Likewise.
3098 (flags_from_mode): Likewise.
3099 * config/s390/predicates.md (s390_alc_comparison): Likewise.
3100 (s390_slb_comparison): Likewise.
3101 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
3102 (s390_vector_mode_supported_p): Likewise.
3103 (s390_cc_modes_compatible): Likewise.
3104 (s390_match_ccmode_set): Likewise.
3105 (s390_canonicalize_comparison): Likewise.
3106 (s390_emit_compare_and_swap): Likewise.
3107 (s390_branch_condition_mask): Likewise.
3108 (s390_rtx_costs): Likewise.
3109 (s390_secondary_reload): Likewise.
3110 (__SECONDARY_RELOAD_CASE): Likewise.
3111 (s390_expand_cs): Likewise.
3112 (s390_preferred_simd_mode): Likewise.
3113 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
3114 * config/sh/sh.c (sh_print_operand): Likewise.
3115 (dump_table): Likewise.
3116 (sh_secondary_reload): Likewise.
3117 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3118 * config/sh/sh.md (casesi_worker_1): Likewise.
3119 (casesi_worker_2): Likewise.
3120 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
3121 (fcc_comparison_operator): Likewise.
3122 * config/sparc/sparc.c (sparc_expand_move): Likewise.
3123 (emit_soft_tfmode_cvt): Likewise.
3124 (sparc_preferred_simd_mode): Likewise.
3125 (output_cbranch): Likewise.
3126 (sparc_print_operand): Likewise.
3127 (sparc_expand_vec_perm_bmask): Likewise.
3128 (vector_init_bshuffle): Likewise.
3129 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
3130 (spu_vector_mode_supported_p): Likewise.
3131 (spu_expand_insv): Likewise.
3132 (spu_emit_branch_or_set): Likewise.
3133 (spu_handle_vector_attribute): Likewise.
3134 (spu_builtin_splats): Likewise.
3135 (spu_builtin_extract): Likewise.
3136 (spu_builtin_promote): Likewise.
3137 (spu_expand_sign_extend): Likewise.
3138 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
3139 (tilegx_simd_int): Likewise.
3140 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
3141 (tilepro_simd_int): Likewise.
3142 * config/v850/v850.c (const_double_split): Likewise.
3143 (v850_print_operand): Likewise.
3144 (ep_memory_offset): Likewise.
3145 * config/vax/vax.c (vax_rtx_costs): Likewise.
3146 (vax_output_int_move): Likewise.
3147 (vax_output_int_add): Likewise.
3148 (vax_output_int_subtract): Likewise.
3149 * config/visium/predicates.md (visium_branch_operator): Likewise.
3150 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
3151 (visium_print_operand_address): Likewise.
3152 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
3153 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
3154 (xtensa_expand_conditional_branch): Likewise.
3155 (xtensa_copy_incoming_a7): Likewise.
3156 (xtensa_output_literal): Likewise.
3157 * dfp.c (decimal_real_maxval): Likewise.
3158 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
3160 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3161 Alan Hayward <alan.hayward@arm.com>
3162 David Sherwood <david.sherwood@arm.com>
3164 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
3165 (mode_nunits_inline): Likewise.
3166 (mode_inner_inline): Likewise.
3167 (mode_unit_size_inline): Likewise.
3168 (mode_unit_precision_inline): Likewise.
3169 (emit_insn_modes_h): Likewise. Also emit a #define of the
3171 (emit_mode_wider): Add an E_ prefix to mode names.
3172 (emit_mode_complex): Likewise.
3173 (emit_mode_inner): Likewise.
3174 (emit_mode_adjustments): Likewise.
3175 (emit_mode_int_n): Likewise.
3176 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
3177 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
3178 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
3179 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
3180 (CRC32_BUILTIN, ENTRY): Likewise.
3181 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
3182 (aarch64_pop_regs): Likewise.
3183 (aarch64_process_components): Likewise.
3184 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
3185 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
3186 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
3187 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
3188 * config/arm/arm.c (arm_init_libfuncs): Likewise.
3189 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
3191 * config/i386/i386-builtin.def (pcmpestr): Likewise.
3192 (pcmpistr): Likewise.
3193 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
3194 * config/mmix/mmix.c (mmix_output_condition): Likewise.
3195 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
3197 * config/rl78/rl78.c (mduc_regs): Likewise.
3198 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3199 (htm_expand_builtin): Likewise.
3200 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
3201 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
3202 * config/xtensa/xtensa.c (print_operand): Likewise.
3203 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
3204 (NUM_MODE_VECTOR_INT): Likewise.
3205 * genoutput.c (null_operand): Likewise.
3206 (output_operand_data): Likewise.
3207 * genrecog.c (print_parameter_value): Likewise.
3208 * lra.c (debug_operand_data): Likewise.
3210 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3212 * dwarf2out.c (scompare_loc_descriptor_wide)
3213 (scompare_loc_descriptor_narrow): New functions, split out from...
3214 (scompare_loc_descriptor): ...here.
3215 * expmed.c (emit_store_flag_int): New function, split out from...
3216 (emit_store_flag): ...here.
3218 2017-08-30 Richard Biener <rguenther@suse.de>
3220 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
3221 (dwarf2out_early_finish): Move setting of AT_pubnames from
3222 early debug output to early finish.
3224 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3226 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
3227 and -mdata-region to the assembler.
3229 2017-08-30 Richard Biener <rguenther@suse.de>
3231 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
3233 (gen_subprogram_die): Add DW_AT_object_pointer only early.
3234 (dwarf2out_early_global_decl): Only generate a DIE for the
3235 abstract origin if it doesn't already exist or is a declaration DIE.
3236 (resolve_addr): Do not add the linkage name twice when
3237 generating a stub DIE for the DW_TAG_GNU_call_site target.
3239 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
3241 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3242 Use machine_mode rather than int for arg1_mode.
3244 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
3247 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
3248 that the second argument of the built-in functions to unpack
3249 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
3250 switch statement instead a lot of if statements.
3251 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
3252 Allow 64-bit values to be in Altivec registers as well as
3253 traditional floating point registers.
3254 (pack<mode>, FMOVE128_VSX iterator): Likewise.
3256 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
3258 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
3259 MAX_REGS_PER_ADDRESS == 1.
3261 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
3263 * config/i386/i386.opt (flag_fentry): Do not init to -1.
3264 * config/i386/i386.c (ix86_option_override_internal): Simplify
3265 setting of opts->x_flag_entry.
3267 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3268 Jakub Jelinek <jakub@redhat.com>
3269 Richard Biener <rguenther@suse.de>
3271 PR tree-optimization/81503
3272 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
3273 folded constant fits in the target type; reorder tests for clarity.
3275 2017-08-29 Martin Liska <mliska@suse.cz>
3277 * passes.def: Include pass_lower_switch.
3278 * stmt.c (dump_case_nodes): Remove and move to
3279 tree-switch-conversion.
3280 (case_values_threshold): Likewise.
3281 (expand_switch_as_decision_tree_p): Likewise.
3282 (emit_case_decision_tree): Likewise.
3283 (expand_case): Likewise.
3284 (balance_case_nodes): Likewise.
3285 (node_has_low_bound): Likewise.
3286 (node_has_high_bound): Likewise.
3287 (node_is_bounded): Likewise.
3288 (emit_case_nodes): Likewise.
3289 (struct simple_case_node): New struct.
3290 (add_case_node): Remove.
3291 (emit_case_dispatch_table): Use vector instead of case_list.
3292 (reset_out_edges_aux): Remove.
3293 (compute_cases_per_edge): Likewise.
3294 (expand_case): Build list of simple_case_node.
3295 (expand_sjlj_dispatch_table): Use it.
3296 * tree-switch-conversion.c (struct case_node): Moved from
3297 stmt.c and adjusted.
3298 (emit_case_nodes): Likewise.
3299 (node_has_low_bound): Likewise.
3300 (node_has_high_bound): Likewise.
3301 (node_is_bounded): Likewise.
3302 (case_values_threshold): Likewise.
3303 (reset_out_edges_aux): Likewise.
3304 (compute_cases_per_edge): Likewise.
3305 (add_case_node): Likewise.
3306 (dump_case_nodes): Likewise.
3307 (balance_case_nodes): Likewise.
3308 (expand_switch_as_decision_tree_p): Likewise.
3309 (emit_jump): Likewise.
3310 (emit_case_decision_tree): Likewise.
3311 (try_switch_expansion): Likewise.
3312 (do_jump_if_equal): Likewise.
3313 (emit_cmp_and_jump_insns): Likewise.
3314 (fix_phi_operands_for_edge): New function.
3315 (record_phi_operand_mapping): Likewise.
3316 (class pass_lower_switch): New pass.
3317 (pass_lower_switch::execute): New function.
3318 (make_pass_lower_switch): Likewise.
3319 (conditional_probability):
3320 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
3321 * tree-pass.h: Add make_pass_lower_switch.
3323 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3326 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
3329 2017-08-29 Richard Biener <rguenther@suse.de>
3331 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
3332 we do not add a DW_AT_inline attribute twice.
3333 (gen_subprogram_die): Remove code setting DW_AT_inline on
3334 DECL_ABSTRACT_P nodes.
3336 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
3338 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
3339 calls to internal functions.
3340 (gimplify_modify_expr): Likewise.
3341 * tree-call-cdce.c (use_internal_fn): Likewise.
3342 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
3343 (convert_to_divmod): Set the nothrow flag.
3344 * tree-if-conv.c (predicate_mem_writes): Likewise.
3345 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3346 (vectorizable_call): Likewise.
3347 (vectorizable_store): Likewise.
3348 (vectorizable_load): Likewise.
3349 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
3350 (vect_recog_mask_conversion_pattern): Likewise.
3352 2017-08-29 Martin Liska <mliska@suse.cz>
3355 * gcc.c (driver_handle_option): Add new argument.
3356 * opts-common.c (handle_option): Pass
3357 target_option_override_hook.
3358 * opts-global.c (lang_handle_option): Add new option.
3359 (set_default_handlers): Add new argument.
3360 (decode_options): Likewise.
3361 * opts.c (target_handle_option): Likewise.
3362 (common_handle_option): Call target_option_override_hook.
3363 * opts.h (struct cl_option_handler_func): Add hook for
3364 target option override.
3365 (struct cl_option_handlers): Likewise.
3366 (set_default_handlers): Add new argument.
3367 (decode_options): Likewise.
3368 (common_handle_option): Likewise.
3369 (target_handle_option): Likewise.
3370 * toplev.c (toplev::main): Pass targetm.target_option.override
3373 2017-08-29 Richard Biener <rguenther@suse.de>
3374 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3376 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
3377 life to the active subtree.
3379 2017-08-28 Jeff Law <law@redhat.com>
3381 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
3382 derive_equivalences.
3383 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
3385 (edge_info::derive_equivalences): New private member function
3387 * tree-ssa-dom.c (class edge_info): Changed from a struct
3388 to a class. Add ctor/dtor, methods and data members.
3389 (edge_info::edge_info): Renamed from allocate_edge_info.
3390 Initialize additional members.
3391 (edge_info::~edge_info): New.
3392 (free_dom_edge_info): Delete the edge info.
3393 (record_edge_info): Use new class & associated member functions.
3394 Tighten forms for testing for edge equivalences.
3395 (record_temporary_equivalences): Iterate over the simple
3396 equivalences rather than assuming there's only one per edge.
3397 (cprop_into_successor_phis): Iterate over the simple
3398 equivalences rather than assuming there's only one per edge.
3399 (optimize_stmt): Use operand_equal_p rather than pointer
3400 equality for mini-DSE code.
3402 2017-08-28 Nathan Sidwell <nathan@acm.org>
3404 * gcc.c (execute): Fold SIGPIPE handling into switch
3405 statement. Adjust internal error message.
3407 2017-08-28 Richard Biener <rguenther@suse.de>
3410 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
3411 Do nothing for removed DIEs.
3413 2017-08-28 Richard Biener <rguenther@suse.de>
3415 PR tree-optimization/81977
3416 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
3419 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
3422 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
3423 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
3424 using targetm.gen_mem_thread_fence.
3426 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
3429 * config/i386/i386.md (*<btsc><mode>): Change operand 2
3430 predicate to register_operand. Reorder operands.
3431 (*btr<mode>): Ditto.
3432 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
3433 (*btr<mode>_mask): Ditto.
3435 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
3437 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
3438 * config/rs6000/xmmintrin.h: New file.
3439 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
3441 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3444 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
3445 parameter and_insn and return it.
3446 (recombine_lvx_pattern): Insert a copy to ensure availability of
3447 the base register of the copied masking operation at the point of
3448 the instruction replacement.
3449 (recombine_stvx_pattern): Likewise.
3451 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
3453 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
3454 undocumented switches.
3455 (-mpower9-dform-vector): Likewise.
3456 (-mpower9-dform): Likewise.
3457 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
3458 comments to delete references to -mpower9-dform* switches.
3459 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
3460 Delete reference to -mpower9-dform* switches, test for
3461 -mpower9-vector instead.
3462 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
3463 (OTHER_P9_VECTOR_MASKS): Likewise.
3464 (POWERPC_MASKS): Likewise.
3465 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
3466 tests against -mpower9-dform* to -mpower9-vector. Delete code
3467 that checked for -mpower9-dform* consistancy with other options.
3468 Add test for -mpower9-misc to enable other power9 switches.
3469 (rs6000_init_hard_regno_mode_ok): Likewise.
3470 (rs6000_option_override_internal): Likewise.
3471 (rs6000_emit_prologue): Likewise.
3472 (rs6000_emit_epilogue): Likewise.
3473 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
3474 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
3475 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
3477 (emit_fusion_p9_store): Likewise.
3478 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
3479 resetting these macros if the assembler does not support ISA 3.0
3481 (TARGET_P9_DFORM_VECTOR): Likewise.
3482 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
3483 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
3485 2017-08-25 Alan Modra <amodra@gmail.com>
3487 PR rtl-optimization/81747
3488 * cse.c (cse_extended_basic_block): Don't attempt to record
3489 equivalences for degenerate conditional branches that branch
3490 to their fall-through.
3492 2017-08-24 Martin Sebor <msebor@redhat.com>
3495 * gimple-fold.c (size_must_be_zero_p): New function.
3496 (gimple_fold_builtin_memory_op): Call it.
3498 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
3500 * config/rs6000/mm_malloc.h: New file.
3502 2017-08-24 Bin Cheng <bin.cheng@arm.com>
3504 PR tree-optimization/81913
3505 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
3506 analysis when either IVs in condition can wrap.
3508 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
3510 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
3511 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
3513 2017-08-24 Richard Biener <rguenther@suse.de>
3516 * targhooks.c (default_target_can_inline_p): Properly
3517 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
3518 is present and always compare.
3519 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
3520 infer -mfpmath=sse from TARGET_SSE_P.
3521 (ix86_can_inline_p): Properly use target_option_default_node when
3522 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
3524 2017-08-24 Richard Biener <rguenther@suse.de>
3527 * dwarf2out.c (output_die): Handle flag_generate_offload like
3529 (output_comp_unit): Likewise.
3530 (gen_array_type_die): Likewise.
3531 (dwarf2out_early_finish): Likewise.
3532 (note_variable_value_in_expr): Likewise.
3533 (dwarf2out_finish): Likewise. Adjust assert.
3534 * cgraphunit.c (symbol_table::compile): Move setting of
3535 flag_generate_offload earlier ...
3536 (symbol_table::finalize_compilation_unit): ... here, before
3537 early debug finalization.
3539 2017-08-24 Richard Biener <rguenther@suse.de>
3541 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
3542 and ipa-fnsummary.h.
3543 (ix86_can_inline_p): When ix86_fpmath flags do not match
3544 check whether the callee uses FP math at all.
3546 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
3549 * tree-ssanames.c (get_nonzero_bits): Use element_precision
3550 instead of TYPE_PRECISION.
3552 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
3553 Alan Hayward <alan.hayward@arm.com>
3554 David Sherwood <david.sherwood@arm.com>
3556 * combine.c (make_extraction): Use subreg_offset_from_lsb.
3558 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
3560 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
3562 (ix86_frame::stack_realign_allocate): New field.
3563 (struct machine_frame_state): Modify comments.
3564 (machine_frame_state::sp_realigned_fp_end): New field.
3565 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
3567 (sp_valid_at): Add assertion to assure no attempt to access invalid
3568 offset of a realigned stack.
3569 (fp_valid_at): Likewise.
3570 (choose_baseaddr): Modify comments.
3571 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
3572 ix86_expand_prologue.
3573 (ix86_expand_prologue): Modify stack realignment and allocation.
3574 (ix86_expand_epilogue): Modify comments.
3575 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
3576 avx2_runtime, avx512f, and avx512f_runtime.
3578 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
3580 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
3581 (mstackrealign): Do not init to -1.
3582 * config/i386/i386.c (ix86_option_override_internal):
3583 Check opts_set, not opts when setting default value of
3584 opts->x_ix86_force_align_arg_pointer.
3586 2017-08-23 Richard Biener <rguenther@suse.de>
3588 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
3589 lang_hooks.decl_printable_name.
3590 * print-rtl-function.c (print_rtx_function): Likewise.
3591 * tree-pretty-print.c (dump_function_header): Likewise.
3593 2017-08-23 Richard Biener <rguenther@suse.de>
3596 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
3597 -g0 at compile-time.
3599 2017-08-23 Tamar Christina <tamar.christina@arm.com>
3602 * doc/sourcebuild.texi (Other hardware attributes):
3605 2017-08-23 Tamar Christina <tamar.christina@arm.com>
3608 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
3609 Fix single-use check.
3611 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
3613 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
3615 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
3617 * doc/install.texi: Modify to add more details on running selected
3620 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
3622 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
3623 is combined with -mabi=ms.
3624 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
3627 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3629 PR tree-optimization/81488
3630 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
3631 and cached_basis fields.
3632 (MAX_SPREAD): New constant.
3633 (alloc_cand_and_find_basis): Initialize new fields.
3634 (clear_visited): New function.
3635 (create_phi_basis_1): Rename from create_phi_basis, set visited
3636 and cached_basis fields.
3637 (create_phi_basis): New wrapper function.
3638 (phi_add_costs_1): Rename from phi_add_costs, add spread
3639 parameter, set visited field, short-circuit when limits reached.
3640 (phi_add_costs): New wrapper function.
3641 (record_phi_increments_1): Rename from record_phi_increments, set
3643 (record_phi_increments): New wrapper function.
3644 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
3645 (phi_incr_cost): New wrapper function.
3646 (all_phi_incrs_profitable_1): Rename from
3647 all_phi_incrs_profitable, set visited field.
3648 (all_phi_incrs_profitable): New wrapper function.
3650 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
3651 Alan Hayward <alan.hayward@arm.com>
3652 David Sherwood <david.sherwood@arm.com>
3654 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
3655 that takes the outer and inner modes.
3656 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
3657 comparison as the canonical test for a paradoxical subreg.
3658 * combine.c (simplify_set): Use paradoxical_subreg_p.
3659 (make_extraction): Likewise.
3660 (force_to_mode): Likewise.
3661 (rtx_equal_for_field_assignment_p): Likewise.
3662 (gen_lowpart_for_combine): Likewise.
3663 (simplify_comparison): Likewise.
3664 * cse.c (equiv_constant): Likewise.
3665 * expmed.c (store_bit_field_1): Likewise.
3666 * final.c (alter_subreg): Likewise.
3667 * fwprop.c (propagate_rtx): Likewise.
3668 (forward_propagate_subreg): Likewise.
3669 * ira-conflicts.c (ira_build_conflicts): Likewise.
3670 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
3671 * lra-constraints.c (curr_insn_transform): Likewise.
3672 (split_reg): Likewise.
3673 * lra-eliminations.c (move_plus_up): Likewise.
3674 (lra_eliminate_regs_1): Likewise.
3675 * recog.c (general_operand): Likewise.
3676 * ree.c (combine_reaching_defs): Likewise.
3677 * reload.c (push_reload): Likewise.
3678 (find_reloads): Likewise.
3679 * reload1.c (elimination_effects): Likewise.
3680 (compute_reload_subreg_offset): Likewise.
3681 (choose_reload_regs): Likewise.
3682 * rtlanal.c (subreg_lsb_1): Likewise.
3683 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3684 (simplify_subreg): Likewise.
3685 * var-tracking.c (track_loc_p): Likewise.
3686 * emit-rtl.c (byte_lowpart_offset): Likewise.
3687 (paradoxical_subreg_p): Delete out-of-line definition.
3689 2017-08-22 Jeff Law <law@redhat.com>
3691 PR tree-optimization/81741
3692 PR tree-optimization/71947
3693 * tree-ssa-dom.c: Include tree-inline.h.
3694 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
3695 equivalences if one is more expensive to compute than the other.
3696 * tree-ssa-scopedtables.h (class const_or_copies): Make
3697 record_const_or_copy_raw method private.
3698 (class avail_exprs_stack): New method simplify_binary_operation.
3699 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
3700 avail_exprs_stack::simplify_binary_operation as needed.
3701 (avail_exprs_stack::simplify_binary_operation): New function.
3703 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
3705 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
3706 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
3707 (DOT_SYMBOLS): Likewise.
3708 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
3709 (RELOCATABLE_NEEDS_FIXUP): Likewise.
3710 (RS6000_ABI_NAME): Likewise.
3711 (TARGET_CMODEL): Likewise.
3712 (TOC_SECTION_ASM_OP): Likewise.
3713 (SET_CMODEL): New macro.
3714 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
3716 2017-08-22 Richard Biener <rguenther@suse.de>
3718 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
3719 to allow for free-lang-data replacements similar to verify_type_variant.
3721 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
3722 Alan Hayward <alan.hayward@arm.com>
3723 David Sherwood <david.sherwood@arm.com>
3725 * config/aarch64/aarch64.md (casesi): Use DImode rather than
3726 VOIDmode for the LABEL_REF.
3728 2017-08-22 Richard Biener <rguenther@suse.de>
3730 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
3732 2017-08-22 Richard Biener <rguenther@suse.de>
3734 * common.opt (feliminate-dwarf2-dups): Ignore.
3735 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
3736 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
3737 same_die_p_wrap, compute_section_prefix,
3738 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
3739 (comdat_symbol_id, comdat_symbol_number): Likewise.
3740 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
3742 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
3743 (output_die): Mark unreachable path unreachable.
3744 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
3745 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
3746 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
3747 (dwarf2out_early_finish): Likewise.
3749 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
3751 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
3753 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
3756 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
3757 not VAR_P. Filter attribute warnings with OPT_Wattributes.
3758 (avr_attribute_table) <io, io_low, address>: Initialize
3759 .decl_required with true.
3761 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
3763 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
3764 undocumented debugging options.
3765 (-mvsx-scalar-double): Likewise.
3766 (-mallow-df-permute): Likewise.
3767 (-mvectorize-builtins): Likewise.
3768 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
3769 (rs6000_builtin_vectorized_function): Likewise.
3770 (rs6000_builtin_md_vectorized_function): Likewise.
3771 (rs6000_opt_vars): Likewise.
3773 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
3776 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
3777 (*btrq_imm): Rename from *btrq.
3778 (*btcq_imm): Rename from *btcq.
3779 (btsc): New code attribute.
3780 (*<btsc><mode>): New insn pattern.
3781 (*btr<mode>): Ditto.
3782 (*<btsc><mode>_mask): New insn_and_split pattern.
3783 (*btr<mode>_mask): Ditto.
3785 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3786 Alan Hayward <alan.hayward@arm.com>
3787 David Sherwood <david.sherwood@arm.com>
3789 * function.c (pad_below): Simplify padding calculation.
3791 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3792 Alan Hayward <alan.hayward@arm.com>
3793 David Sherwood <david.sherwood@arm.com>
3795 * target.def (function_prologue): Remove frame size argument.
3796 (function_epilogue): Likewise.
3797 * doc/tm.texi: Regenerate.
3798 * final.c (final_start_function): Update call to function_prologue.
3799 (final_end_function): Update call to function_epilogue.
3800 (default_function_pro_epilogue): Remove frame size argument.
3801 * output.h (default_function_pro_epilogue): Likewise.
3802 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
3803 (arm_output_function_prologue): Likewise.
3804 * config/frv/frv.c (frv_function_prologue): Likewise.
3805 (frv_function_epilogue): Likewise.
3806 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
3807 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
3808 (ia64_output_function_epilogue): Likewise.
3809 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
3810 (m32r_output_function_epilogue): Likewise.
3811 * config/microblaze/microblaze.c (microblaze_function_prologue)
3812 (microblaze_function_epilogue): Likewise.
3813 * config/mips/mips.c (mips_output_function_prologue): Likewise.
3814 (mips_output_function_epilogue): Likewise.
3815 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
3816 (mmix_target_asm_function_epilogue): Likewise.
3817 * config/msp430/msp430.c (msp430_start_function): Likewise.
3818 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
3819 (nds32_asm_function_epilogue): Likewise.
3820 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
3821 * config/pa/pa.c (pa_output_function_prologue): Likewise.
3822 (pa_output_function_epilogue): Likewise.
3823 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
3824 (rs6000_output_function_epilogue): Likewise.
3825 * config/rl78/rl78.c (rl78_start_function): Likewise.
3826 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
3827 (rs6000_output_function_epilogue): Likewise.
3828 * config/rx/rx.c (rx_output_function_prologue): Likewise.
3829 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
3830 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
3831 (sparc_asm_function_epilogue): Likewise.
3833 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3835 * tree.h (type_has_mode_precision_p): New function.
3836 * convert.c (convert_to_integer_1): Use it.
3837 * expr.c (expand_expr_real_2): Likewise.
3838 (expand_expr_real_1): Likewise.
3839 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
3840 * match.pd: Likewise.
3841 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
3842 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
3843 * tree-tailcall.c (process_assignment): Likewise.
3844 * tree-vect-loop.c (vectorizable_reduction): Likewise.
3845 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
3846 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
3847 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
3848 (vectorizable_assignment): Likewise.
3849 (vectorizable_shift): Likewise.
3850 (vectorizable_operation): Likewise.
3851 * tree-vrp.c (register_edge_assert_for_2): Likewise.
3853 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
3855 * match.pd: Add pow (C, x) simplification.
3857 2017-08-21 Richard Biener <rguenther@suse.de>
3859 PR tree-optimization/81900
3860 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
3861 for blocks with abnormal predecessors.
3862 (compute_antic): Do not set visited flag prematurely.
3864 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
3867 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
3869 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
3871 * stor-layout.h (vector_type_mode): Move to...
3872 * tree.h (vector_type_mode): ...here.
3873 * stor-layout.c (vector_type_mode): Move to...
3874 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
3876 2017-08-21 Richard Biener <rguenther@suse.de>
3878 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
3879 register_external_die hooks.
3880 (debug_false_tree_charstarstar_uhwistar): Declare.
3881 (debug_nothing_tree_charstar_uhwi): Likewise.
3882 * debug.c (do_nothing_debug_hooks): Adjust.
3883 (debug_false_tree_charstarstar_uhwistar): New do nothing.
3884 (debug_nothing_tree_charstar_uhwi): Likewise.
3885 * dbxout.c (dbx_debug_hooks): Adjust.
3886 (xcoff_debug_hooks): Likewise.
3887 * sdbout.c (sdb_debug_hooks): Likewise.
3888 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3889 * dwarf2out.c (macinfo_label_base): New global.
3890 (dwarf2out_register_external_die): New function for the
3891 register_external_die hook.
3892 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
3893 (dwarf2_debug_hooks): Use them.
3894 (dwarf2_lineno_debug_hooks): Adjust.
3895 (struct die_struct): Add with_offset flag.
3896 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
3897 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
3898 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
3899 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
3900 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
3901 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
3902 defining section names for the early LTO debug variants.
3903 (reset_indirect_string): New helper.
3904 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
3905 (print_dw_val): Add support for offsetted symbol references.
3906 (get_ultimate_context): Split out from is_cxx.
3907 (is_cxx): Use get_ultimate_context.
3908 (is_fortran): Add decl overload.
3909 (compute_comp_unit_symbol): Split out worker from
3910 compute_section_prefix.
3911 (compute_section_prefix): Call compute_comp_unit_symbol and
3912 set comdat_type_p here.
3913 (output_die): Skip DIE symbol output for the LTO added one.
3914 Handle DIE symbol references with offset.
3915 (output_comp_unit): Guard section name mangling properly.
3916 For LTO debug sections emit a symbol at the section beginning
3917 which we use to refer to its DIEs.
3918 (add_abstract_origin_attribute): For DIEs registered via
3919 dwarf2out_register_external_die directly refer to the early
3920 DIE rather than indirectly through the shadow one we created.
3921 Remove obsolete call to dwarf2out_abstract_function for
3922 non-function/block origins.
3923 (gen_array_type_die): When generating early LTO debug do
3924 not emit DW_AT_string_length.
3925 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
3926 late when in LTO. As suggested place a gcc_unreachable for
3927 the DECL_ABSTRACT_P case.
3928 (gen_subprogram_die): Avoid another specification DIE
3929 for early built declarations/definitions for the late LTO case.
3930 (gen_variable_die): Add type references for late duplicated VLA dies
3932 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
3933 we have the abstract instance already.
3934 (process_scope_var): Adjust decl DIE contexts in LTO which
3935 first puts them in limbo.
3936 (gen_decl_die): Do not generate type DIEs late apart from
3937 types for VLAs or for decls we do not yet have a DIE. Do not
3938 call dwarf2out_abstract_function late.
3939 (dwarf2out_early_global_decl): Make sure to create DIEs
3940 for abstract instances of a decl first.
3941 (dwarf2out_late_global_decl): Adjust comment.
3942 (output_macinfo_op): With multiple macro sections use
3943 macinfo_label_base to distinguish labels.
3944 (output_macinfo): Likewise. Update macinfo_label_base.
3945 Pass in the line info label.
3946 (note_variable_value_in_expr): When generating LTO resolve
3947 all variable values here by generating DIEs as needed.
3948 (init_sections_and_labels): Add early LTO debug flag parameter
3949 and generate different sections and names if set. Add generation
3950 counter for the labels so we can have multiple of them.
3951 (reset_dies): Helper to allow DIEs to be output multiple times.
3952 (dwarf2out_finish): When outputting DIEs to the fat part of an
3953 LTO object first reset DIEs.
3954 (dwarf2out_early_finish): Output early DIEs when generating LTO.
3955 (modified_type_die): Check for decl_ultimate_origin being self
3957 (gen_type_die_with_usage): Likewise.
3958 (gen_typedef_die): Allow decl_ultimate_origin being self.
3959 (set_decl_abstract_flags): Remove.
3960 (set_block_abstract_flags): Likewise.
3961 (dwarf2out_abstract_function): Treat the early generated DIEs
3962 as the abstract copy and only add DW_AT_inline and
3963 DW_AT_artificial here and call set_decl_origin_self.
3964 If the DIE has an abstract origin don't do anything.
3965 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
3966 if we have none yet (Go fails to build one, PR78628).
3967 (variably_modified_type_p): Prevent endless recursion for Ada
3968 cyclic pointer types.
3969 * lto-streamer-in.c: Include debug.h.
3970 (dref_queue): New global.
3971 (lto_read_tree_1): Stream in DIE references.
3972 (lto_input_tree): Register DIE references.
3973 (input_function): Stream DECL_DEBUG_ARGS.
3974 * lto-streamer-out.c: Include debug.h.
3975 (lto_write_tree_1): Output DIE references.
3976 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
3977 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
3978 (output_function): Stream DECL_DEBUG_ARGS.
3979 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
3980 Stream DECL_ABSTRACT_ORIGIN.
3981 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
3982 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
3983 DECL_CONTEXT for file-scope decls.
3984 * lto-streamer.h (struct dref_entry): Declare.
3985 (dref_queue): Likewise.
3986 * cfgexpand.c (pass_expand::execute): Do not call the
3987 outlining_inline_function hook here.
3988 * lto-wrapper.c (debug_obj): New global.
3989 (tool_cleanup): Unlink it if required.
3990 (debug_objcopy): New function.
3991 (run_gcc): Handle early debug sections in the IL files by
3992 extracting them to separate files, partially linkin them and
3993 feeding the result back as result to the linker.
3994 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
3995 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
3996 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
3997 sections into a separate segment.
3998 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
4000 (darwin_asm_dwarf_section): Likewise.
4001 (darwin_asm_output_dwarf_offset): Likewise.
4002 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
4004 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4005 Alan Hayward <alan.hayward@arm.com>
4006 David Sherwood <david.sherwood@arm.com>
4008 * read-md.h (md_reader::record_potential_iterator_use): Replace
4009 pointer argument with an rtx and an index.
4010 * read-rtl.c (iterator_group::apply_iterator): Likewise.
4011 (apply_mode_iterator): Likewise.
4012 (apply_code_iterator): Likewise.
4013 (apply_int_iterator): Likewise.
4014 (apply_subst_iterator): Likewise.
4015 (record_iterator_use): Likewise.
4016 (record_attribute_use): Likewise.
4017 (md_reader::record_potential_iterator_use): Likewise. Update calls
4018 to record_iterator_use and apply_iterator.
4019 (iterator_use): Replace ptr with x and index.
4020 (attribute_use): Likewise.
4021 (apply_attribute_uses): Update calls to apply_iterator.
4022 (apply_iterators): Likewise. Update initialization of iterator_use.
4023 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
4024 and record_potential_iterator_use.
4025 (rtx_reader::read_rtx_operand): Likewise.
4027 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
4028 Alan Hayward <alan.hayward@arm.com>
4029 David Sherwood <david.sherwood@arm.com>
4031 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
4034 2017-08-21 Richard Biener <rguenther@suse.de>
4037 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
4038 at struct end conservatively when comparing common bases.
4040 2017-08-21 Richard Biener <rguenther@suse.de>
4042 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
4043 (mem_ref_in_stmt): Remove.
4044 (determine_max_movement): Use ref index to get at the reference.
4045 (invariantness_dom_walker::before_dom_children): Deal with
4046 lim data already initialized.
4047 (gather_mem_refs_stmt): Initialize lim data and record ref index.
4049 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4051 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
4052 (TARGET_ISA_ROUND): Ditto.
4053 (TARGET_ROUND): Ditto.
4054 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
4055 * config/i386/i386.md: Ditto.
4056 * config/i386/sse.md: Ditto.
4057 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
4058 with OPTION_MASK_ISA_SSE4_1.
4060 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
4063 * doc/extend.texi (x86 Built-in Functions): Correct the name of
4064 __builtin_ia32_lzcnt_u16.
4066 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4069 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
4070 (rs6000_set_current_function): Rewrite function to use it.
4072 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
4075 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
4076 and TYPE_WARN_IF_NOT_ALIGN.
4077 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
4078 (handle_warn_if_not_align): New.
4079 (place_union_field): Call handle_warn_if_not_align.
4080 (place_field): Call handle_warn_if_not_align.
4081 Copy TYPE_WARN_IF_NOT_ALIGN.
4082 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
4083 (layout_type): Likewise.
4084 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
4086 (tree_decl_common): Add warn_if_not_align.
4087 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
4088 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
4089 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
4090 (DECL_WARN_IF_NOT_ALIGN): Likewise.
4091 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
4092 * doc/extend.texi: Document warn_if_not_aligned attribute.
4093 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
4095 2017-08-17 Martin Liska <mliska@suse.cz>
4098 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
4099 (get_data_dependence): Use it as pointer type.
4100 (distribute_loop): Likewise.
4102 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4104 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
4105 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
4106 (p8_vmrgow_<mode>_direct): New define_insn.
4107 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
4108 handle endianness for vmrgew and vmrgow permute patterns.
4110 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4112 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
4113 * config/rs6000/rs6000-cpus.def: Remove comment.
4114 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
4115 (POWERPC_MASKS): Likewise.
4116 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
4117 use of TARGET_VSX_TIMODE.
4118 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
4119 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
4120 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
4121 (rs6000_option_override_internal): Remove dead code.
4122 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
4123 (rs6000_legitimize_reload_address): Likewise.
4124 (rs6000_legitimate_address_p): Likewise.
4125 (rs6000_opt_masks): Delete "vsx-timode".
4126 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
4127 from function comment.
4128 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
4129 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
4130 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
4132 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
4133 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
4134 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
4137 (define_peephole2): Likewise.
4139 2017-08-17 Martin Sebor <msebor@redhat.com>
4142 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
4143 past the end of an array.
4144 (test_pp_format): Add test cases.
4146 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
4148 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
4149 missing ECF_NOTHROW flags.
4151 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
4154 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
4155 operands residing in integer registers.
4156 (*vsx_le_perm_load_<mode>): Likewise.
4157 (*vsx_le_perm_store_<mode>): Likewise.
4158 (define_peephole2): Add peepholes to optimize the above.
4160 2017-08-17 Marek Polacek <polacek@redhat.com>
4163 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
4164 to mimic what shorten_compare did. Change the return type to bool.
4165 (fold_cond_expr_with_comparison): Update call to
4166 operand_equal_for_comparison_p.
4167 (fold_ternary_loc): Likewise.
4169 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
4171 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
4173 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
4175 2017-08-17 Richard Biener <rguenther@suse.de>
4177 * tree-ssa-structalias.c (solve_graph): When propagating
4178 to successors update the graphs succ edges and avoid duplicate work.
4180 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
4183 * config/i386/i386.c (ix86_option_override_internal): Save target
4184 specific options after ix86_stack_protector_guard_reg was changed.
4186 2017-08-17 Richard Biener <rguenther@suse.de>
4188 PR tree-optimization/81827
4189 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
4190 (new_var_info): Initialize it conservatively.
4191 (get_call_vi): Mark register vars.
4192 (new_scalar_tmp_constraint_exp): Likewise.
4193 (handle_rhs_call): Likewise.
4194 (handle_const_call): Likewise.
4195 (create_function_info_for): Likewise.
4196 (solve_constraints): Sort varinfos to separate register from
4197 non-register vars to pack points-to solution bitmaps during
4200 2017-08-17 Marek Polacek <polacek@redhat.com>
4202 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
4204 2017-08-17 Richard Biener <rguenther@suse.de>
4206 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
4207 to true when overflow is undefined and we saturated the result.
4209 2017-08-17 Alan Modra <amodra@gmail.com>
4212 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
4213 Don't use store multiple if only one reg needs saving.
4214 (interesting_frame_related_regno): New function.
4215 (rs6000_frame_related): Don't emit frame info for regs that
4217 (rs6000_emit_epilogue): Likewise.
4219 2017-08-16 Nathan Sidwell <nathan@acm.org>
4221 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
4222 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
4223 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
4224 * tree.c (free_lang_data_in_type): Use else-if chain. Always
4225 clear TYPE_LANG_1. Remove obsolete member-function stripping.
4226 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
4227 (verify_type): Adjust for TYPE_BINFO move.
4228 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
4229 process TYPE_BINFO directly.
4230 (hash_tree): Likewise.
4231 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
4233 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
4236 2017-08-16 David Malcolm <dmalcolm@redhat.com>
4238 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
4240 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
4243 * config/i386/i386.md (*anddi_1_btr): Change predicates of
4244 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
4245 Add ix86_binary_operator_ok to insn constraint.
4246 (*iordi_1_bts): Ditto.
4247 (*xordi_1_btc): Ditto.
4248 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
4249 Update corresponding peephole2 pattern.
4253 2017-08-16 Bin Cheng <bin.cheng@arm.com>
4255 PR tree-optimization/81832
4256 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
4257 copy loop header which has IFN_LOOP_DIST_ALIAS call.
4259 2017-08-16 Marek Polacek <polacek@redhat.com>
4262 * fold-const.c (fold_indirect_ref_1): Restore original behavior
4263 regarding size_zero_node.
4265 2017-08-16 Martin Liska <mliska@suse.cz>
4268 * config.gcc: Respect previously set extra_objs in case
4271 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
4273 PR tree-optimization/81835
4274 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
4275 the phi SSA_NAME. Check that the condition in a COND_EXPR does
4276 not depend on the phi.
4278 2017-08-16 Alan Modra <amodra@gmail.com>
4280 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
4283 2017-08-16 Alan Modra <amodra@gmail.com>
4285 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
4286 (save_reg_p): ..into this. Update all callers.
4287 (first_reg_to_save): Simplify.
4289 2017-08-16 Alan Modra <amodra@gmail.com>
4291 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
4294 2017-08-15 Joseph Myers <joseph@codesourcery.com>
4298 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
4299 constant count if that count is less than 32.
4301 2017-08-15 Nathan Sidwell <nathan@acm.org>
4303 * gcc.c (execute): Emit friendlier message if inferior is killed
4304 by an external cause.
4306 2017-08-15 Richard Biener <rguenther@suse.de>
4308 PR tree-optimization/81790
4309 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
4310 CONSTRUCTORs from simplifying and VN.
4312 2017-08-14 Martin Sebor <msebor@redhat.com>
4314 * builtin-attrs.def: Add comments.
4316 2017-08-14 Martin Sebor <msebor@redhat.com>
4319 * doc/extend.texi (attribute nonstring): Document new attribute.
4321 2017-08-14 Martin Sebor <msebor@redhat.com>
4324 * tree-diagnostic.c (default_tree_printer): Handle %G.
4325 * gimple-pretty-print.h (percent_G_format): Declare new function.
4326 * gimple-pretty-print.c (percent_G_format): Define.
4327 * tree-pretty-print.c (percent_K_format): Add argument.
4329 2017-08-14 Martin Sebor <msebor@redhat.com>
4331 PR translation/79998
4332 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
4333 Remove a stray space.
4335 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
4338 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
4339 (*iordi_1_bts): Ditto.
4340 (*xordi_1_btc): Ditto.
4342 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4345 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
4346 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
4348 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
4349 quoted strings, and make more translator-friendly.
4350 (darwin_rs6000_override_options): Likewise.
4351 (rs6000_option_override_internal): Likewise.
4352 (rs6000_return_in_memory): Fix overlong line.
4353 (init_cmulative_args): Use quoted strings, and make more
4354 translator-friendly.
4355 (rs6000_pass_by_reference): Fix overlong line.
4356 (def_builtin): Use quoted strings.
4357 (altivec_expand_predicate_builtin): Use quoted strings, and make
4358 more translator-friendly.
4359 (htm_expand_builtin): Use quoted strings.
4360 (cpu_expand_builtin): Use quoted strings, and make more
4361 translator-friendly.
4362 (altivec_expand_builtin): Likewise.
4363 (paired_expand_predicate_builtin): Likewise.
4364 (rs6000_invalid_builtin): Likewise.
4365 (builtin_function_type): Use quoted strings.
4366 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
4367 more translator-friendly.
4368 (rs6000_trampoline_init): Likewise.
4369 (rs6000_handle_altivec_attribute): Likewise.
4370 (rs6000_inner_target_options): Use quoted strings.
4371 (rs6000_disable_incompatible_switches): Likewise.
4372 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
4373 strings, and make more translator-friendly.
4374 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
4376 2017-08-14 Bin Cheng <bin.cheng@arm.com>
4378 PR tree-optimization/81799
4379 * tree-loop-distribution.c (version_loop_by_alias_check): Force
4380 cond_expr to simple gimple operand.
4382 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
4385 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
4387 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
4391 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
4392 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
4393 TARGET_GASISR_PROLOGUES.
4394 * config/avr/avr.c (avr_option_override): Same.
4395 (avr_pass_pre_proep::execute): Same.
4397 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
4400 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
4401 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
4403 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
4405 * config/i386/i386.md (*load_tp_<mode>): Redefine as
4406 define_insn_and_split. Split to a memory load from 0 in
4407 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
4408 using PTR mode iterator.
4409 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
4410 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
4411 (*add_tp_<mode>): Redefine as define_insn_and_split.
4412 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
4413 address space. Merge with *add_tp_x32 using PTR mode iterator.
4414 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
4415 Split to an add with a memory load from 0 in
4416 DEFAULT_TLS_SEG_REG address space.
4418 2017-08-12 Andrew Pinski <apinski@cavium.com>
4420 * config/aarch64/aarch64-option-extensions.def (rdma):
4421 Fix feature string to what Linux prints out in /proc/cpuinfo.
4423 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
4426 * dwarf2out.c (modified_type_die): For C typedef types that have
4427 an ultimate origin, process the ultimate origin instead of the
4429 (gen_typedef_die): Assert that input DECLs have no ultimate
4431 (gen_type_die_with_usage): For typedef variants that have an
4432 ultimate origin, just call gen_decl_die on the original DECL.
4433 (process_scope_var): Avoid creating DIEs for local typedefs and
4434 concrete static variables.
4436 2017-08-12 Alan Modra <amodra@gmail.com>
4440 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
4441 match gnu-user.h startfile.
4442 (ENDFILE_LINUX_SPEC): Similarly.
4444 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
4447 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
4449 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
4451 2017-08-11 Tamar Christina <tamar.christina@arm.com>
4452 * config/aarch64/aarch64.md (mov<mode>): Change.
4453 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4454 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
4455 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
4457 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
4459 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
4460 for storage order barriers.
4462 2017-08-11 Martin Liska <mliska@suse.cz>
4465 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
4466 variables of void type.
4468 2017-08-11 Martin Liska <mliska@suse.cz>
4470 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
4471 TARGET_SUPPORTS_ALIASES.
4472 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
4473 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
4474 (optimize_weakref): Likewise.
4475 * symtab.c (symtab_node::noninterposable_alias): Likewise.
4476 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
4477 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
4479 2017-08-11 Martin Liska <mliska@suse.cz>
4482 * config/i386/i386.c (make_resolver_func): Do complete
4483 refactoring of the function.
4485 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4488 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
4489 * config/i386/i386.c (ix86_stack_protect_guard): Use
4490 ix86_stack_protect_guard_symbol_str to generate varible declaration.
4491 * doc/invoke.texi (x86 Options): Document
4492 -mstack-protector-guard-symbol= option.
4494 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
4496 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
4497 * config/i386/i386.c (ix86_split_stack_guard): New function.
4498 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
4499 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
4500 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
4501 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
4502 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
4503 (split_stack_space_check): Call ix86_split_stack_guard.
4505 2017-08-10 Martin Sebor <msebor@redhat.com>
4507 * print-tree.c (print_node): Print location using the established
4509 Replace spaces with colons.
4510 (debug_raw, debug): Ditto.
4512 2017-08-10 Martin Sebor <msebor@redhat.com>
4515 * pretty-print.c (pp_format): Correct the handling of %s precision.
4517 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
4520 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
4522 (ix86_finalize_stack_frame_flags): This. Also clear
4523 frame_pointer_needed if -fno-omit-frame-pointer is used without
4525 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
4526 with ix86_finalize_stack_frame_flags.
4527 (ix86_expand_epilogue): Likewise.
4528 (ix86_expand_split_stack_prologue): Likewise.
4529 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
4531 2017-08-10 Martin Liska <mliska@suse.cz>
4534 * c-attribs.c (handle_target_attribute):
4535 Report warning for an empty string argument of target attribute.
4537 2017-08-09 Jakub Jelinek <jakub@redhat.com>
4540 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
4542 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
4543 or DECL_NONLOCAL labels.
4544 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
4545 or DECL_NONLOCAL labels here.
4547 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
4549 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
4550 to indicate when early gimple folding has been disabled.
4551 (rs6000_gimple_fold_builtin): Add debug content.
4552 (rs6000_invalid_builtin): Fix whitespace.
4553 (rs6000_expand_builtin): Fix whitespace.
4554 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
4556 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
4559 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
4560 SAVE_MULTIPLE if not all the registers that saves, should be saved.
4562 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
4564 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
4565 (qdf24xx): Likewise.
4566 * config/aarch64/aarch64.md: Include falkor.md.
4567 * config/aarch64/falkor.md: New.
4569 2017-08-09 Marek Polacek <polacek@redhat.com>
4572 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
4573 * diagnostic.c (emit_diagnostic): Add a comment.
4574 (emit_diagnostic_valist): New function.
4576 2017-08-09 Marek Polacek <polacek@redhat.com>
4579 * input.c (make_location): New overload.
4580 * input.h (make_location): Declare.
4582 2017-08-08 Alan Modra <amodra@gmail.com>
4583 H.J. Lu <hongjiu.lu@intel.com>
4586 * gcc.c (NO_PIE_SPEC): Delete.
4587 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
4589 (LINK_PIE_SPEC): ..to here.
4590 (LINK_COMMAND_SPEC): Support -no-pie.
4591 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
4592 chain of crtbegin*.o selection, update for PIE_SPEC changes and
4594 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
4595 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
4596 (ENDFILE_CRTEND_SPEC): Similarly.
4598 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
4601 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
4602 (mstack-protector-guard-offset=): Ditto.
4603 * config/i386/i386.c (ix86_option_override): Handle
4604 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
4606 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
4607 ix86_stack_protect_guard_offset variables.
4608 (TARGET_STACK_PROTECT_GUARD): Always define.
4609 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
4610 and -mstack-protector-guard-offset= options.
4612 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4614 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
4615 boundary case for the last candidate.
4617 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4619 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
4620 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
4622 2017-08-08 Tamar Christina <tamar.christina@arm.com>
4625 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
4626 * config/aarch64/aarch64-builtins.c
4627 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
4628 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
4629 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
4631 2017-08-08 Tamar Christina <tamar.christina@arm.com>
4632 Andrew Pinski <pinskia@gmail.com>
4635 * internal-fn.def (XORSIGN): New.
4636 * optabs.def (xorsign_optab): New.
4637 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
4638 (convert_expand_mult_copysign): New.
4639 (pass_optimize_widening_mul::execute): Call
4640 convert_expand_mult_copysign.
4642 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4644 PR tree-optimization/81354
4645 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
4646 Insert on edges rather than explicitly creating landing pads.
4647 (analyze_candidates_and_replace): Commit edge inserts.
4649 2017-08-08 Richard Biener <rguenther@suse.de>
4652 * tree-ssa-loop-niter.c: Include tree-dfa.h.
4653 (expand_simple_operations): Also look through ADDR_EXPRs with
4654 MEM_REF bases treating them as POINTER_PLUS_EXPR.
4656 2017-08-08 Richard Biener <rguenther@suse.de>
4658 PR tree-optimization/81723
4659 * tree-vect-slp.c (struct bst_traits): New hash traits.
4660 (bst_fail): New global.
4661 (vect_build_slp_tree_2): New worker, split out from ...
4662 (vect_build_slp_tree): ... this now wrapping it with using
4663 bst_fail set to cache SLP tree build fails. Properly handle
4665 (vect_analyze_slp_instance): Allocate and free bst_fail.
4667 2017-08-08 Martin Liska <mliska@suse.cz>
4670 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
4671 LABEL_DECLs that can be from a different function.
4673 2017-08-08 Bin Cheng <bin.cheng@arm.com>
4675 PR tree-optimization/81744
4676 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
4677 loop's number of iterations.
4679 2017-08-08 Martin Liska <mliska@suse.cz>
4681 * asan.c: Include header files.
4682 * attribs.c (build_decl_attribute_variant): New function moved
4684 (build_type_attribute_qual_variant): Likewise.
4685 (cmp_attrib_identifiers): Likewise.
4686 (simple_cst_list_equal): Likewise.
4687 (omp_declare_simd_clauses_equal): Likewise.
4688 (attribute_value_equal): Likewise.
4689 (comp_type_attributes): Likewise.
4690 (build_type_attribute_variant): Likewise.
4691 (lookup_ident_attribute): Likewise.
4692 (remove_attribute): Likewise.
4693 (merge_attributes): Likewise.
4694 (merge_type_attributes): Likewise.
4695 (merge_decl_attributes): Likewise.
4696 (merge_dllimport_decl_attributes): Likewise.
4697 (handle_dll_attribute): Likewise.
4698 (attribute_list_equal): Likewise.
4699 (attribute_list_contained): Likewise.
4700 * attribs.h (lookup_attribute): New function moved from tree.[ch].
4701 (lookup_attribute_by_prefix): Likewise.
4702 * bb-reorder.c: Include header files.
4703 * builtins.c: Likewise.
4704 * calls.c: Likewise.
4705 * cfgexpand.c: Likewise.
4706 * cgraph.c: Likewise.
4707 * cgraphunit.c: Likewise.
4708 * convert.c: Likewise.
4709 * dwarf2out.c: Likewise.
4710 * final.c: Likewise.
4711 * fold-const.c: Likewise.
4712 * function.c: Likewise.
4713 * gimple-expr.c: Likewise.
4714 * gimple-fold.c: Likewise.
4715 * gimple-pretty-print.c: Likewise.
4716 * gimple.c: Likewise.
4717 * gimplify.c: Likewise.
4718 * hsa-common.c: Likewise.
4719 * hsa-gen.c: Likewise.
4720 * internal-fn.c: Likewise.
4721 * ipa-chkp.c: Likewise.
4722 * ipa-cp.c: Likewise.
4723 * ipa-devirt.c: Likewise.
4724 * ipa-fnsummary.c: Likewise.
4725 * ipa-inline.c: Likewise.
4726 * ipa-visibility.c: Likewise.
4728 * lto-cgraph.c: Likewise.
4729 * omp-expand.c: Likewise.
4730 * omp-general.c: Likewise.
4731 * omp-low.c: Likewise.
4732 * omp-offload.c: Likewise.
4733 * omp-simd-clone.c: Likewise.
4734 * opts-global.c: Likewise.
4735 * passes.c: Likewise.
4736 * predict.c: Likewise.
4737 * sancov.c: Likewise.
4738 * sanopt.c: Likewise.
4739 * symtab.c: Likewise.
4740 * toplev.c: Likewise.
4741 * trans-mem.c: Likewise.
4742 * tree-chkp.c: Likewise.
4743 * tree-eh.c: Likewise.
4744 * tree-into-ssa.c: Likewise.
4745 * tree-object-size.c: Likewise.
4746 * tree-parloops.c: Likewise.
4747 * tree-profile.c: Likewise.
4748 * tree-ssa-ccp.c: Likewise.
4749 * tree-ssa-live.c: Likewise.
4750 * tree-ssa-loop.c: Likewise.
4751 * tree-ssa-sccvn.c: Likewise.
4752 * tree-ssa-structalias.c: Likewise.
4753 * tree-ssa.c: Likewise.
4754 * tree-streamer-in.c: Likewise.
4755 * tree-vectorizer.c: Likewise.
4756 * tree-vrp.c: Likewise.
4758 * ubsan.c: Likewise.
4759 * varasm.c: Likewise.
4760 * varpool.c: Likewise.
4761 * tree.c: Remove functions moved to attribs.[ch].
4763 * config/aarch64/aarch64.c: Add attrs.h header file.
4764 * config/alpha/alpha.c: Likewise.
4765 * config/arc/arc.c: Likewise.
4766 * config/arm/arm.c: Likewise.
4767 * config/avr/avr.c: Likewise.
4768 * config/bfin/bfin.c: Likewise.
4769 * config/c6x/c6x.c: Likewise.
4770 * config/cr16/cr16.c: Likewise.
4771 * config/cris/cris.c: Likewise.
4772 * config/darwin.c: Likewise.
4773 * config/epiphany/epiphany.c: Likewise.
4774 * config/fr30/fr30.c: Likewise.
4775 * config/frv/frv.c: Likewise.
4776 * config/ft32/ft32.c: Likewise.
4777 * config/h8300/h8300.c: Likewise.
4778 * config/i386/winnt.c: Likewise.
4779 * config/ia64/ia64.c: Likewise.
4780 * config/iq2000/iq2000.c: Likewise.
4781 * config/lm32/lm32.c: Likewise.
4782 * config/m32c/m32c.c: Likewise.
4783 * config/m32r/m32r.c: Likewise.
4784 * config/m68k/m68k.c: Likewise.
4785 * config/mcore/mcore.c: Likewise.
4786 * config/microblaze/microblaze.c: Likewise.
4787 * config/mips/mips.c: Likewise.
4788 * config/mmix/mmix.c: Likewise.
4789 * config/mn10300/mn10300.c: Likewise.
4790 * config/moxie/moxie.c: Likewise.
4791 * config/msp430/msp430.c: Likewise.
4792 * config/nds32/nds32-isr.c: Likewise.
4793 * config/nds32/nds32.c: Likewise.
4794 * config/nios2/nios2.c: Likewise.
4795 * config/nvptx/nvptx.c: Likewise.
4796 * config/pa/pa.c: Likewise.
4797 * config/pdp11/pdp11.c: Likewise.
4798 * config/powerpcspe/powerpcspe.c: Likewise.
4799 * config/riscv/riscv.c: Likewise.
4800 * config/rl78/rl78.c: Likewise.
4801 * config/rx/rx.c: Likewise.
4802 * config/s390/s390.c: Likewise.
4803 * config/sh/sh.c: Likewise.
4804 * config/sol2.c: Likewise.
4805 * config/sparc/sparc.c: Likewise.
4806 * config/spu/spu.c: Likewise.
4807 * config/stormy16/stormy16.c: Likewise.
4808 * config/tilegx/tilegx.c: Likewise.
4809 * config/tilepro/tilepro.c: Likewise.
4810 * config/v850/v850.c: Likewise.
4811 * config/vax/vax.c: Likewise.
4812 * config/visium/visium.c: Likewise.
4813 * config/xtensa/xtensa.c: Likewise.
4815 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
4818 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
4819 constraints since the -mupper-regs-* switches have been
4821 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
4822 into a vector from a double word element that was extracted from
4823 another vector, and eliminate extra XXPERMDI instructions.
4824 (vsx_concat_<mode>_2): Likewise.
4825 (vsx_concat_<mode>_3): Likewise.
4826 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
4827 concat to allow optimizing inserts from previous extracts.
4829 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
4831 * config/i386/i386.c (ix86_stack_protect_guard): Generate
4832 memory reference to a SSP offset in TLS address space.
4833 (ix86_print_operand) <case '@'>: Remove.
4834 (ix86_print_operand_punct_valid_p): Remove '@' code.
4835 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
4837 (stack_tls_protect_set_<mode>): Remove.
4838 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
4839 (stack_tls_protect_test_<mode>): Remove.
4840 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
4842 2017-08-07 Olivier Hainque <hainque@adacore.com>
4845 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
4847 2017-08-07 Douglas Rupp <rupp@adacore.com>
4849 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
4850 variable was referenced as multidir in command.
4852 2017-08-07 Jakub Jelinek <jakub@redhat.com>
4855 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
4858 2017-08-07 Martin Liska <mliska@suse.cz>
4860 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
4861 * config/rl78/rl78.c: Add include of attribs.h.
4862 * config/sh/sh.c: Likewise.
4863 * config/v850/v850.c: Likewise.
4865 2017-08-07 Tom de Vries <tom@codesourcery.com>
4868 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
4870 2017-08-07 Martin Liska <mliska@suse.cz>
4872 * config/mips/mips.c: Include attribs.h.
4874 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
4877 * doc/invoke.texi: Document change in behvaior for -Ofast for
4880 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
4882 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
4884 (aarch64_pop_regs): Likewise.
4885 (aarch64_gen_load_pair): Likewise.
4886 (aarch64_save_callee_saves): Likewise.
4887 (aarch64_restore_callee_saves): Likewise.
4889 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4891 * config/i386/i386.c: Revert the last change.
4893 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4896 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
4898 (ix86_finalize_stack_frame_flags): This. Also clear
4899 frame_pointer_needed if -fno-omit-frame-pointer is used without
4901 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
4902 with ix86_finalize_stack_frame_flags.
4903 (ix86_expand_epilogue): Likewise.
4904 (ix86_expand_split_stack_prologue): Likewise.
4906 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
4909 * config/i386/i386.c (get_builtin_code_for_version): Set priority
4910 to P_AES for Westmere.
4912 2017-08-07 Jonathan Yong <10walls@gmail.com>
4914 * config/i386/mingw.opt (fset-stack-executable): Removed.
4915 * config/i386/cygming.opt (fset-stack-executable): Moved
4917 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
4919 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
4921 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
4923 2017-08-07 Marek Polacek <polacek@redhat.com>
4926 * fold-const.c (fold_indirect_ref_1): Check type_domain.
4928 2017-08-07 Martin Liska <mliska@suse.cz>
4930 * attribs.h (canonicalize_attr_name): New function.
4931 (cmp_attribs): Move from c-format.c and adjusted.
4932 (is_attribute_p): Moved from tree.h.
4933 * tree-inline.c: Add new includes.
4934 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
4935 (private_is_attribute_p): Remove.
4936 (private_lookup_attribute): Likewise.
4937 (private_lookup_attribute_by_prefix): Simplify.
4938 (remove_attribute): Use is_attribute_p.
4939 * tree.h: Remove removed declarations.
4941 2017-08-07 Jakub Jelinek <jakub@redhat.com>
4944 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
4945 instead of computing it in the function. Formatting fix.
4946 (expand_case): Don't rely on default_edge being the first edge,
4947 clear it if removing it, pass default_edge to
4948 emit_case_dispatch_table.
4949 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
4952 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
4954 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
4955 insn in the function, emit NOP after the insn.
4957 2017-08-06 Tom de Vries <tom@codesourcery.com>
4959 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
4962 2017-08-06 Tom de Vries <tom@codesourcery.com>
4964 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
4967 2017-08-04 Yury Gribov <tetra2005@gmail.com>
4969 PR tree-optimization/57371
4970 * match.pd: New pattern.
4972 2017-08-04 Marek Polacek <polacek@redhat.com>
4975 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
4976 perform the computation in offset_int.
4978 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
4980 PR tree-optimization/81136
4981 * tree-vectorizer.h: Include tree-hash-traits.h.
4982 (vec_base_alignments): New typedef.
4983 (vec_info): Add a base_alignments field.
4984 (vect_record_base_alignments): Declare.
4985 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
4987 (DR_IS_CONDITIONAL_IN_STMT): New macro.
4988 (create_data_ref): Add an is_conditional_in_stmt argument.
4989 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
4990 the is_conditional_in_stmt field.
4991 (data_ref_loc): Add an is_conditional_in_stmt field.
4992 (get_references_in_stmt): Set the is_conditional_in_stmt field.
4993 (find_data_references_in_stmt): Update call to create_data_ref.
4994 (graphite_find_data_references_in_stmt): Likewise.
4995 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
4996 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
4997 (vect_record_base_alignment): New function.
4998 (vect_record_base_alignments): Likewise.
4999 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
5000 for nested statements even if we fail to compute a misalignment.
5001 Use pooled base alignments for unconditional references.
5002 (vect_find_same_alignment_drs): Compare base addresses instead
5004 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
5005 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
5007 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5009 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
5010 Add an explicit name for the enum. Use auto_vec for slp_instances
5012 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
5014 (_bb_vec_info): Add a constructor and destructor.
5015 (vinfo_for_stmt): Return NULL for uids of -1 as well.
5016 (destroy_loop_vec_info): Delete.
5017 (vect_destroy_datarefs): Likewise.
5018 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
5019 (vec_info::vec_info): New function.
5020 (vec_info::~vec_info): Likewise.
5021 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
5022 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
5023 destroy_loop_vec_info.
5024 * tree-vect-loop.c (new_loop_vec_info): Replace with...
5025 (_loop_vec_info::_loop_vec_info): ...this.
5026 (destroy_loop_vec_info): Replace with...
5027 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
5028 the stmt_vec_infos. Leave handling of vec_info information to its
5029 destructor. Remove explicit vector releases.
5030 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
5031 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
5032 * tree-vect-slp.c (new_bb_vec_info): Replace with...
5033 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
5034 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
5035 (destroy_bb_vec_info): Replace with...
5036 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
5037 information to its destructor.
5038 (vect_slp_analyze_bb_1): Use new and delete instead of
5039 new_bb_vec_info and destroy_bb_vec_info.
5040 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
5043 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
5045 * tree-data-ref.h (subscript): Add access_fn field.
5046 (data_dependence_relation): Add could_be_independent_p.
5047 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
5048 (same_access_functions): Move to tree-data-ref.c.
5049 * tree-data-ref.c (ref_contains_union_access_p): New function.
5050 (access_fn_component_p): Likewise.
5051 (access_fn_components_comparable_p): Likewise.
5052 (dr_analyze_indices): Add a reference to access_fn_component_p.
5053 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
5055 (constant_access_functions): Likewise.
5056 (add_other_self_distances): Likewise.
5057 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
5058 (initialize_data_dependence_relation): Use XCNEW and remove
5059 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
5060 of access functions that have the same type. Allow the
5061 subsequence to end with different bases in some circumstances.
5062 Record the chosen access functions in SUB_ACCESS_FN.
5063 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
5064 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
5065 (subscript_dependence_tester_1): Likewise dra and drb.
5066 (build_classic_dist_vector): Update calls accordingly.
5067 (subscript_dependence_tester): Likewise.
5068 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
5069 DDR_COULD_BE_INDEPENDENT_P.
5070 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
5071 comp_alias_ddrs instead of may_alias_ddrs.
5072 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
5074 (vect_analyze_data_ref_dependence): Use it if
5075 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
5076 distance vectors if that fails.
5077 (dependence_distance_ge_vf): New function.
5078 (vect_prune_runtime_alias_test_list): Use it. Don't clear
5079 LOOP_VINFO_MAY_ALIAS_DDRS.
5081 2017-08-04 Richard Biener <rguenther@suse.de>
5084 * fold-const.c (fold_binary_loc): Properly restrict
5085 minus_var0 && minus_var1 case when associating undefined overflow
5088 2017-08-04 Tom de Vries <tom@codesourcery.com>
5090 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
5092 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5094 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5095 Don't start diagnostic messages with a capital letter.
5096 * config/rs6000/rs6000.c (rs6000_option_override_internal):
5098 (rs6000_invalid_builtin): Likewise.
5099 (rs6000_trampoline_init): Likewise.
5101 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5104 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
5105 after setting changeable df flags.
5107 2017-08-03 Richard Biener <rguenther@suse.de>
5109 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
5110 up if the use is in USE - X.
5112 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
5114 * toplev.c (dumpfile.h): New include.
5115 (internal_error_reentered): New static function. Use it...
5116 (internal_error_function): ...here to handle reentered internal_error.
5118 2017-08-03 Richard Biener <rguenther@suse.de>
5121 * fold-const.c (split_tree): Add minus_var and minus_con
5122 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
5123 here but always use minus_*.
5124 (associate_trees): Assert we never associate with MINUS_EXPR
5125 and NULL first operand. Do not recurse for PLUS_EXPR operands
5126 when associating as MINUS_EXPR either.
5127 (fold_binary_loc): Track minus_var and minus_con.
5129 2017-08-03 Tom de Vries <tom@codesourcery.com>
5132 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
5133 ACCEL_COMPILER, apply finish_options on
5134 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
5136 2017-08-03 Tom de Vries <tom@codesourcery.com>
5139 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
5140 function_entry_patch_area_size > 0.
5142 2017-08-03 Jakub Jelinek <jakub@redhat.com>
5145 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
5146 instead of 10??LU, perform unit multiplication in wide_int,
5147 don't change alloc_object_size_limit if the limit is larger
5150 PR tree-optimization/81655
5151 PR tree-optimization/81588
5152 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
5153 the case when ranges[i].low and high are 1 for unsigned type with
5157 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
5158 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
5160 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5162 * tree-vrp.h: Add include guard.
5164 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
5167 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
5168 (ud2): New insn pattern.
5169 * config/i386/i386.c (ix86_expand_epilogue):
5170 For naked functions, generate ud2 instead of trap insn.
5172 2017-08-02 Marek Polacek <polacek@redhat.com>
5175 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
5177 2017-08-02 Tom de Vries <tom@codesourcery.com>
5178 Cesar Philippidis <cesar@codesourcery.com>
5180 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
5181 Add missing edge probabilities.
5183 2017-08-02 Tamar Christina <tamar.christina@arm.com>
5185 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
5188 2017-08-02 Jakub Jelinek <jakub@redhat.com>
5191 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
5192 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
5193 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
5194 prologue_seq sequences - if any.
5196 2017-08-02 Richard Biener <rguenther@suse.de>
5198 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
5199 via vectors if supported, integer extracts via punning if supported
5200 or otherwise vector extracts.
5202 2017-08-02 Richard Biener <rguenther@suse.de>
5204 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
5206 (bitmap_insert_into_set): ... this.
5208 2017-08-02 Richard Biener <rguenther@suse.de>
5210 PR tree-optimization/81633
5212 2015-08-17 Alan Hayward <alan.hayward@arm.com>
5214 PR tree-optimization/71752
5215 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
5217 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
5219 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
5220 (machine_function::call_ms2sysv_pad_out): Remove field.
5221 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
5222 (ix86_compute_frame_layout): Likewise.
5224 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
5227 * config/i386/i386.c (ix86_set_func_type): Disallow naked
5228 attribute with interrupt attribute.
5230 2017-08-01 Andrew Pinski <apinski@cavium.com>
5232 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
5233 BIT_INSERT_EXPR's operand 1
5234 to see if the types precision matches.
5236 2017-08-01 Martin Liska <mliska@suse.cz>
5239 * builtins.c (expand_builtin_memcpy_args): Remove.
5240 (expand_builtin_memcpy): Call newly added function
5241 expand_builtin_memory_copy_args.
5242 (expand_builtin_memcpy_with_bounds): Likewise.
5243 (expand_builtin_mempcpy): Remove last argument.
5244 (expand_builtin_mempcpy_with_bounds): Likewise.
5245 (expand_builtin_memory_copy_args): New function created from
5246 expand_builtin_mempcpy_args with small modifications.
5247 (expand_builtin_mempcpy_args): Remove.
5248 (expand_builtin_stpcpy): Remove unused argument.
5249 (expand_builtin): Likewise.
5250 (expand_builtin_with_bounds): Likewise.
5252 2017-08-01 Martin Liska <mliska@suse.cz>
5255 Make mempcpy more optimal (PR middle-end/70140).
5257 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5260 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
5261 __builtin_vec_cmpne verify both arguments are compatible vectors
5262 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
5263 verify arg1_type is a pointer or array type. For __builtin_vec_st,
5264 move computation of aligned to after checking the argument types.
5268 * config/rs6000/vsx.md (vextract_fp_from_shorth,
5269 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
5272 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
5273 Doug Rupp <rupp@adacore.com>
5274 Olivier Hainque <hainque@adacore.com>
5276 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
5277 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
5279 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
5280 for TARGET_OS_CPP_BUILTIN.
5281 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
5282 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
5284 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
5285 passing required abi options to the assembler for EABI configurations.
5286 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
5287 of .text.hot and .text.unlikely sections for kernel modules when
5288 using ARM style exceptions.
5289 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
5290 options. Add EXTRA_CC1_SPEC.
5291 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
5293 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
5295 (ARM_TARGET2_DWARF_FORMAT): Define.
5296 * config/arm/t-vxworks: Adjust multilib control to removal of the
5297 Diab command line options.
5299 2017-08-01 Martin Liska <mliska@suse.cz>
5301 PR gcov-profile/81561
5302 * gcov.c (unblock): Make unblocking safe as we need to preserve
5303 index correspondence of blocks and block_lists.
5305 2017-08-01 Richard Biener <rguenther@suse.de>
5307 PR tree-optimization/81181
5308 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
5309 (compute_antic): ... end of iteration here.
5311 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
5313 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
5314 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
5315 (ftree-slp-vectorize): Likewise.
5316 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
5317 can no longer be set independent of flag_tree_loop_vectorize.
5318 * omp-general.c (emp_max_vf): Likewise.
5319 * opts.c (enable_fdo_optimizations): Remove references to
5320 flag_tree_vectorize, these are now implicit.
5321 (common_handle_option): Remove handling for OPT_ftree_vectorize,
5322 and leave it for the options machinery.
5324 2017-08-01 Martin Liska <mliska@suse.cz>
5327 * builtins.c (expand_builtin_memcpy_args): Remove.
5328 (expand_builtin_memcpy): Call newly added function
5329 expand_builtin_memory_copy_args.
5330 (expand_builtin_memcpy_with_bounds): Likewise.
5331 (expand_builtin_mempcpy): Remove last argument.
5332 (expand_builtin_mempcpy_with_bounds): Likewise.
5333 (expand_builtin_memory_copy_args): New function created from
5334 expand_builtin_mempcpy_args with small modifications.
5335 (expand_builtin_mempcpy_args): Remove.
5336 (expand_builtin_stpcpy): Remove unused argument.
5337 (expand_builtin): Likewise.
5338 (expand_builtin_with_bounds): Likewise.
5340 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5343 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
5344 print "ds:" only for immediates in generic address space.
5346 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
5349 * config/i386/i386.c (ix86_funciton_naked): New prototype.
5350 (ix86_function_ok_for_sibcall): Return false for naked functions.
5352 2017-08-01 Richard Biener <rguenther@suse.de>
5354 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
5355 (compute_antic): Seed worklist with exit block predecessors.
5356 * cfganal.c (dfs_find_deadend): For a cycle return the source
5357 of the edge closing it.
5359 2017-08-01 Tamar Christina <tamar.christina@arm.com>
5361 * config/aarch64/aarch64.c
5362 (aarch64_can_const_movi_rtx_p): Move 0 check.
5364 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5366 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
5367 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
5369 * match.pd: Ditto in address comparison pattern.
5371 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5373 PR tree-optimization/81627
5374 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
5375 closed ssa form for store-store chain.
5377 2017-08-01 Bin Cheng <bin.cheng@arm.com>
5379 PR tree-optimization/81620
5380 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
5381 for store-store chain.
5383 2017-08-01 Jakub Jelinek <jakub@redhat.com>
5385 PR tree-optimization/81588
5386 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
5387 ranges[i].in_p, invert comparison code ccode. For >/>=,
5388 swap rhs1 and rhs2 and comparison code unconditionally,
5389 for </<= don't do that. Don't swap rhs1/rhs2 again if
5390 ranges[i].in_p, instead invert comparison code ccode if
5391 opcode or oe->rank is BIT_IOR_EXPR.
5394 * optabs.def (vec_extract_optab, vec_init_optab): Change from
5395 a direct optab to conversion optab.
5396 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
5397 with GET_MODE_INNER as last argument instead of optab_handler.
5398 * expmed.c (extract_bit_field_1): Likewise. Use vector from
5399 vector extraction if possible and optab is available.
5400 * expr.c (store_constructor): Use convert_optab_handler instead
5401 of optab_handler. Use vector initialization from smaller
5402 vectors if possible and optab is available.
5403 * tree-vect-stmts.c (vectorizable_load): Likewise.
5404 * doc/md.texi (vec_extract, vec_init): Document that the optabs
5406 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
5407 of vec_init from half-sized vectors with the same element mode.
5408 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
5409 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
5410 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
5411 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
5412 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
5413 after mode in gen_vec_extract* calls.
5414 (vec_extract<mode>): Renamed to ...
5415 (vec_extract<mode><ssescalarmodelower>): ... this.
5416 (vec_extract<mode><ssehalfvecmodelower>): New expander.
5417 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
5418 element mode after mode in gen_vec_init* calls.
5419 (VEC_INIT_HALF_MODE): New mode iterator.
5420 (vec_init<mode>): Renamed to ...
5421 (vec_init<mode><ssescalarmodelower>): ... this.
5422 (vec_init<mode><ssehalfvecmodelower>): New expander.
5423 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
5424 (vec_extractv2sfsf): ... this.
5425 (vec_initv2sf): Renamed to ...
5426 (vec_initv2sfsf): ... this.
5427 (vec_extractv2si): Renamed to ...
5428 (vec_extractv2sisi): ... this.
5429 (vec_initv2si): Renamed to ...
5430 (vec_initv2sisi): ... this.
5431 (vec_extractv4hi): Renamed to ...
5432 (vec_extractv4hihi): ... this.
5433 (vec_initv4hi): Renamed to ...
5434 (vec_initv4hihi): ... this.
5435 (vec_extractv8qi): Renamed to ...
5436 (vec_extractv8qiqi): ... this.
5437 (vec_initv8qi): Renamed to ...
5438 (vec_initv8qiqi): ... this.
5439 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
5440 (vec_init<mode>): Renamed to ...
5441 (vec_init<mode><VEC_base_l>): ... this.
5442 (vec_extract<mode>): Renamed to ...
5443 (vec_extract<mode><VEC_base_l>): ... this.
5444 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
5445 (vec_initv2sfsf): ... this.
5446 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
5447 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5448 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
5449 element mode after mode in gen_vec_init* calls.
5450 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
5451 (vec_init<mode><Vel>): ... this.
5452 (vec_extract<mode>): Renamed to ...
5453 (vec_extract<mode><Vel>): ... this.
5454 * config/aarch64/iterators.md (Vel): New mode attribute.
5455 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
5456 Add element mode after mode in gen_vec_extract* calls.
5457 * config/s390/vector.md (non_vec_l): New mode attribute.
5458 (vec_extract<mode>): Renamed to ...
5459 (vec_extract<mode><non_vec_l>): ... this.
5460 (vec_init<mode>): Renamed to ...
5461 (vec_init<mode><non_vec_l>): ... this.
5462 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
5463 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
5465 * config/arm/iterators.md (V_elem_l): New mode attribute.
5466 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
5467 (vec_extract<mode><V_elem_l>): ... this.
5468 (vec_extractv2di): Renamed to ...
5469 (vec_extractv2didi): ... this.
5470 (vec_init<mode>): Renamed to ...
5471 (vec_init<mode><V_elem_l>): ... this.
5472 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
5473 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
5474 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
5475 Add element mode after gen_vec_extract* calls.
5476 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
5477 (vec_init<mode><unitmode>): ... this.
5478 (vec_extract<mode>): Renamed to ...
5479 (vec_extract<mode><unitmode>): ... this.
5480 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
5481 (vec_init<mode><unitmode>): ... this.
5482 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
5483 (vec_initv2sfsf): ... this.
5484 (vec_extractv2sf): Renamed to ...
5485 (vec_extractv2sfsf): ... this.
5486 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
5487 Add element mode after gen_vec_extract* calls.
5488 * config/mips/mips.md (unitmode): New mode iterator.
5489 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
5490 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
5491 * config/spu/spu.md (inner_l): New mode attribute.
5492 (vec_init<mode>): Renamed to ...
5493 (vec_init<mode><inner_l>): ... this.
5494 (vec_extract<mode>): Renamed to ...
5495 (vec_extract<mode><inner_l>): ... this.
5496 * config/sparc/sparc.md (veltmode): New mode iterator.
5497 (vec_init<VMALL:mode>): Renamed to ...
5498 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
5499 * config/ia64/vect.md (vec_initv2si): Renamed to ...
5500 (vec_initv2sisi): ... this.
5501 (vec_initv2sf): Renamed to ...
5502 (vec_initv2sfsf): ... this.
5503 (vec_extractv2sf): Renamed to ...
5504 (vec_extractv2sfsf): ... this.
5505 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
5506 (vec_init<mode>): Renamed to ...
5507 (vec_init<mode><VEC_base_l>): ... this.
5508 (vec_extract<mode>): Renamed to ...
5509 (vec_extract<mode><VEC_base_l>): ... this.
5510 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
5511 (vec_initv2sfsf): ... this.
5512 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
5513 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
5514 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
5515 gen_vec_init* calls.
5517 2017-08-01 Richard Biener <rguenther@suse.de>
5519 PR tree-optimization/81297
5520 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
5521 TREE_OVERFLOW from INTEGER_CSTs.
5523 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
5525 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
5527 2017-07-31 Carl Love <cel@us.ibm.com>
5529 * config/rs6000/rs6000-c: Add support for built-in functions
5530 vector signed char vec_xl_be (signed long long, signed char *);
5531 vector unsigned char vec_xl_be (signed long long, unsigned char *);
5532 vector signed int vec_xl_be (signed long long, signed int *);
5533 vector unsigned int vec_xl_be (signed long long, unsigned int *);
5534 vector signed long long vec_xl_be (signed long long, signed long long *);
5535 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
5536 vector signed short vec_xl_be (signed long long, signed short *);
5537 vector unsigned short vec_xl_be (signed long long, unsigned short *);
5538 vector double vec_xl_be (signed long long, double *);
5539 vector float vec_xl_be (signed long long, float *);
5540 * config/rs6000/altivec.h (vec_xl_be): Add #define.
5541 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
5542 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
5544 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
5545 (altivec_expand_builtin): Add switch statement to call
5546 altivec_expand_xl_be for each builtin.
5547 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
5548 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
5549 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
5550 __builtin_vsx_le_be_v16qi.
5551 * doc/extend.texi: Update the built-in documentation file for the
5552 new built-in functions.
5554 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5557 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
5559 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5561 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5563 * config.gcc: Add z14.
5564 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
5565 CPU model numbers for z13s and z14.
5566 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
5568 * config/s390/s390-opts.h (enum processor_type): Rename
5569 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5570 * config/s390/s390.c (processor_table): Add field for CPU name to
5571 be passed to Binutils.
5572 (s390_asm_output_machine_for_arch): Use the new field in
5573 processor_table for Binutils.
5574 (s390_expand_builtin): Replace arch12 with z14.
5575 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
5576 (s390_get_sched_attrmask): Likewise.
5577 (s390_get_unit_mask): Likewise.
5578 * config/s390/s390.opt: Add z14 to processor_type enum.
5580 2017-07-31 Martin Jambor <mjambor@suse.cz>
5583 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
5584 regardless of optimization level.
5586 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
5587 Martin Liska <mliska@suse.cz>
5589 * predict.def: Remove old comment and adjust probability.
5590 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
5593 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
5596 * config/i386/i386.c (ix86_function_naked): New function.
5597 (ix86_can_use_return_insn_p): Return false for naked functions.
5598 (ix86_expand_prologue): Skip prologue for naked functions.
5599 (ix86_expand_epilogue): Skip epilogue for naked functions
5600 and emit trap instruction.
5601 (ix86_warn_func_return): New function.
5602 (ix86_attribute_table): Add "naked" attribute specification.
5603 (TARGET_WARN_FUNC_RETURN): Define.
5604 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
5606 2017-07-31 Martin Liska <mliska@suse.cz>
5608 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
5609 (dump_gimple_bb_header): Always dump BB info.
5610 (pp_cfg_jump): Do not append info about BB when dumping a jump.
5612 2017-07-31 Martin Liska <mliska@suse.cz>
5615 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
5616 also with current_function_decl non-null equality.
5618 2017-07-31 Jakub Jelinek <jakub@redhat.com>
5621 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
5622 change type to the element type, instead add eltype variable and
5623 use it where we are interested in the element type.
5625 PR tree-optimization/81603
5626 * ipa-polymorphic-call.c
5627 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
5628 offset arithmetic in offset_int, bail out if the resulting bit offset
5629 doesn't fit into shwi.
5631 2017-07-31 Martin Liska <mliska@suse.cz>
5633 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
5634 (gimplify_save_expr): Fix comment.
5636 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
5639 * config/i386/i386.c (ix86_function_arg): Update arguments for
5641 (ix86_compute_frame_layout): Set the initial stack offset to
5642 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
5643 INCOMING_FRAME_SP_OFFSET.
5644 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
5645 stack before exception handler returns.
5646 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
5647 the 'ERROR_CODE' for exception handler.
5649 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
5651 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
5652 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
5653 (ASM_OUTPUT_REG_POP): Ditto.
5654 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
5655 instead of asm_fprintf to output pure string.
5657 2017-07-29 Jakub Jelinek <jakub@redhat.com>
5659 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
5660 to imported_module_or_decl hook.
5661 (debug_nothing_tree_tree_tree_bool): Remove.
5662 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
5663 * debug.c (do_nothing_debug_hooks): Use
5664 debug_nothing_tree_tree_tree_bool_bool instead of
5665 debug_nothing_tree_tree_tree_bool.
5666 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
5667 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
5668 * sdbout.c (sdb_debug_hooks): Likewise.
5669 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
5670 (gen_namespace_die): Add DW_AT_export_symbols attribute if
5672 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
5673 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
5674 attribute, don't add anything.
5676 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5678 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
5679 (fold_build2_stat_loc): Likewise.
5680 (fold_build3_stat_loc): Likewise.
5681 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
5682 (fold_build1_loc): Remove macro.
5683 (fold_build2_loc): Likewise.
5684 (fold_build3_loc): Likewise.
5686 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5688 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
5689 (gimple_build_debug_bind_source_stat): Likewise.
5690 * gimple.h (gimple_build_debug_bind): Remove macro.
5691 (gimple_build_debug_bind_source): Likewise.
5693 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5695 * bitmap.c (bitmap_alloc): Adjust.
5696 (bitmap_gc_alloc): Likewise.
5697 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
5699 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5701 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
5702 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
5703 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
5704 (bitmap_gc_alloc_stat): Likewise.
5705 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
5707 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5709 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
5710 * rtl.h (shallow_copy_rtx): Remove macro.
5712 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5714 * emit-rtl.c (gen_raw_REG): Adjust.
5715 * gengenrtl.c (gendef): Likewise.
5716 * rtl.c (rtx_alloc_stat): Remove _stat from name.
5717 * rtl.h (rtx_alloc): Remove macro.
5719 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5721 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
5722 (build_tree_list_stat): Likewise.
5723 * tree.h (build_tree_list): Remove macro.
5724 (build_tree_list_vec): Likewise.
5726 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5728 * tree.c (make_vector_stat): Remove _stat from name.
5729 (build_vector_stat): Likewise.
5730 * tree.h (make_vector_stat): Remove macro.
5731 (build_vector_stat): Likewise.
5733 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5735 * tree.h (build_var_debug_value): Remove prototype.
5737 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5739 * tree.c (tree_cons_stat): Remove _stat from name.
5740 * tree.h (tree_cons): Remove macro.
5742 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5744 * tree.c (build_vl_exp_stat): Remove _stat from name.
5745 * tree.h (build_vl_exp): Remove macro.
5747 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5749 * tree.c (build_decl_stat): Remove _stat from name.
5750 * tree.h (build_decl): Remove macro.
5752 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5754 * gimple.c (gimple_build_with_ops_stat): Adjust.
5755 (gimple_alloc_stat): Remove _stat from name.
5756 * gimple.h (gimple_alloc): Remove macro.
5758 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5760 * tree.c (make_tree_vec_stat): Remove _stat from name.
5761 (grow_tree_vec_stat): Likewise.
5762 * tree.h (make_tree_vec_stat): Adjust prototype.
5763 (grow_tree_vec_stat): Likewise.
5764 (make_tree_vec): Remove macro.
5765 (grow_tree_vec): Likewise.
5767 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5769 * fold-const.c (fold_build1_stat_loc): Adjust.
5770 (fold_build2_stat_loc): Likewise.
5771 (fold_build3_stat_loc): Likewise.
5772 * tree.c (build0_stat): Remove _stat from name.
5773 (build1_stat): Likewise.
5774 (build2_stat): Likewise.
5775 (build3_stat): Likewise.
5776 (build4_stat): Likewise.
5777 (build5_stat): Likewise.
5778 * tree.h (build1_loc): Remove macro, and rename _stat function
5780 (build2_loc): Likewise.
5781 (build3_loc): Likewise.
5782 (build4_loc): Likewise.
5783 (build5_loc): Likewise.
5785 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5787 * tree.c (make_int_cst_stat): Remove _stat from name.
5788 * tree.h (make_int_cst_stat): Adjust prototype.
5789 (make_int_cst): Remove macro.
5791 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5793 * tree.c (make_tre_binfo_stat): Remove _stat from name.
5794 * tree.h (make_tree_binfo_stat): Adjust prototype.
5795 (make_tree_binfo): Remove.
5797 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5799 * tree.c (copy_node_stat): Rename to copy_node.
5800 (build_distinct_type_copy): Adjust.
5801 * tree.h (copy_node_stat): Adjust prototype.
5802 (copy_node): Remove macro.
5804 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
5806 * tree.c (make_node_stat): rename to make_node.
5807 (build_tree_list_stat): Adjust.
5808 (build0_stat): Likewise.
5809 (build2_stat): Likewise.
5810 (build3_stat): Likewise.
5811 (build4_stat): Likewise.
5812 (build5_stat): Likewise.
5813 (build_decl_stat): Likewise.
5814 * tree.h (make_node_stat): Adjust prototype.
5815 (make_node): remove macro.
5817 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
5819 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
5820 (PPC_FEATURE2_SCV): Likewise.
5821 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
5823 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5825 * config/aarch64/aarch64.c
5826 (aarch64_internal_mov_immediate): Add new special pattern.
5827 * config/aarch64/aarch64.md (*movdi_aarch64):
5828 Add reg/32bit const mov case.
5830 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5831 Richard Sandiford <richard.sandiford@linaro.org>
5833 * config/aarch64/aarch64.md (mov<mode>): Generalize.
5834 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
5835 Add integer and movi cases.
5836 (movi-split-hf-df-sf split, fp16): New.
5837 (enabled): Added TARGET_FP_F16INST.
5838 * config/aarch64/iterators.md (GPF_HF): New.
5839 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
5841 2017-07-28 Tamar Christina <tamar.christina@arm.com>
5843 * config/aarch64/aarch64.c
5844 (aarch64_simd_container_mode): Add prototype.
5845 (aarch64_expand_mov_immediate): Add HI support.
5846 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
5847 (aarch64_can_const_movi_rtx_p): New.
5848 (aarch64_preferred_reload_class):
5849 Remove restrictions of using FP registers for certain SIMD operations.
5850 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
5851 (aarch64_valid_floating_const): Add integer move validation.
5852 (aarch64_simd_imm_scalar_p): Remove.
5853 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
5854 (aarch64_legitimate_constant_p): Expand list of supported cases.
5855 * config/aarch64/aarch64-protos.h
5856 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
5857 (aarch64_reinterpret_float_as_int): New.
5858 (aarch64_simd_imm_scalar_p): Remove.
5859 * config/aarch64/constraints.md (Uvi): New.
5860 (Dd): Split into Ds and new Dd.
5861 * config/aarch64/aarch64.md (*movsi_aarch64):
5863 (*movdi_aarch64): Add SIMD mov case.
5865 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5867 * tree-predcom.c: (struct chain): Handle store-store chain in which
5868 stores for elimination only store loop invariant values.
5869 (execute_pred_commoning_chain): Ditto.
5870 (prepare_initializers_chain_store_elim): Ditto.
5871 (prepare_finalizers): Ditto.
5872 (is_inv_store_elimination_chain): New function.
5873 (initialize_root_vars_store_elim_1): New function.
5875 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5877 * tree-predcom.c: Revise general description of the pass.
5878 (enum chain_type): New enum type for store elimination.
5879 (struct chain): New field supporting store elimination.
5880 (struct component): Ditto.
5881 (dump_chain): Dump store-stores chain.
5882 (release_chain): Release resources.
5883 (split_data_refs_to_components): Compute and create component
5884 contains only stores for elimination.
5885 (get_chain_last_ref_at): New function.
5886 (make_invariant_chain): Initialization.
5887 (make_rooted_chain): Specify chain type in parameter and record it.
5888 (add_looparound_copies): Skip for store-stores chain.
5889 (determine_roots_comp): Compute type of chain and pass it to
5891 (initialize_root_vars_store_elim_2): New function.
5892 (finalize_eliminated_stores): New function.
5893 (remove_stmt): Handle store for elimination.
5894 (execute_pred_commoning_chain): Execute predictive commoning on
5896 (determine_unroll_factor): Skip unroll for store-stores chain.
5897 (prepare_initializers_chain_store_elim): New function.
5898 (prepare_initializers_chain): Hanlde store-store chain.
5899 (prepare_finalizers_chain, prepare_finalizers): New function.
5900 (tree_predictive_commoning_loop): Return integer value indicating
5901 if loop is unrolled or lcssa form is corrupted.
5902 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
5904 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5906 * tree-predcom.c (initialize_root): Delete.
5907 (execute_pred_commoning_chain): Initialize root vars and replace
5908 reference of non-combined chain directly, rather than call above
5911 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5913 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
5914 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
5916 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5918 * tree-predcom.c (struct chain): New field init_seq.
5919 (release_chain): Release init_seq.
5920 (prepare_initializers_chain): Record intialization stmts in above
5922 (insert_init_seqs): New function.
5923 (tree_predictive_commoning_loop): Call insert_init_seqs.
5925 2017-07-28 Bin Cheng <bin.cheng@arm.com>
5927 * tree-predcom.c (determine_roots_comp): Skip trivial components.
5929 2017-07-28 Richard Biener <rguenther@suse.de>
5931 * match.pd: Remove superfluous :c.
5932 * genmatch.c (simplify::id): Add member.
5933 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
5935 (current_id): New global.
5936 (dt_node::parent): Move from ...
5937 (dt_operand::parent): ... here. Add for_id member.
5938 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
5939 (decision_tree::find_node): Relax order requirement when
5940 merging DT_TRUE nodes to ones inbetween the current simplify
5941 and the one we try to merge with. Add diagnostic whenever
5942 we need to enforce pattern order by not merging.
5943 (decision_tree::insert): Set current_id.
5944 (decision_tree::print_node): Dump parent node and for_id.
5945 (parser::last_id): Add member.
5946 (parser::push_simplify): Assign unique id.
5947 (parser::parser): Initialize last_id.
5949 2017-07-28 Martin Liska <mliska@suse.cz>
5952 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
5953 gimple_build_debug_bind.
5955 2017-07-28 Richard Biener <rguenther@suse.de>
5957 PR tree-optimization/81502
5958 * match.pd: Add pattern combining BIT_INSERT_EXPR with
5960 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
5962 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
5963 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
5964 for BIT_FIELD_REF args.
5965 * fold-const.c (make_bit_field_ref): Likewise.
5966 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
5968 2017-07-28 Jakub Jelinek <jakub@redhat.com>
5971 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
5972 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
5973 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
5974 Or it into SANITIZER_UNDEFINED.
5975 * ubsan.c: Include gimple-fold.h and varasm.h.
5976 (ubsan_expand_ptr_ifn): New function.
5977 (instrument_pointer_overflow): New function.
5978 (maybe_instrument_pointer_overflow): New function.
5979 (instrument_object_size): Formatting fix.
5980 (pass_ubsan::execute): Call instrument_pointer_overflow
5981 and maybe_instrument_pointer_overflow.
5982 * internal-fn.c (expand_UBSAN_PTR): New function.
5983 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
5984 * sanitizer.def (__ubsan_handle_pointer_overflow,
5985 __ubsan_handle_pointer_overflow_abort): New builtins.
5986 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
5987 * internal-fn.def (UBSAN_PTR): New internal function.
5988 * opts.c (sanitizer_opts): Add pointer-overflow.
5989 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
5990 * fold-const.c (build_range_check): Compute pointer range check in
5991 integral type if pointer arithmetics would be needed. Formatting
5994 2017-07-28 Martin Liska <mliska@suse.cz>
5997 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
5998 parameters that are of a variable-length.
6000 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
6002 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
6004 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
6005 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
6006 (CRT_CALL_STATIC_FUNCTION): Likewise.
6007 (ASM_DEFAULT_SPEC): New define.
6008 (ASM_SPEC32): Likewise.
6009 (ASM_SPEC64): Likewise.
6010 (ASM_SPEC_COMMON): Likewise.
6011 (ASM_SPEC): Likewise.
6012 (INVALID_64BIT): Likewise.
6013 (LINK_OS_DEFAULT_SPEC): Likewise.
6014 (LINK_OS_SPEC32): Likewise.
6015 (LINK_OS_SPEC64): Likewise.
6016 (POWERPC_LINUX): Likewise.
6017 (PTRDIFF_TYPE): Likewise.
6018 (RESTORE_FP_PREFIX): Likewise.
6019 (RESTORE_FP_SUFFIX): Likewise.
6020 (SAVE_FP_PREFIX): Likewise.
6021 (SAVE_FP_SUFFIX): Likewise.
6022 (SIZE_TYPE): Likewise.
6023 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
6024 (TARGET_64BIT): Likewise.
6025 (TARGET_64BIT): Likewise.
6026 (TARGET_AIX): Likewise.
6027 (WCHAR_TYPE_SIZE): Likewise.
6028 (WCHAR_TYPE): Undefine.
6029 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
6030 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
6031 (CPP_OS_RTEMS_SPEC): Delete.
6032 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
6033 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
6035 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
6037 2017-07-28 Jakub Jelinek <jakub@redhat.com>
6039 PR tree-optimization/81578
6040 * tree-parloops.c (build_new_reduction): Bail out if
6041 reduction_code isn't one of the standard OpenMP reductions.
6042 Move the details printing after that decision.
6044 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6046 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
6047 related to reload_in_progress.
6048 (splat_input_operand): Likewise.
6049 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
6051 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
6053 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
6054 (TARGET_INSTANTIATE_DECLS): Likewise.
6055 (legitimate_indexed_address_p): Delete reload_in_progress code.
6056 (rs6000_debug_legitimate_address_p): Likewise.
6057 (rs6000_eliminate_indexed_memrefs): Likewise.
6058 (rs6000_emit_le_vsx_store): Likewise.
6059 (rs6000_emit_move_si_sf_subreg): Likewise.
6060 (rs6000_emit_move): Likewise.
6061 (register_to_reg_type): Likewise.
6062 (rs6000_pre_atomic_barrier): Likewise.
6063 (rs6000_machopic_legitimize_pic_address): Likewise.
6064 (rs6000_allocate_stack_temp): Likewise.
6065 (rs6000_address_for_fpconvert): Likewise.
6066 (rs6000_address_for_altivec): Likewise.
6067 (rs6000_secondary_memory_needed_rtx): Delete function.
6068 (rs6000_check_sdmode): Likewise.
6069 (rs6000_alloc_sdmode_stack_slot): Likewise.
6070 (rs6000_instantiate_decls): Likewise.
6071 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
6072 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
6073 Delete reload_in_progress.
6074 (*vec_reload_and_plus_<mptrsize>): Likewise.
6075 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
6076 (vsx_div_v2di): Likewise.
6077 (vsx_udiv_v2di): Likewise.
6079 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6081 * config/rs6000/rs6000.opt (mlra): Replace with stub.
6082 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
6083 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
6084 (rs6000_debug_reg_global): Delete print of LRA status.
6085 (rs6000_option_override_internal): Delete dead LRA related code.
6086 (rs6000_lra_p): Delete function.
6087 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
6089 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
6091 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
6092 * config/riscv/rtems.h: New file.
6094 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6095 Sudakshina Das <sudi.das@arm.com>
6097 * config/aarch64/aarch64.md
6098 (define_split for and<mode>3nr_compare): Move
6099 non aarch64_logical_operand to a register.
6100 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
6101 register immediate operand to a register.
6102 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
6104 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
6107 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
6109 2017-07-27 Richard Biener <rguenther@suse.de>
6111 PR tree-optimization/81573
6112 PR tree-optimization/81494
6113 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
6114 multi defuse cycle case.
6116 2017-07-27 Richard Biener <rguenther@suse.de>
6118 PR tree-optimization/81571
6119 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
6122 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
6124 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
6125 earlier and only if MASK_FPU is set. Adjust formatting.
6127 2017-07-27 Martin Liska <mliska@suse.cz>
6129 * opt-functions.awk: Add validation of value of Init.
6130 * optc-gen.awk: Pass new argument.
6132 2017-07-27 Martin Liska <mliska@suse.cz>
6134 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
6135 Fix wrong condition.
6137 2017-07-27 Martin Liska <mliska@suse.cz>
6139 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
6140 BBs and edges seen by autoFDO.
6142 2017-07-27 Richard Biener <rguenther@suse.de>
6144 PR tree-optimization/81502
6145 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
6146 with incompatible but same sized type.
6147 (execute_update_addresses_taken): Likewise.
6149 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
6151 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
6152 flag_tree_loop_vectorize rather than flag_tree_vectorize.
6154 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6157 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
6158 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
6159 Change s_operand to memory_operand.
6161 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
6163 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
6164 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
6165 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
6166 Emit instructions rather than returning an expression. Handle TFmode
6167 and KFmode by casting to TImode.
6168 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
6169 (rs6000_emit_le_vsx_store): Likewise.
6170 * config/rs6000/vsx.md (VSX_TI): New iterator.
6171 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
6172 (*vsx_le_undo_permute_<mode>): Likewise.
6173 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
6174 emit the split sequence.
6175 (*vsx_le_perm_store_<mode>): Likewise.
6177 2017-07-27 Jakub Jelinek <jakub@redhat.com>
6179 PR tree-optimization/81555
6180 PR tree-optimization/81556
6181 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
6182 if true, force CHANGED for the recursive invocation.
6183 (reassociate_bb): Remember original length of ops array, pass
6184 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
6186 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
6187 attributes for noipa attribute. For naked attribute use
6188 lookup_attribute first before lookup_attribute_spec.
6189 * final.c (rest_of_handle_final): Disable IPA RA for functions with
6191 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
6192 for functions with noipa attribute.
6193 (cgraph_externally_visible_p): Return true for functions with noipa
6195 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
6196 for functions with noipa attribute.
6197 * doc/extend.texi: Document noipa function attribute.
6198 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
6199 also for functions with noipa attribute.
6200 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
6202 2017-07-26 Andrew Pinski <apinski@cavium.com>
6204 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
6205 vec_unalign_load_cost and vec_unalign_store_cost.
6207 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
6209 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
6210 -mvsx-small-integer option.
6211 (ISA_3_0_MASKS_IEEE): Likewise.
6212 (OTHER_VSX_VECTOR_MASKS): Likewise.
6213 (POWERPC_MASKS): Likewise.
6214 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
6215 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
6216 code, only testing for DImode being allowed in non-VSX floating
6218 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
6219 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
6221 (rs6000_option_override_internal): Delete -mvsx-small-integer.
6222 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
6223 TARGET_P8_VECTOR test.
6224 (rs6000_secondary_reload_simple_move): Likewise.
6225 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
6226 since TARGET_P9_VECTOR was already tested.
6227 (rs6000_opt_masks): Remove -mvsx-small-integer.
6228 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
6229 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6231 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
6232 test for TARGET_VEXTRACTUB was used, and that uses
6234 (p9 extract splitter): Likewise.
6235 (vsx_extract_<mode>_di_p9): Likewise.
6236 (vsx_extract_<mode>_store_p9): Likewise.
6237 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
6238 for TARGET_P9_VECTOR was used. Delete code that is now dead with
6239 the elimination of TARGET_VSX_SMALL_INTEGER.
6240 (vsx_extract_<mode>_p8): Likewise.
6241 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
6242 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
6243 (vsx_set_<mode>_p9): Likewise.
6244 (vsx_set_v4sf_p9): Likewise.
6245 (vsx_set_v4sf_p9_zero): Likewise.
6246 (vsx_insert_extract_v4sf_p9): Likewise.
6247 (vsx_insert_extract_v4sf_p9_2): Likewise.
6248 * config/rs6000/rs6000.md (sign extend splitter): Change
6249 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
6250 (floatsi<mode>2_lfiwax_mem): Likewise.
6251 (floatunssi<mode>2_lfiwzx_mem): Likewise.
6252 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
6253 since a test for TARGET_P9_VECTOR was used.
6254 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6255 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
6256 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6257 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6258 TARGET_P8_VECTOR test.
6259 (fix_trunc<mode>si2_stfiwx): Likewise.
6260 (fix_trunc<mode>si2_internal): Likewise.
6261 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
6262 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6264 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6265 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
6266 TARGET_P8_VECTOR test.
6267 (fixuns_trunc<mode>si2_stfiwx): Likewise.
6268 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
6269 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
6271 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
6272 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
6273 since a test for TARGET_P9_VECTOR was used.
6274 (splitter for loading small constants): Likewise.
6276 2017-07-26 Andrew Pinski <apinski@cavium.com>
6278 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
6281 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
6284 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
6285 (fp_valid_at): Likewise.
6287 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6289 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
6290 (qdf24xx_addrcost_table): Likewise.
6291 (cortexa57_tunings): Update to use generic_branch_cost.
6292 (cortexa72_tunings): Likewise.
6293 (cortexa73_tunings): Likewise.
6294 (qdf24xx_tunings): Likewise.
6296 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
6298 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
6299 (thunderx2t99_branch_cost): Likewise.
6300 (cortexa35_tunings): Update to use generic_branch_cost.
6301 (cortexa53_tunings): Likewise.
6302 (cortexa57_tunings): Likewise.
6303 (cortexa72_tunings): Likewise.
6304 (cortexa73_tunings): Likewise.
6305 (thunderx2t99_tunings): Likewise.
6307 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6309 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
6310 (sparc_option_override): Honour MASK_FSMULD.
6311 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
6312 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
6313 * config/sparc/sparc.opt (mfsmuld): New option.
6314 * doc/invoke.texi (mfsmuld): Document option.
6316 2017-07-26 Marek Polacek <polacek@redhat.com>
6319 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
6321 2017-07-26 Richard Biener <rguenther@suse.de>
6323 * gimple-match-head.c (do_valueize): Return OP if valueize
6325 (get_def): New helper to get at the def stmt of a SSA name
6327 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
6328 do_valueize to get at the def stmt.
6329 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
6331 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
6334 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
6336 2017-07-26 Martin Liska <mliska@suse.cz>
6339 * function.c (expand_function_start): Make expansion of
6340 nonlocal_goto_save_area after parm_birth_insn.
6342 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
6344 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
6345 from all CPU target flags enable members.
6347 2017-07-26 Richard Biener <rguenther@suse.de>
6349 * genmatch.c (dt_simplify::gen): Make iterator vars const.
6350 (decision_tree::gen): Make 'type' const.
6351 (write_predicate): Likewise.
6353 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6355 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
6356 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
6357 (rs6000_option_override_internal): Likewise.
6358 (rs6000_expand_vector_set): Likewise.
6359 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
6360 (TARGET_UPPER_REGS_SF): Likewise.
6361 (TARGET_UPPER_REGS_DI): Likewise.
6362 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
6363 (TARGET_DIRECT_MOVE_64BIT): Likewise.
6364 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
6365 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
6366 (Splitters for DI constants in Altivec registers): Likewise.
6367 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
6368 (vsx_set_v4sf_p9): Likewise.
6369 (vsx_set_v4sf_p9_zero): Likewise.
6370 (vsx_insert_extract_v4sf_p9): Likewise.
6371 (vsx_insert_extract_v4sf_p9_2): Likewise.
6373 2017-07-25 Carl Love <cel@us.ibm.com>
6375 * doc/extend.texi: Update the built-in documentation file for the
6376 existing built-in functions
6377 vector signed char vec_cnttz (vector signed char);
6378 vector unsigned char vec_cnttz (vector unsigned char);
6379 vector signed short vec_cnttz (vector signed short);
6380 vector unsigned short vec_cnttz (vector unsigned short);
6381 vector signed int vec_cnttz (vector signed int);
6382 vector unsigned int vec_cnttz (vector unsigned int);
6383 vector signed long long vec_cnttz (vector signed long long);
6384 vector unsigned long long vec_cnttz (vector unsigned long long);
6386 2017-07-25 Andrew Pinski <apinski@cavium.com>
6388 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
6389 accesses where the use is for the first operand of a BIT_INSERT.
6391 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6394 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6395 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6397 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
6399 * config/i386/gstabs.h: Delete.
6400 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
6402 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
6404 * config/i386/i386.c (ix86_decompose_address): Do not check for
6405 register RTX when looking at index_reg or base_reg.
6406 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
6408 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
6410 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
6411 to update EH info here.
6413 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6415 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
6417 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6419 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
6421 2017-07-25 Torsten Duwe <duwe@suse.de>
6423 * common.opt: Introduce -fpatchable-function-entry
6424 command line option, and its variables function_entry_patch_area_size
6425 and function_entry_patch_area_start.
6426 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
6427 including a two-value parser.
6428 * target.def (print_patchable_function_entry): New target hook.
6429 * targhooks.h (default_print_patchable_function_entry): New function.
6430 * targhooks.c (default_print_patchable_function_entry): Likewise.
6431 * toplev.c (process_options): Switch off IPA-RA if
6432 patchable function entries are being generated.
6433 * varasm.c (assemble_start_function): Look at the
6434 patchable-function-entry command line switch and current
6435 function attributes and maybe generate NOP instructions by
6436 calling the print_patchable_function_entry hook.
6437 * doc/extend.texi: Document patchable_function_entry attribute.
6438 * doc/invoke.texi: Document -fpatchable_function_entry
6439 command line option.
6440 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
6442 * doc/tm.texi: Re-generate.
6444 2017-07-25 Jakub Jelinek <jakub@redhat.com>
6447 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
6448 TARGET_AVX512DQ rather than TARGET_AVX512BW.
6450 2017-07-25 Tamar Christina <tamar.christina@arm.com>
6452 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
6454 2017-07-25 Richard Biener <rguenther@suse.de>
6456 PR tree-optimization/81455
6457 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
6458 not walk in cycles when looking for guards.
6460 2017-07-25 Richard Biener <rguenther@suse.de>
6462 PR tree-optimization/81529
6463 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
6464 when optimizing backedge uses.
6466 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
6468 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
6470 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
6471 to dl_section_ref. On AIX, append an expression to subtract
6472 the size of the section length to dl_section_ref.
6474 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
6476 * configure.ac: If any of the config.* scripts fail, exit 1.
6477 * configure: Regenerate.
6479 2017-07-25 Richard Biener <rguenther@suse.de>
6482 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
6483 of immediate uses, be more verbose on errors.
6485 2017-07-25 Richard Biener <rguenther@suse.de>
6487 PR tree-optimization/81510
6488 * tree-vect-loop.c (vect_is_simple_reduction): When the
6489 reduction stmt is not inside the loop bail out.
6491 2017-07-25 Richard Biener <rguenther@suse.de>
6493 PR tree-optimization/81303
6494 * tree-vect-loop-manip.c (vect_loop_versioning): Build
6495 profitability check against LOOP_VINFO_NITERSM1.
6497 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
6499 * domwalk.c (cmp_bb_postorder): Simplify.
6500 (sort_bbs_postorder): New function. Use it...
6501 (dom_walker::walk): ...here to optimize common cases.
6503 2017-07-25 Martin Liska <mliska@suse.cz>
6506 * ipa-visibility.c (function_and_variable_visibility): Make the
6507 redirection just on target that supports aliasing.
6508 Fix GNU coding style.
6510 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
6513 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
6515 * config/arm/rtems.h: Likewise.
6516 * config/bfin/rtems.h: Likewise.
6517 * config/i386/rtemself.h: Likewise.
6518 * config/lm32/rtems.h: Likewise.
6519 * config/m32c/rtems.h: Likewise.
6520 * config/m68k/rtemself.h: Likewise.
6521 * config/microblaze/rtems.h: Likewise.
6522 * config/mips/rtems.h: Likewise.
6523 * config/moxie/rtems.h: Likewise.
6524 * config/nios2/rtems.h: Likewise.
6525 * config/powerpcspe/rtems.h: Likewise.
6526 * config/rs6000/rtems.h: Likewise.
6527 * config/rtems.h: Likewise.
6528 * config/sh/rtems.h: Likewise.
6529 * config/sh/rtemself.h: Likewise.
6530 * config/sparc/rtemself.h: Likewise.
6532 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6535 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
6536 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
6537 * tree-ssa-structalias.c (alias_get_name): Same.
6539 2017-07-25 Bin Cheng <bin.cheng@arm.com>
6542 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
6543 instructions if no du chain is found.
6545 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
6547 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
6549 2017-07-25 Richard Biener <rguenther@suse.de>
6552 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
6555 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
6557 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
6559 (ISA_2_7_MASKS_SERVER): Likewise.
6560 (ISA_3_0_MASKS_IEEE): Likewise.
6561 (OTHER_P8_VECTOR_MASKS): Likewise.
6562 (OTHER_VSX_VECTOR_MASKS): Likewise.
6563 (POWERPC_MASKS): Likewise.
6564 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
6565 duplicate list of options.
6566 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
6567 explicit -mupper-regs options.
6568 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
6569 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
6570 alias for -mupper-regs-df.
6571 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
6572 (rs6000_init_hard_regno_mode_ok): Likewise.
6573 (rs6000_option_override_internal): Likewise.
6574 (rs6000_opt_masks): Likewise.
6575 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
6576 options in terms of whether -mvsx or -mpower8-vector was used.
6577 (TARGET_UPPER_REGS_DI): Likewise.
6578 (TARGET_UPPER_REGS_SF): Likewise.
6579 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
6580 -mupper-regs-* options.
6582 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6584 * passes.c (emergency_dump_function): Print some empty lines and a
6585 header before the RTL dump.
6587 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
6589 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
6591 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
6594 * config/aarch64/aarch64.c (aarch64_classify_symbol):
6595 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
6597 2017-07-24 Carl Love <cel@us.ibm.com>
6599 * config/rs6000/rs6000-c.c: Add support for built-in functions
6600 vector float vec_extract_fp32_from_shorth (vector unsigned short);
6601 vector float vec_extract_fp32_from_shortl (vector unsigned short);
6602 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
6603 vec_extract_fp_from_shortl): Add defines for the two builtins.
6604 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
6605 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
6607 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
6608 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
6609 * doc/extend.texi: Update the built-in documentation file for the
6610 new built-in function.
6612 2017-07-24 Jakub Jelinek <jakub@redhat.com>
6615 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
6617 * doc/generic.texi: Likewise.
6618 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
6619 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
6621 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
6623 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
6624 (aarch64_mls_elt_merge<mode>): Likewise.
6626 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
6628 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
6629 having __cxa_atexit.
6631 2017-07-23 Michael Collison <michael.collison@arm.com>
6633 * config/arm/arm.c (arm_option_override): Deprecate
6634 use of -mstructure-size-boundary.
6635 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
6636 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
6638 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6641 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
6642 Reduce cost estimate for direct moves.
6644 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
6647 * config/i386/i386.c (ix86_option_override_internal): Disable
6648 BMI, BMI2 and TBM instructions for -m16.
6650 2017-07-21 Carl Love <cel@us.ibm.com>
6652 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6653 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6654 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
6655 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
6656 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
6657 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
6658 VMULOSW): New enum "unspec" values.
6659 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
6660 altivec_vmulosw): New patterns.
6661 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
6662 VMULOSW): Add definitions.
6664 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
6666 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
6667 (qdf24xx): Likewise.
6668 * config/aarch64/aarch64-options-extensions.def (rdma); New.
6669 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
6670 (AARCH64_FL_V8_1): Renumber.
6671 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
6672 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
6673 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
6674 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
6675 rdma to feature modifiers list.
6677 2017-07-21 Yury Gribov <tetra2005@gmail.com>
6680 * ipa-visibility (function_and_variable_visibility): Convert
6681 recursive PLT call to direct call if appropriate.
6683 2017-07-21 Andrew Pinski <apinski@cavium.com>
6685 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
6686 operand 1 to see if the types precision matches.
6687 * fold-const.c (operand_equal_p): Likewise.
6689 2017-07-21 Richard Biener <rguenther@suse.de>
6691 PR tree-optimization/81303
6692 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
6693 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
6694 (vect_peeling_hash_get_lowest_cost): Adjust.
6695 (vect_enhance_data_refs_alignment): Likewise. Use
6696 vect_get_peeling_costs_all_drs to compute the penalty for no
6697 peeling to match up costs.
6699 2017-07-21 Richard Biener <rguenther@suse.de>
6701 PR tree-optimization/81500
6702 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
6703 we didn't identify a reduction path.
6705 2017-07-21 Tom de Vries <tom@codesourcery.com>
6706 Cesar Philippidis <cesar@codesourcery.com>
6708 PR gcov-profile/81442
6709 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
6712 2017-07-21 Tom de Vries <tom@codesourcery.com>
6715 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
6717 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
6718 nvptx_override_options_after_change.
6720 2017-07-21 Ulrich Drepper <drepper@redhat.com>
6722 * dwarf2out.c (output_file_names): Avoid double testing for
6725 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
6727 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
6729 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
6731 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
6733 (try_crossjump_to_edge): Do not punt on partitioned functions.
6735 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
6737 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
6738 Put all BBs reachable only via paths crossing cold region to cold
6740 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
6742 2016-07-21 Richard Biener <rguenther@suse.de>
6744 PR tree-optimization/81303
6745 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
6746 into account prologue and epilogue iterations when raising
6747 min_profitable_iters to sth at least covering one vector iteration.
6749 2017-07-21 Tamar Christina <tamar.christina@arm.com>
6751 * config/arm/arm.c (arm_test_cpu_arch_dat):
6754 2017-07-20 Nathan Sidwell <nathan@acm.org>
6756 Remove TYPE_METHODS.
6757 * tree.h (TYPE_METHODS): Delete.
6758 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
6759 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
6760 (dbxout_type_methods): Scan TYPE_FIELDS.
6761 (dbxout_type): Don't check TYPE_METHODS here.
6762 * function.c (use_register_for_decl): Always ignore register for
6763 class types when not optimizing.
6764 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
6765 * tree.c (free_lang_data_in_type): Stitch out member functions and
6766 templates from TYPE_FIELDS.
6767 (build_distinct_type_copy, verify_type_variant,
6768 verify_type): Member fns are on TYPE_FIELDS.
6769 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
6770 * tree-pretty-print.c (dump_generic_node): Likewise.
6772 2017-07-20 Jakub Jelinek <jakub@redhat.com>
6775 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
6776 V2TImode and V4TImode.
6777 (ix86_expand_vector_extract): Likewise.
6778 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
6779 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
6780 (ssescalarmode): Handle V4TImode and V2TImode.
6781 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
6782 (*vec_extractv2ti, *vec_extractv4ti): New insns.
6783 (VEXTRACTI128_MODE): New mode iterator.
6784 (splitter for *vec_extractv?ti first element): New.
6785 (VEC_INIT_MODE): New mode iterator.
6786 (vec_init<mode>): Consolidate 3 expanders into one using
6787 VEC_INIT_MODE mode iterator.
6789 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
6791 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
6792 non_spilled_static_chain_regno_p.
6794 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
6796 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
6798 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
6800 * bb-reorder.c (connect_traces): Allow copying of blocks within
6803 2017-07-20 Richard Biener <rguenther@suse.de>
6805 * gimple.h (gimple_phi_result): Add gphi * overload.
6806 (gimple_phi_result_ptr): Likewise.
6807 (gimple_phi_arg): Likewise. Adjust index assert to only
6808 allow actual argument accesses rather than all slots available
6810 (gimple_phi_arg_def): Add gphi * overload.
6811 * tree-phinodes.c (make_phi_node): Initialize only actual
6813 (resize_phi_node): Clear memory not covered by old node,
6814 do not initialize excess argument slots.
6815 (reserve_phi_args_for_new_edge): Initialize new argument slot
6818 2017-07-20 Bin Cheng <bin.cheng@arm.com>
6820 PR tree-optimization/81388
6822 2016-07-21 Bin Cheng <bin.cheng@arm.com>
6824 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
6825 by removing computation of may_be_zero.
6827 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
6828 Tom de Vries <tom@codesourcery.com>
6831 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
6832 when gimple level profile disagrees with what RTL expander did.
6834 2017-07-20 Richard Biener <rguenther@suse.de>
6836 PR tree-optimization/61171
6837 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
6838 (vect_analyze_stmt): Add slp instance parameter.
6839 (vectorizable_reduction): Likewise.
6840 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
6841 (vect_is_simple_reduction): Deal with chains not detected
6842 as SLP reduction chain, specifically not properly associated
6843 chains containing a mix of plus/minus.
6844 (get_reduction_op): Remove.
6845 (get_initial_defs_for_reduction): Simplify, pass in whether
6846 this is a reduction chain, pass in the SLP node for the PHIs.
6847 (vect_create_epilog_for_reduction): Get the SLP instance as
6849 (vectorizable_reduction): Get the SLP instance as arg.
6850 During analysis remember the SLP node with the PHIs in the
6851 instance. Simplify getting at the vectorized reduction PHIs.
6852 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
6853 through SLP instance.
6854 (vect_slp_analyze_operations): Likewise.
6855 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
6856 (vect_transform_stmt): Likewise.
6858 2017-07-20 Tom de Vries <tom@codesourcery.com>
6860 PR tree-optimization/81489
6861 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
6862 read of phi arg location to before loop that modifies phi.
6864 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
6866 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
6869 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6872 * except.c (execute): Fix ordering issue.
6874 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
6876 PR rtl-optimization/81423
6877 * combine.c (make_compound_operation_int): Don't try to optimize
6878 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
6880 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
6882 PR rtl-optimization/81423
6883 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
6884 with a constant that is -1 in the truncated to mode.
6886 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6888 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
6889 (determine_unlikely_bbs): ... here.
6890 * predict.h (propagate_unlikely_bbs_forward): Declare.
6891 * cfgexpand.c (pass_expand::execute): Use it.
6892 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
6894 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
6895 propagate_unlikely_bbs_forward.
6897 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
6900 * except.c (maybe_add_nop_after_section_switch): New function.
6903 2017-07-19 Tom de Vries <tom@codesourcery.com>
6905 * gimple.h (gimple_phi_set_arg): Make assert more strict.
6907 2017-07-19 Tom de Vries <tom@codesourcery.com>
6909 * gimple.h (gimple_phi_arg): Make assert more strict.
6911 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
6913 * config.gcc (powerpc*-*-*): Add mmintrin.h.
6914 * config/rs6000/mmintrin.h: New file.
6915 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
6917 2017-07-19 Jakub Jelinek <jakub@redhat.com>
6919 PR tree-optimization/81346
6920 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
6922 2017-07-19 Tom de Vries <tom@codesourcery.com>
6924 * config/nvptx/nvptx.md (VECIM): Add V2DI.
6926 2017-07-19 Tom de Vries <tom@codesourcery.com>
6928 * config/nvptx/nvptx-modes.def: Add V2DImode.
6929 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
6930 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
6931 (nvptx_output_mov_insn): Handle lack of mov.b128.
6932 (nvptx_print_operand): Handle 'H' and 'L' codes.
6933 (nvptx_vector_mode_supported): Allow V2DImode.
6934 (nvptx_preferred_simd_mode): New function.
6935 (nvptx_data_alignment): New function.
6936 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
6937 nvptx_preferred_simd_mode.
6938 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
6940 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
6942 2017-07-19 Tom de Vries <tom@codesourcery.com>
6944 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
6945 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
6946 (nvptx_vector_mode_supported): New function. Allow V2SImode.
6947 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
6948 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
6949 (mov<VECIM>_insn): New define_insn.
6950 (define_expand "mov<VECIM>): New define_expand.
6952 2017-07-19 Tom de Vries <tom@codesourcery.com>
6954 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
6956 2017-07-19 Jakub Jelinek <jakub@redhat.com>
6958 PR tree-optimization/81346
6959 * fold-const.h (fold_div_compare, range_check_type): Declare.
6960 * fold-const.c (range_check_type): New function.
6961 (build_range_check): Use range_check_type.
6962 (fold_div_compare): No longer static, rewritten into
6963 a match.pd helper function.
6964 (fold_comparison): Don't call fold_div_compare here.
6965 * match.pd (X / C1 op C2): New optimization using fold_div_compare
6968 2017-07-19 Nathan Sidwell <nathan@acm.org>
6970 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
6971 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
6972 * tree.c (find_decls_types_r, verify_type): Use
6973 TYPE_{MIN,MAX}_VALUE_RAW.
6974 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
6975 (hash_tree): Likewise.
6976 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
6978 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
6981 2017-07-18 Tom de Vries <tom@codesourcery.com>
6984 * omp-expand.c (expand_omp_for_static_chunk): Handle
6985 equal-argument loop exit phi.
6987 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
6990 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
6991 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
6992 operand 2 predicate.
6993 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
6994 operand 2 predicate.
6995 (ror,rol -> rorx splitters): Use const_int_operand as
6996 operand 2 predicate.
6998 2017-06-18 Richard Biener <rguenther@suse.de>
7000 PR tree-optimization/81410
7001 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
7002 the gap in the ! slp_perm SLP case after each group.
7004 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7007 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
7010 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
7013 * predict.c (set_even_probabilities): Cleanup; do not affect
7014 probabilities that are already known.
7015 (combine_predictions_for_bb): Call even when count is set.
7017 2017-07-18 Nathan Sidwell <nathan@acm.org>
7019 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
7022 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7025 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
7026 optimization for loop niter analysis.
7028 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
7031 * config/avr/avr.c (avr_optimize_casesi): Don't use
7032 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
7034 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
7036 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
7037 body_cost_vec from _vect_peel_extended_info.
7038 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
7039 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
7042 2017-07-18 Bin Cheng <bin.cheng@arm.com>
7044 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
7046 2017-07-18 Richard Biener <rguenther@suse.de>
7048 PR tree-optimization/80620
7049 PR tree-optimization/81403
7050 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
7051 info when re-using a VN table entry.
7053 2017-07-18 Richard Biener <rguenther@suse.de>
7055 PR tree-optimization/81418
7056 * tree-vect-loop.c (vectorizable_reduction): Properly compute
7057 vectype_in. Verify that with lane-reducing reduction operations
7058 we have a single def-use cycle.
7060 2017-07-17 Carl Love <cel@us.ibm.com>
7062 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
7064 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7065 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7066 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7067 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7068 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7069 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
7070 VMULOSW): New enum "unspec" values.
7071 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
7072 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
7073 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
7074 altivec_vmulosw): New patterns.
7075 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7076 VMULOSW): Add definitions.
7077 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
7079 * config/alpha/alpha.c: Include predict.h.
7081 2017-07-17 Yury Gribov <tetra2005@gmail.com>
7083 * tree-vrp.c (compare_assert_loc): Fix comparison function
7084 to return predictable results.
7086 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7088 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
7091 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
7092 * doc/invoke.texi (mexpand-adddi): Update text.
7094 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7096 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
7097 that also clobbers the CC register. The old expand code is moved
7099 (*arc_clzsi2): ... here.
7100 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
7101 the CC register. The old expand code is moved to ...
7102 (arc_ctzsi2): ... here.
7104 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7106 * config/arc/arc.opt (mindexed-loads): Use initial value
7107 TARGET_INDEXED_LOADS_DEFAULT.
7108 (mauto-modify-reg): Use initial value
7109 TARGET_AUTO_MODIFY_REG_DEFAULT.
7110 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7111 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7112 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
7113 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
7115 2017-07-17 Martin Liska <mliska@suse.cz>
7118 * opts.c (finish_options): Do not allow -fgnu-tm
7119 w/ -fsanitize={kernel-,}address. Say sorry.
7121 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7124 * tree-loop-distribution.c (classify_partition): Only assert on
7125 numer of iterations.
7126 (merge_dep_scc_partitions): Delete prameter. Update function call.
7127 (distribute_loop): Remove code handling loop with unknown niters.
7128 (pass_loop_distribution::execute): Skip loop with unknown niters.
7130 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7133 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
7134 function sort_partitions_by_post_order.
7136 2017-07-17 Bin Cheng <bin.cheng@arm.com>
7138 PR tree-optimization/81374
7139 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
7140 the max index of basic blocks, rather than number of basic blocks.
7142 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7144 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
7146 (arc_legitimate_pic_operand_p): Likewise.
7147 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
7149 (arc_needs_pcl_p): Likewise.
7150 (arc_legitimate_pc_offset_p): Likewise.
7151 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
7152 function is also used in constrains.md.
7153 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
7154 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
7155 PLUS. Only return true/false in known cases, otherwise assert.
7156 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
7157 is already called in arc_legitimate_constant_p.
7158 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
7160 (LEGITIMATE_PIC_OPERAND_P): Use
7161 arc_raw_symbolic_reference_mentioned_p function.
7162 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
7167 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
7168 Andrew Burgess <andrew.burgess@embecosm.com>
7170 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
7171 (arc_return_address_register): New function.
7172 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
7173 (arc_handle_fndecl_attribute): Add naked attribute.
7174 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
7175 (TARGET_WARN_FUNC_RETURN): Likewise.
7176 (arc_allocate_stack_slots_for_args): New function.
7177 (arc_warn_func_return): Likewise.
7178 (machine_function): Change type fn_type.
7179 (arc_compute_function_type): Consider new naked function type,
7180 change function return type.
7181 (arc_must_save_register): Adapt to handle new
7182 arc_compute_function_type's return type.
7183 (arc_expand_prologue): Likewise.
7184 (arc_expand_epilogue): Likewise.
7185 (arc_return_address_regs): Delete.
7186 (arc_return_address_register): New function.
7187 (arc_epilogue_uses): Use above function.
7188 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
7189 (arc_function_type): Change encoding, add naked type.
7190 (ARC_INTERRUPT_P): Change to handle the new encoding.
7191 (ARC_FAST_INTERRUPT_P): Likewise.
7192 (ARC_NORMAL_P): Define.
7193 (ARC_NAKED_P): Likewise.
7194 (arc_compute_function_type): Delete prototype.
7195 * config/arc/arc.md (in_ret_delay_slot): Use
7196 arc_return_address_register function.
7197 (simple_return): Likewise.
7198 (p_return_i): Likewise.
7200 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7202 PR tree-optimization/81428
7203 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
7204 can't be built for those types.
7206 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7208 Remove stuff dead since r239246.
7210 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
7211 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
7212 (avr_inform_devices): Remove dead stuff.
7214 2017-07-17 Tamar Christina <tamar.christina@arm.com>
7216 * config/arm/arm_neon.h: Fix softp typo.
7218 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7220 PR tree-optimization/81365
7221 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
7222 aggregate moves onto bb predecessor edges, make sure there are no
7223 loads that could alias the lhs in between the start of bb and the
7226 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
7229 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
7230 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
7231 [LSHIFTRT, outer_code = TRUNCATE]: Same.
7233 2017-07-17 Jakub Jelinek <jakub@redhat.com>
7235 PR tree-optimization/81396
7236 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
7237 (init_symbolic_number): Initialize it to 1.
7238 (perform_symbolic_merge): Add n_ops from both operands into the new
7240 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
7241 without base_addr as useless if they need more than one operation.
7242 (bswap_replace): Handle !bswap case for NULL base_addr.
7244 2017-07-17 Tom de Vries <tom@codesourcery.com>
7247 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
7250 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
7252 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
7253 conditional builtin define __FIX_LEON3FT_B2BST.
7255 2017-07-17 Daniel Cederman <cederman@gaisler.com>
7257 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
7258 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
7261 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7263 PR rtl-optimization/81424
7264 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
7265 to remove potential trapping from operands if -fnon-call-exceptions.
7267 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7269 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
7270 profile_proability for scalling.
7271 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
7273 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7275 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
7277 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7279 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
7280 fixpoint arithmetics.
7282 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7284 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
7285 fixpoint arithmetics.
7287 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7289 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
7290 fixpoint arithmetics.
7292 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
7294 * profile-count.h (profile_probability::from_reg_br_prob_note,
7295 profile_probability::to_reg_br_prob_note): New functions.
7296 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
7297 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
7298 * predict.c (probability_reliable_p): Update.
7299 (edge_probability_reliable_p): Update.
7300 (br_prob_note_reliable_p): Update.
7301 (invert_br_probabilities): Update.
7302 (add_reg_br_prob_note): New function.
7303 (combine_predictions_for_insn): Update.
7304 * asan.c (asan_clear_shadow): Update.
7305 * cfgbuild.c (compute_outgoing_frequencies): Update.
7306 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
7307 (update_br_prob_note): Update.
7308 (rtl_verify_edges): Update.
7309 (purge_dead_edges): Update.
7310 (fixup_reorder_chain): Update.
7311 * emit-rtl.c (try_split): Update.
7312 * ifcvt.c (cond_exec_process_insns): Update.
7313 (cond_exec_process_if_block): Update.
7314 (dead_or_predicable): Update.
7315 * internal-fn.c (expand_addsub_overflow): Update.
7316 (expand_neg_overflow): Update.
7317 (expand_mul_overflow): Update.
7318 * loop-doloop.c (doloop_modify): Update.
7319 * loop-unroll.c (compare_and_jump_seq): Update.
7320 * optabs.c (emit_cmp_and_jump_insn_1): Update.
7321 * predict.h: Update.
7322 * reorg.c (mostly_true_jump): Update.
7324 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
7325 * config/alpha/alpha.c (emit_unlikely_jump): Update.
7326 * config/arc/arc.c: (emit_unlikely_jump): Update.
7327 * config/arm/arm.c: (emit_unlikely_jump): Update.
7328 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
7329 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
7330 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
7331 (ix86_print_operand): Update.
7332 (ix86_split_fp_branch): Update.
7333 (predict_jump): Update.
7334 * config/ia64/ia64.c (ia64_print_operand): Update.
7335 * config/mmix/mmix.c (mmix_print_operand): Update.
7336 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
7337 (rs6000_expand_split_stack_prologue): Update.
7338 * config/rs6000/rs6000.c: Update.
7339 * config/s390/s390.c (s390_expand_vec_strlen): Update.
7340 (s390_expand_vec_movstr): Update.
7341 (s390_expand_cs_tdsi): Update.
7342 (s390_expand_split_stack_prologue): Update.
7343 * config/sh/sh.c (sh_print_operand): Update.
7344 (expand_cbranchsi4): Update.
7345 (expand_cbranchdi4): Update.
7346 * config/sparc/sparc.c (output_v9branch): Update.
7347 * config/spu/spu.c (get_branch_target): Update.
7348 (ea_load_store_inline): Update.
7349 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
7350 * config/tilepro/tilepro.c: Update.
7352 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
7354 * gimplify.c (mostly_copy_tree_r): Revert latest change.
7355 (gimplify_save_expr): Likewise.
7357 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7359 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
7361 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7363 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
7365 * timevar.def (TV_IPA_FNSUMMARY): Define.
7367 2017-07-16 Daniel Cederman <cederman@gaisler.com>
7369 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
7370 to back store errata sensitive sequence from being generated.
7371 (sqrtdf2_fix): Likewise.
7373 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7375 * tree-ssa-threadupdate.c (compute_path_counts,
7376 update_joiner_offpath_counts): Use profile_probability.
7378 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
7381 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7383 * config/arm/arm-c.c (arm_cpu_builtins): Define
7384 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7386 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
7388 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7389 array entries to represent __ieee128 versions of the
7390 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
7391 scalar_extract_sig, and scalar_insert_exp built-in functions.
7392 (altivec_resolve_overloaded_builtin): Add special case handling
7393 for the __builtin_scalar_insert_exp function, as represented by
7394 the P9V_BUILTIN_VEC_VSIEDP constant.
7395 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
7396 exponent support for __ieee128 argument.
7397 (VSESQP): Add scalar extract signature support for __ieee128
7399 (VSTDCNQP): Add scalar test negative support for __ieee128
7401 (VSIEQP): Add scalar insert exponent support for __int128 argument
7402 with __ieee128 result.
7403 (VSIEQPF): Add scalar insert exponent support for __ieee128
7404 argument with __ieee128 result.
7405 (VSTDCQP): Add scalar test data class support for __ieee128
7407 (VSTDCNQP): Add overload support for scalar test negative with
7409 (VSTDCQP): Add overload support for scalar test data class
7411 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
7413 (UNSPEC_VSX_SIEXPQP): New constant.
7414 (xsxexpqp): New insn for VSX scalar extract exponent quad
7416 (xsxsigqp): New insn for VSX scalar extract significand quad
7418 (xsiexpqpf): New insn for VSX scalar insert exponent quad
7419 precision with floating point argument.
7420 (xststdcqp): New expand for VSX scalar test data class quad
7422 (xststdcnegqp): New expand for VSX scalar test negative quad
7424 (xststdcqp): New insn to match expansions for VSX scalar test data
7425 class quad precision and VSX scalar test negative quad precision.
7426 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
7427 special case operand checking to enforce that second operand of
7428 VSX scalar test data class with quad precision argument is a 7-bit
7430 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
7431 prototypes and descriptions of __ieee128 versions of
7432 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
7433 scalar_test_data_class, and scalar_test_neg built-in functions.
7435 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7437 PR tree-optimization/81162
7438 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
7439 replace a negate with an add.
7441 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
7443 * doc/invoke.texi (arm/-mcpu): Document +crypto.
7445 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7447 * config/arm/arm-c.c (arm_cpu_builtins): Define
7448 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
7450 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7452 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
7453 (armv8-r): Set ARM Cortex-R52 as default CPU.
7454 * config/arm/arm-tables.opt: Regenerate.
7455 * config/arm/arm-tune.md: Regenerate.
7456 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7458 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
7459 extension for -mcpu=cortex-r52.
7461 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
7463 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
7464 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
7465 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
7466 (fp-armv8): Define it as FP_ARMv8 only.
7467 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
7468 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
7470 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
7471 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
7472 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
7473 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
7474 than TARGET_FPU_ARMV8.
7475 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
7476 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
7477 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
7479 * config/arm/neon.md (neon_vrint): Likewise.
7480 (neon_vcvt): Likewise.
7481 (neon_<fmaxmin_op><mode>): Likewise.
7482 (<fmaxmin><mode>3): Likewise.
7483 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
7484 * config/arm/predicates.md (arm_cond_move_operator): Check against
7485 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
7487 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
7489 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
7492 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7494 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
7495 loop in comment with memset.
7497 2017-07-14 Martin Liska <mliska@suse.cz>
7499 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
7500 * dwarf2out.c (is_java): Remove the function.
7501 (output_pubname): Remove usage of the function.
7502 (lower_bound_default): Remove usage of DW_LANG_Java.
7503 (gen_compile_unit_die): Likewise.
7504 * gcc.c: Remove compiler defaults for .java and .zip files.
7505 * gimple-expr.c (remove_suffix): Change as there's no longer
7506 extension than 4-letter one.
7507 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
7508 (gimplify_save_expr): Likewise.
7509 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
7510 as it's possible even for other languages than Java.
7511 * langhooks.h (struct lang_hooks): Remove Java from a comment.
7512 * lto-opts.c (lto_write_options): Remove reference to Java.
7513 * opts.c (strip_off_ending): Update file extension handling.
7514 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
7515 * tree-eh.c (lower_resx): Likewise.
7516 * tree.c (free_lang_data_in_type): Remove dead code.
7517 (find_decls_types_r): Likewise.
7518 (build_common_builtin_nodes): Remove Java from a comment.
7519 (verify_type): Remove dead code.
7520 * varasm.c (assemble_external): Remove Java from a comment.
7522 2017-07-14 Martin Liska <mliska@suse.cz>
7524 * opts.c (finish_options): Add quotes.
7525 (common_handle_option): Likewise.
7527 2017-07-14 Martin Liska <mliska@suse.cz>
7529 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
7530 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
7532 * dwarf2out.c (lower_bound_default): Do not handle
7534 (gen_compile_unit_die): Likewise.
7535 * gcc.c: Remove default extension binding for GNU Pascal.
7536 * stmt.c: Remove Pascal language from a comment.
7537 * xcoffout.c: Likewise.
7539 2017-07-13 David Malcolm <dmalcolm@redhat.com>
7542 * diagnostic-show-locus.c (fixit_cmp): New function.
7543 (layout::layout): Sort m_fixit_hints.
7544 (column_range::column_range): Assert that the values are valid.
7545 (struct char_span): New struct.
7546 (correction::overwrite): New method.
7547 (struct source_line): New struct.
7548 (line_corrections::add_hint): Add assertions. Reimplement memcpy
7549 calls in terms of classes source_line and char_span, and
7550 correction::overwrite.
7551 (selftest::test_overlapped_fixit_printing_2): New function.
7552 (selftest::diagnostic_show_locus_c_tests): Call it.
7554 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
7556 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
7557 early if there is no lhs.
7559 2017-07-13 Martin Liska <mliska@suse.cz>
7561 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
7562 (gen_reference_type_die): Likewise.
7563 * stor-layout.c: Remove Pascal-related comment.
7565 2017-07-13 Martin Liska <mliska@suse.cz>
7567 * opts.c (finish_options): Add quotes to error messages.
7568 (parse_sanitizer_options): Likewise.
7570 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
7572 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
7574 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
7576 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
7578 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
7580 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
7582 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
7584 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
7587 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
7588 provides the hardware capability bits, define the macro
7589 __BUILTIN_CPU_SUPPORTS__.
7590 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
7591 if GLIBC does not provide the hardware capability bits. Add a
7592 gcc_unreachable call if the built-in cpu function is neither
7593 __builtin_cpu_is nor __builtin_cpu_supports.
7594 (rs6000_get_function_versions_dispatcher): Change the warning
7595 that an old GLIBC is used which does not export the capability
7596 bits to be an error.
7597 * doc/extend.texi (target_clones attribute): Document the
7598 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
7599 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
7600 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
7601 the macros defined by GCC if the newer GLIBC is available.
7603 2017-07-12 Jeff Law <law@redhat.com>
7605 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
7606 remaining includes slightly.
7607 * config/riscv/riscv-builtins.c: Include profile-count.h.
7609 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
7612 * config/avr/avr.c (avr_set_current_function): In diagnostic
7613 messages: Quote keywords and (parts of) identifiers.
7614 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
7617 2017-07-12 Carl Love <cel@us.ibm.com>
7619 * config/rs6000/rs6000-c.c: Add support for built-in functions
7620 vector bool char vec_revb (vector bool char);
7621 vector bool short vec_revb (vector short char);
7622 vector bool int vec_revb (vector bool int);
7623 vector bool long long vec_revb (vector bool long long);
7624 * doc/extend.texi: Update the built-in documentation file for the
7625 new built-in functions.
7627 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7629 * config/s390/s390.md: Remove movcc splitter.
7631 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7633 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
7634 load/store on condition.
7636 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
7639 * config/avr/avr.c (avr_encode_section_info)
7640 [progmem && !TREE_READONLY]: Error if progmem object needs
7643 2017-07-11 Michael Collison <michael.collison@arm.com>
7645 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
7648 2017-07-11 Carl Love <cel@us.ibm.com>
7650 * config/rs6000/rs6000-c.c: Add support for builtins
7651 vector unsigned int vec_parity_lsbb (vector signed int);
7652 vector unsigned int vec_parity_lsbb (vector unsigned int);
7653 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
7654 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
7655 vector unsigned long long vec_parity_lsbb (vector signed long long);
7656 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
7657 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
7658 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
7659 * doc/extend.texi: Update the built-in documentation file for the
7660 new built-in functions.
7662 2017-07-11 David Malcolm <dmalcolm@redhat.com>
7664 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
7665 (layout::m_primary_loc): New field.
7666 (layout::layout): Initialize new field. Move location filtering
7667 logic from here to...
7668 (layout::maybe_add_location_range): ...this new method. Add
7669 support for filtering to just the lines already specified by other
7671 (layout::will_show_line_p): New method.
7672 (gcc_rich_location::add_location_if_nearby): New method.
7673 (selftest::test_add_location_if_nearby): New test function.
7674 (selftest::diagnostic_show_locus_c_tests): Call it.
7675 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
7678 2017-07-11 Tom de Vries <tom@codesourcery.com>
7680 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
7681 (bb_first_real_insn): New function.
7682 (nvptx_single): Add extra initialization of broadcasted condition
7685 2017-07-11 Nathan Sidwell <nathan@acm.org>
7687 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
7689 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
7691 * doc/extend.texi (AVR Function Attributes): Remove weblink to
7692 Binutils doc as TEXI will mess them up.
7693 * doc/invoke.texi (AVR Options): Same here.
7695 2017-07-11 Daniel Cederman <cederman@gaisler.com>
7697 * config/sparc/sparc.opt (mfix-ut700): New option.
7698 (mfix-gr712rc): Likewise.
7699 (sparc_fix_b2bst): New variable.
7700 * doc/invoke.texi (SPARC options): Document them.
7701 (ARM options): Fix warnings.
7702 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
7703 instructions to prevent sequences that can trigger the store-store
7704 errata for certain LEON3FT processors.
7705 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
7706 (sparc_option_override): Set sparc_fix_b2bst appropriately.
7707 * config/sparc/sparc.md (fix_b2bst): New attribute.
7708 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
7710 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
7713 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
7715 (*rsqrtsf2_sse): Ditto.
7717 (div<mode>3): Macroize insn from divdf3 and divsf3
7718 using MODEF mode iterator.
7720 2017-07-10 Martin Sebor <msebor@redhat.com>
7722 PR tree-optimization/80397
7723 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
7724 instead of testing for equality to INTEGER_TYPE.
7726 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
7728 * config.gcc: Remove uclibc from arc target spec.
7730 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
7732 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
7734 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
7737 * lto-wrapper.c (remove_option): New function.
7738 (merge_and_complain): Merge PIC/PIE options more realistically.
7740 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
7742 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
7746 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
7747 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
7748 * config.in: Regenerate.
7749 * configure: Regenerate.
7750 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
7751 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
7752 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
7753 (TARGET_GASISR_PROLOGUES): ...target mask.
7754 * common/config/avr/avr-common.c
7755 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
7756 Set -mgas-isr-prologues.
7757 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
7758 INSERT_PASS_BEFORE for it.
7759 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
7760 * config/avr/avr.c (avr_option_override)
7761 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
7762 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
7763 (avr_attribute_table) <no_gccisr>: Add new function attribute.
7764 (avr_set_current_function) <is_no_gccisr>: Init machine field.
7765 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
7767 (make_avr_pass_pre_proep): New function.
7768 (emit_push_sfr) <treg>: Add argument to function and use it
7770 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
7771 and set machine->gasisr.yes.
7772 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
7773 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
7774 __gcc_isr.n_pushed to .L__stack_usage.
7775 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
7776 (avr_asm_final_postscan_insn): ...this new static function.
7777 * config/avr/avr.h (machine_function)
7778 <is_no_gccisr, use_L__stack_usage>: New fields.
7779 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
7780 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
7781 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
7782 (gasisr, *gasisr): New expander and insn.
7783 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
7784 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
7785 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
7787 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
7789 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
7792 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
7794 Move jump-tables out of .text again.
7797 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
7798 (ASM_OUTPUT_ADDR_VEC): New function.
7799 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
7800 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
7801 INSN_ADDRESSes as asm comment.
7802 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
7803 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
7804 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
7805 * config/avr/avr.md (*tablejump): Adjust comment.
7806 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
7807 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
7809 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
7810 (avr_output_addr_vec): New proto.
7811 (avr_log_t) <insn_addresses>: New field.
7813 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
7816 * config/i386/i386.c (ix86_function_arg_advance): Set
7817 outgoing_args_on_stack to true if there are outgoing arguments
7819 (ix86_function_arg): Likewise.
7820 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
7821 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
7822 * config/i386/i386.h (machine_function): Add
7823 outgoing_args_on_stack.
7825 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
7827 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
7829 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
7831 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
7833 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
7834 (REAL_H): Remove $(MACHMODE_H).
7835 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
7837 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
7838 $(MACHMODE_H) and double-int.h.
7839 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
7841 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
7844 2017-07-07 Andrew Pinski <apinski@cavium.com>
7846 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
7847 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
7849 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7851 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
7852 Add warning if GCC was not configured to link against a GLIBC that
7853 exports the hardware capability bits.
7854 (make_resolver_func): Make resolver function private and not a
7855 COMDAT function. Create the name with clone_function_name instead
7856 of make_unique_name.
7859 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
7860 correct operand in doing the split.
7862 2017-07-07 Carl Love <cel@us.ibm.com>
7864 * config/rs6000/rs6000-c: Add support for built-in function
7865 vector unsigned short vec_pack_to_short_fp32 (vector float,
7867 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
7868 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
7869 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
7870 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
7871 (convert_4f32_8i16): Add define_expand.
7872 * doc/extend.texi: Update the built-in documentation file for the
7873 new built-in function.
7875 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7877 * config/sparc/m8.md: New file.
7878 * config/sparc/sparc.md: Include m8.md.
7880 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7882 * config/sparc/sparc.opt: New option -mvis4b.
7883 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
7884 (sparc_option_override): Handle VIS4B.
7885 (enum sparc_builtins): Define
7886 SPARC_BUILTIN_DICTUNPACK{8,16,32},
7887 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
7888 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
7889 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
7890 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
7891 (check_constant_argument): New function.
7892 (sparc_vis_init_builtins): Define builtins
7893 __builtin_vis_dictunpack{8,16,32},
7894 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
7895 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
7896 __builtin_vis_fpcmpde{8,16,32}shl and
7897 __builtin_vis_fpcmpur{8,16,32}shl.
7898 (sparc_expand_builtin): Check that the constant operands to
7899 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
7900 constant and in range.
7901 * config/sparc/sparc-c.c (sparc_target_macros): Handle
7903 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
7904 (SPARC_IMM5_P): Likewise.
7905 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
7906 (enabled): Handle vis4b.
7907 (UNSPEC_DICTUNPACK): New unspec.
7908 (UNSPEC_FPCMPSHL): Likewise.
7909 (UNSPEC_FPUCMPSHL): Likewise.
7910 (UNSPEC_FPCMPDESHL): Likewise.
7911 (UNSPEC_FPCMPURSHL): Likewise.
7912 (cpu_feature): New CPU feature `vis4b'.
7913 (dictunpack{8,16,32}): New insns.
7914 (FPCSMODE): New mode iterator.
7915 (fpcscond): New code iterator.
7916 (fpcsucond): Likewise.
7917 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
7918 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
7919 (fpcmpde{8,16,32}{si,di}shl): Likewise.
7920 (fpcmpur{8,16,32}{si,di}shl): Likewise.
7921 * config/sparc/constraints.md: Define constraints `q' for unsigned
7922 2-bit integer constants and `t' for unsigned 5-bit integer
7924 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
7926 (imm5_operand_dictunpack16): Likewise.
7927 (imm5_operand_dictunpack32): Likewise.
7928 (imm2_operand): Likewise.
7929 * doc/invoke.texi (SPARC Options): Document -mvis4b.
7930 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
7931 ditunpack* and fpcmp*shl builtins.
7933 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7935 * config.gcc: Handle m8 in --with-{cpu,tune} options.
7936 * config.in: Add HAVE_AS_SPARC6 define.
7937 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
7939 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
7941 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
7942 (CPP_CPU_SPEC): Handle m8.
7943 (ASM_CPU_SPEC): Likewise.
7944 * config/sparc/sparc-opts.h (enum processor_type): Add
7946 * config/sparc/sparc.c (m8_costs): New struct.
7947 (sparc_option_override): Handle TARGET_CPU_m8.
7948 (sparc32_initialize_trampoline): Likewise.
7949 (sparc64_initialize_trampoline): Likewise.
7950 (sparc_issue_rate): Likewise.
7951 (sparc_register_move_cost): Likewise.
7952 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
7953 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
7954 (ASM_CPU64_DEFAULT_SPEC): Likewise.
7955 (CPP_CPU_SPEC): Handle M8.
7956 (ASM_CPU_SPEC): Likewise.
7957 (AS_M8_FLAG): Define.
7958 * config/sparc/sparc.md: Add m8 to the cpu attribute.
7959 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
7960 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
7962 * configure: Regenerate.
7963 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
7966 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
7968 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
7970 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
7971 ("*movdi_insn_sp32"): Do not set v3pipe.
7972 ("*movsi_insn"): Likewise.
7973 ("*movdi_insn_sp64"): Likewise.
7974 ("*movsf_insn"): Likewise.
7975 ("*movdf_insn_sp32"): Likewise.
7976 ("*movdf_insn_sp64"): Likewise.
7977 ("*zero_extendsidi2_insn_sp64"): Likewise.
7978 ("*sign_extendsidi2_insn"): Likewise.
7979 ("*mov<VM32:mode>_insn"): Likewise.
7980 ("*mov<VM64:mode>_insn_sp64"): Likewise.
7981 ("*mov<VM64:mode>_insn_sp32"): Likewise.
7982 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
7983 ("<vlop:code><VL:mode>3"): Likewise.
7984 ("*not_<vlop:code><VL:mode>3"): Likewise.
7985 ("*nand<VL:mode>_vis"): Likewise.
7986 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
7987 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
7988 ("one_cmpl<VL:mode>2"): Likewise.
7989 ("faligndata<VM64:mode>_vis"): Likewise.
7990 ("alignaddrsi_vis"): Likewise.
7991 ("alignaddrdi_vis"): Likweise.
7992 ("alignaddrlsi_vis"): Likewise.
7993 ("alignaddrldi_vis"): Likewise.
7994 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
7995 ("bmaskdi_vis"): Likewise.
7996 ("bmasksi_vis"): Likewise.
7997 ("bshuffle<VM64:mode>_vis"): Likewise.
7998 ("cmask8<P:mode>_vis"): Likewise.
7999 ("cmask16<P:mode>_vis"): Likewise.
8000 ("cmask32<P:mode>_vis"): Likewise.
8001 ("pdistn<P:mode>_vis"): Likewise.
8002 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8004 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8006 * config/sparc/sparc.md ("subtype"): New insn attribute.
8007 ("*wrgsr_sp64"): Set insn subtype.
8008 ("*rdgsr_sp64"): Likewise.
8009 ("alignaddrsi_vis"): Likewise.
8010 ("alignaddrdi_vis"): Likewise.
8011 ("alignaddrlsi_vis"): Likewise.
8012 ("alignaddrldi_vis"): Likewise.
8013 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
8014 ("fexpand_vis"): Likewise.
8015 ("fpmerge_vis"): Likewise.
8016 ("faligndata<VM64:mode>_vis"): Likewise.
8017 ("bshuffle<VM64:mode>_vis"): Likewise.
8018 ("cmask8<P:mode>_vis"): Likewise.
8019 ("cmask16<P:mode>_vis"): Likewise.
8020 ("cmask32<P:mode>_vis"): Likewise.
8021 ("fchksm16_vis"): Likewise.
8022 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
8023 ("fmean16_vis"): Likewise.
8024 ("fp<plusminus_insn>64_vis"): Likewise.
8025 ("<plusminus_insn>v8qi3"): Likewise.
8026 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
8027 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
8028 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
8029 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
8030 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
8031 ("*movqi_insn"): Likewise.
8032 ("*movhi_insn"): Likewise.
8033 ("*movsi_insn"): Likewise.
8034 ("movsi_pic_gotdata_op"): Likewise.
8035 ("*movdi_insn_sp32"): Likewise.
8036 ("*movdi_insn_sp64"): Likewise.
8037 ("movdi_pic_gotdata_op"): Likewise.
8038 ("*movsf_insn"): Likewise.
8039 ("*movdf_insn_sp32"): Likewise.
8040 ("*movdf_insn_sp64"): Likewise.
8041 ("*zero_extendhisi2_insn"): Likewise.
8042 ("*zero_extendqihi2_insn"): Likewise.
8043 ("*zero_extendqisi2_insn"): Likewise.
8044 ("*zero_extendqidi2_insn"): Likewise.
8045 ("*zero_extendhidi2_insn"): Likewise.
8046 ("*zero_extendsidi2_insn_sp64"): Likewise.
8047 ("ldfsr"): Likewise.
8048 ("prefetch_64"): Likewise.
8049 ("prefetch_32"): Likewise.
8050 ("tie_ld32"): Likewise.
8051 ("tie_ld64"): Likewise.
8052 ("*tldo_ldub_sp32"): Likewise.
8053 ("*tldo_ldub1_sp32"): Likewise.
8054 ("*tldo_ldub2_sp32"): Likewise.
8055 ("*tldo_ldub_sp64"): Likewise.
8056 ("*tldo_ldub1_sp64"): Likewise.
8057 ("*tldo_ldub2_sp64"): Likewise.
8058 ("*tldo_ldub3_sp64"): Likewise.
8059 ("*tldo_lduh_sp32"): Likewise.
8060 ("*tldo_lduh1_sp32"): Likewise.
8061 ("*tldo_lduh_sp64"): Likewise.
8062 ("*tldo_lduh1_sp64"): Likewise.
8063 ("*tldo_lduh2_sp64"): Likewise.
8064 ("*tldo_lduw_sp32"): Likewise.
8065 ("*tldo_lduw_sp64"): Likewise.
8066 ("*tldo_lduw1_sp64"): Likewise.
8067 ("*tldo_ldx_sp64"): Likewise.
8068 ("*mov<VM32:mode>_insn"): Likewise.
8069 ("*mov<VM64:mode>_insn_sp64"): Likewise.
8070 ("*mov<VM64:mode>_insn_sp32"): Likewise.
8072 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8074 * config/sparc/sparc.md ("type"): New insn type viscmp.
8075 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
8077 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
8078 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
8079 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
8080 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
8082 ("n7_vis_logical_11cycle"): Likewise.
8083 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
8084 * config/sparc/niagara2.md ("niag3_vis": Likewise.
8085 * config/sparc/niagara.md ("niag_vis"): Likewise.
8086 * config/sparc/ultra3.md ("us3_fga"): Likewise.
8087 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
8089 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
8091 * config/sparc/sparc.md: New instruction type `bmask'.
8092 (bmaskdi_vis): Use the `bmask' type.
8093 (bmasksi_vis): Likewise.
8094 * config/sparc/ultra3.md (us3_array): Likewise.
8095 * config/sparc/niagara7.md (n7_array): Likewise.
8096 * config/sparc/niagara4.md (n4_array): Likewise.
8097 * config/sparc/niagara2.md (niag2_vis): Likewise.
8098 (niag3_vis): Likewise.
8099 * config/sparc/niagara.md (niag_vis): Likewise.
8101 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8103 * ipa-comdats.c: Remove optimize check from gate.
8104 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
8105 for functions not optimized.
8106 (ipa_fn_summary_read): Skip optimize check.
8107 (ipa_fn_summary_write): Likewise.
8108 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
8110 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
8112 (can_inline_edge_p): Check flag_pcc_struct_return for match.
8113 (check_callers): Give up on caller which is not optimized.
8114 (inline_small_functions): Likewise.
8115 (ipa_inline): Do not give up when not optimizing.
8116 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
8117 away unoptimizes cdtors.
8118 (whole_program_function_and_variable_visibility): Do
8119 ipa_discover_readonly_nonaddressable_vars in LTO mode.
8120 * ipa.c (process_references): Do not check optimize.
8121 (symbol_table::remove_unreachable_nodes): Update optimize check.
8122 (set_writeonly_bit): Update optimize check.
8123 (pass_ipa_cdtor_merge::gate): Do not check optimize.
8124 (pass_ipa_single_use::gate): Remove.
8126 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8128 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
8129 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
8130 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
8131 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
8132 permute_load, permute_store, adjust_extract, adjust_splat,
8133 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
8134 replace_swap_with_copy, dump_swap_insn_table,
8135 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
8136 recombine_lvx_pattern, recombine_stvx_pattern,
8137 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
8138 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
8139 to file rs6000-p8swap.c.
8140 * config/rs6000/rs6000-p8swap.c: New file.
8141 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
8142 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
8143 and rs6000*-*-* targets.
8145 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8147 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
8149 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8151 * lto-wrapper.c (merge_and_complain): Do not merge
8152 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
8153 fsigned_zeros, ftrapping_math, fwrapv.
8154 (append_compiler_options): Do not track these options.
8155 (append_linker_options): Likewie
8157 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8159 * cgraphunit.c (cgraph_node::finalize_function): When
8160 !flag_toplevel_reorde set no_reorder flag.
8161 (varpool_node::finalize_decl): Likewise.
8162 (symbol_table::compile): Drop no toplevel reorder path.
8164 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8166 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
8167 edges; zero probability is not better than uninitialized.
8169 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8171 * asan.h (asan_sanitize_allocas_p): Declare.
8172 * asan.c (asan_sanitize_allocas_p): New function.
8173 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
8174 (handle_builtin_alloca): Likewise.
8175 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
8176 if !asan_sanitize_allocas_p.
8177 * params.def (asan-instrument-allocas): Add new option.
8178 * params.h (ASAN_PROTECT_ALLOCAS): Define.
8179 * opts.c (common_handle_option): Disable allocas sanitization for
8182 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
8184 * asan.c: Include gimple-fold.h.
8185 (get_last_alloca_addr): New function.
8186 (handle_builtin_stackrestore): Likewise.
8187 (handle_builtin_alloca): Likewise.
8188 (asan_emit_allocas_unpoison): Likewise.
8189 (get_mem_refs_of_builtin_call): Add new parameter, remove const
8190 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
8191 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
8192 (instrument_builtin_call): Pass gimple iterator to
8193 get_mem_refs_of_builtin_call.
8194 (last_alloca_addr): New global.
8195 * asan.h (asan_emit_allocas_unpoison): Declare.
8196 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
8197 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
8198 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
8199 if function calls alloca.
8200 * gimple-fold.c (replace_call_with_value): Remove static keyword.
8201 * gimple-fold.h (replace_call_with_value): Declare.
8202 * internal-fn.c: Include asan.h.
8203 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
8204 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
8206 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8208 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
8209 (C_SELFTEST_FLAGS): New.
8210 (CPP_SELFTEST_FLAGS): New.
8211 (SELFTEST_DEPS): New, from deps of s-selftest.
8212 (C_SELFTEST_DEPS): New, from deps of s-selftest.
8213 (CPP_SELFTEST_DEPS): New.
8214 (selftest): Add dependency on s-selftest-c++.
8215 (s-selftest): Rename to...
8216 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
8217 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
8218 than SELFTEST_FLAGS.
8219 (selftest-gdb): Rename to...
8220 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
8222 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
8223 (selftest-valgrind): Rename to...
8224 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
8226 (selftest-valgrind): Reintroduce as an alias for
8227 selftest-c-valgrind.
8228 (s-selftest-c++): New.
8229 (selftest-c++-gdb): New.
8230 (selftest-c++-valgrind): New.
8232 2017-07-06 Olivier Hainque <hainque@adacore.com>
8234 * gcc.c (process_command): When deciding if undefined variables
8235 should be ignored when processing specs, accept "gcc -v" as well.
8237 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
8239 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
8240 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
8242 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
8244 * config/arm/arm-cpus.in (armv8-r): Add new entry.
8245 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
8246 * config/arm/arm-tables.opt: Regenerate.
8247 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
8249 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
8251 2017-07-06 Carl Love <cel@us.ibm.com>
8253 * ChangeLog: Clean up from mid air collision
8255 2017-07-06 Carl Love <cel@us.ibm.com>
8257 * config/rs6000/rs6000-c.c: Add support for built-in functions
8258 vector signed int vec_subc (vector signed int, vector signed int);
8259 vector signed __int128 vec_subc (vector signed __int128,
8260 vector signed __int128);
8261 vector unsigned __int128 vec_subc (vector unsigned __int128,
8262 vector unsigned __int128);
8263 vector signed int vec_sube (vector signed int, vector signed int,
8265 vector unsigned int vec_sube (vector unsigned int,
8266 vector unsigned int,
8267 vector unsigned int);
8268 vector signed __int128 vec_sube (vector signed __int128,
8269 vector signed __int128,
8270 vector signed__int128);
8271 vector unsigned __int128 vec_sube (vector unsigned __int128,
8272 vector unsigned __int128,
8273 vector unsigned __int128);
8274 vector signed int vec_subec (vector signed int, vector signed int,
8276 vector unsigned int vec_subec (vector unsigned int,
8277 vector unsigned int,
8278 vector unsigned int);
8279 vector signed __int128 vec_subec (vector signed __int128,
8280 vector signed __int128,
8281 vector signed__int128);
8282 vector unsigned __int128 vec_subec (vector unsigned __int128,
8283 vector unsigned __int128,
8284 vector unsigned __int128);
8285 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
8286 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
8287 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
8288 BU_ALTIVEC_OVERLOAD_X definitions.
8289 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
8290 * doc/extend.texi: Update the built-in documentation file for the new
8293 2017-07-06 David Malcolm <dmalcolm@redhat.com>
8296 * diagnostic-show-locus.c (layout::layout): Use start and finish
8297 spelling location for the start and finish of each range.
8298 * genmatch.c (linemap_client_expand_location_to_spelling_point):
8299 Add unused aspect param.
8300 * input.c (expand_location_1): Add "aspect" param, and use it
8301 to access the correct part of the location.
8302 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
8304 (expand_location_to_spelling_point): Likewise.
8305 (linemap_client_expand_location_to_spelling_point): Add "aspect"
8306 param, and pass it to expand_location_1.
8308 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
8310 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
8311 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
8312 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
8313 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
8314 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
8315 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
8316 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
8317 _mm_maskz_getmant_ss): New intrinsics.
8318 (__builtin_ia32_getexpss128_mask): Changed to ...
8319 __builtin_ia32_getexpss128_round ... this.
8320 (__builtin_ia32_getexpsd128_mask): Changed to ...
8321 __builtin_ia32_getexpsd128_round ... this.
8322 * config/i386/i386-builtin-types.def
8323 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
8324 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
8325 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
8326 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
8327 __builtin_ia32_getmantss_mask_round): New builtins.
8328 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
8329 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
8330 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
8331 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
8332 * config/i386/sse.md
8333 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
8334 avx512f_sgetexp<mode><mask_scalar_name>
8335 <round_saeonly_scalar_name> ... this.
8336 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
8337 %0, %1, %2<round_saeonly_op3>}): Changed to ...
8338 vgetexp<ssescalarmodesuffix>
8339 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8340 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
8341 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
8342 avx512f_vgetmant<mode><mask_scalar_name>
8343 <round_saeonly_scalar_name> ... this.
8344 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
8345 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
8346 vgetmant<ssescalarmodesuffix>
8347 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
8348 %0<mask_scalar_operand4>, %1, %2
8349 <round_saeonly_scalar_mask_op4>, %3} ... this.
8350 * config/i386/subst.md (mask_scalar_operand4,
8351 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
8352 round_saeonly_scalar_nimm_predicate): New subst attributes.
8354 2017-07-06 Julia Koval <julia.koval@intel.com>
8356 * config/i386/i386.c (ix86_erase_embedded_rounding):
8357 Remove code for old rounding pattern.
8359 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
8361 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
8363 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
8365 * doc/sourcebuild.texi (Test Directives, Variants of
8366 dg-require-support): Add documentation for dg-require-stack-check.
8368 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
8370 * config/i386/subst.md (mask_scalar, round_scalar,
8371 round_saeonly_scalar): New meta-templates.
8372 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
8373 round_scalar_mask_operand3, round_scalar_mask_op3,
8374 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
8375 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
8376 round_saeonly_scalar_constraint,
8377 round_saeonly_scalar_prefix): New subst attribute.
8378 * config/i386/sse.md
8379 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
8380 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
8381 <round_scalar_name> ... this.
8382 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
8383 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
8384 <round_scalar_name> ... this.
8385 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
8386 <sse>_vm<code><mode>3<mask_scalar_name>
8387 <round_saeonly_scalar_name> ... this.
8388 (v<plusminus_mnemonic><ssescalarmodesuffix>
8389 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8390 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8391 v<plusminus_mnemonic><ssescalarmodesuffix>
8392 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8393 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8394 (v<multdiv_mnemonic><ssescalarmodesuffix>
8395 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
8396 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
8397 v<multdiv_mnemonic><ssescalarmodesuffix>
8398 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8399 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
8400 (v<maxmin_float><ssescalarmodesuffix>
8401 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
8402 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
8403 v<maxmin_float><ssescalarmodesuffix>
8404 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
8405 %0<mask_scalar_operand3>, %1, %<iptr>2
8406 <round_saeonly_scalar_mask_op3>} ... this.
8408 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
8410 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
8411 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
8413 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
8414 Alan Hayward <alan.hayward@arm.com>
8415 David Sherwood <david.sherwood@arm.com>
8417 * combine.c (simplify_if_then_else): Remove "enum" before
8419 * compare-elim.c (can_eliminate_compare): Likewise.
8420 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
8422 (aarch64_lookup_simd_builtin_type): Likewise.
8423 (aarch64_simd_builtin_type): Likewise.
8424 (aarch64_init_simd_builtin_types): Likewise.
8425 (aarch64_simd_expand_args): Likewise.
8426 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
8428 (aarch64_reverse_mask): Likewise.
8429 (aarch64_simd_emit_reg_reg_move): Likewise.
8430 (aarch64_gen_adjusted_ldpstp): Likewise.
8431 (aarch64_ccmp_mode_to_code): Likewise.
8432 (aarch64_operands_ok_for_ldpstp): Likewise.
8433 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8434 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
8436 (aarch64_min_divisions_for_recip_mul): Likewise.
8437 (aarch64_reassociation_width): Likewise.
8438 (aarch64_get_condition_code_1): Likewise.
8439 (aarch64_simd_emit_reg_reg_move): Likewise.
8440 (aarch64_simd_attr_length_rglist): Likewise.
8441 (aarch64_reverse_mask): Likewise.
8442 (aarch64_operands_ok_for_ldpstp): Likewise.
8443 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8444 (aarch64_gen_adjusted_ldpstp): Likewise.
8445 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
8447 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
8448 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
8449 (arm_lookup_simd_builtin_type): Likewise.
8450 (arm_simd_builtin_type): Likewise.
8451 (arm_init_simd_builtin_types): Likewise.
8452 (arm_expand_builtin_args): Likewise.
8453 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
8454 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
8455 (ft32_setup_incoming_varargs): Likewise.
8456 (ft32_function_arg): Likewise.
8457 (ft32_function_arg_advance): Likewise.
8458 (ft32_pass_by_reference): Likewise.
8459 (ft32_arg_partial_bytes): Likewise.
8460 (ft32_valid_pointer_mode): Likewise.
8461 (ft32_addr_space_pointer_mode): Likewise.
8462 (ft32_addr_space_legitimate_address_p): Likewise.
8463 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
8465 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
8466 (ix86_emit_outlined_ms2sysv_restore): Likewise.
8467 (iamcu_alignment): Likewise.
8468 (canonicalize_vector_int_perm): Likewise.
8469 (ix86_noce_conversion_profitable_p): Likewise.
8470 (ix86_mpx_bound_mode): Likewise.
8471 (ix86_operands_ok_for_move_multiple): Likewise.
8472 * config/microblaze/microblaze-protos.h
8473 (microblaze_expand_conditional_branch_reg): Likewise.
8474 * config/microblaze/microblaze.c
8475 (microblaze_expand_conditional_branch_reg): Likewise.
8476 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8478 (rs6000_reassociation_width): Likewise.
8479 (rs6000_invalid_binary_op): Likewise.
8480 (fusion_p9_p): Likewise.
8481 (emit_fusion_p9_load): Likewise.
8482 (emit_fusion_p9_store): Likewise.
8483 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
8485 (riscv_hard_regno_mode_ok_p): Likewise.
8486 (riscv_address_insns): Likewise.
8487 (riscv_split_symbol): Likewise.
8488 (riscv_legitimize_move): Likewise.
8489 (riscv_function_value): Likewise.
8490 (riscv_hard_regno_nregs): Likewise.
8491 (riscv_expand_builtin): Likewise.
8492 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
8493 (riscv_build_integer): Likewise.
8494 (riscv_split_integer): Likewise.
8495 (riscv_legitimate_constant_p): Likewise.
8496 (riscv_cannot_force_const_mem): Likewise.
8497 (riscv_regno_mode_ok_for_base_p): Likewise.
8498 (riscv_valid_base_register_p): Likewise.
8499 (riscv_valid_offset_p): Likewise.
8500 (riscv_valid_lo_sum_p): Likewise.
8501 (riscv_classify_address): Likewise.
8502 (riscv_legitimate_address_p): Likewise.
8503 (riscv_address_insns): Likewise.
8504 (riscv_load_store_insns): Likewise.
8505 (riscv_force_binary): Likewise.
8506 (riscv_split_symbol): Likewise.
8507 (riscv_force_address): Likewise.
8508 (riscv_legitimize_address): Likewise.
8509 (riscv_move_integer): Likewise.
8510 (riscv_legitimize_const_move): Likewise.
8511 (riscv_legitimize_move): Likewise.
8512 (riscv_address_cost): Likewise.
8513 (riscv_subword): Likewise.
8514 (riscv_output_move): Likewise.
8515 (riscv_canonicalize_int_order_test): Likewise.
8516 (riscv_emit_int_order_test): Likewise.
8517 (riscv_function_arg_boundary): Likewise.
8518 (riscv_pass_mode_in_fpr_p): Likewise.
8519 (riscv_pass_fpr_single): Likewise.
8520 (riscv_pass_fpr_pair): Likewise.
8521 (riscv_get_arg_info): Likewise.
8522 (riscv_function_arg): Likewise.
8523 (riscv_function_arg_advance): Likewise.
8524 (riscv_arg_partial_bytes): Likewise.
8525 (riscv_function_value): Likewise.
8526 (riscv_pass_by_reference): Likewise.
8527 (riscv_setup_incoming_varargs): Likewise.
8528 (riscv_print_operand): Likewise.
8529 (riscv_elf_select_rtx_section): Likewise.
8530 (riscv_save_restore_reg): Likewise.
8531 (riscv_for_each_saved_reg): Likewise.
8532 (riscv_register_move_cost): Likewise.
8533 (riscv_hard_regno_mode_ok_p): Likewise.
8534 (riscv_hard_regno_nregs): Likewise.
8535 (riscv_class_max_nregs): Likewise.
8536 (riscv_memory_move_cost): Likewise.
8537 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
8538 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
8539 (rl78_addr_space_address_mode): Likewise.
8540 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8542 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8543 (rs6000_reassociation_width): Likewise.
8544 (rs6000_invalid_binary_op): Likewise.
8545 (fusion_p9_p): Likewise.
8546 (emit_fusion_p9_load): Likewise.
8547 (emit_fusion_p9_store): Likewise.
8548 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
8549 (ok_for_simple_move_operands): Likewise.
8550 (ok_for_simple_move_strict_operands): Likewise.
8551 (ok_for_simple_arith_logic_operands): Likewise.
8552 (visium_legitimize_reload_address): Likewise.
8553 (visium_select_cc_mode): Likewise.
8554 (output_cbranch): Likewise.
8555 (visium_split_double_move): Likewise.
8556 (visium_expand_copysign): Likewise.
8557 (visium_expand_int_cstore): Likewise.
8558 (visium_expand_fp_cstore): Likewise.
8559 * config/visium/visium.c (visium_pass_by_reference): Likewise.
8560 (visium_function_arg): Likewise.
8561 (visium_function_arg_advance): Likewise.
8562 (visium_libcall_value): Likewise.
8563 (visium_setup_incoming_varargs): Likewise.
8564 (visium_legitimate_constant_p): Likewise.
8565 (visium_legitimate_address_p): Likewise.
8566 (visium_legitimize_address): Likewise.
8567 (visium_secondary_reload): Likewise.
8568 (visium_register_move_cost): Likewise.
8569 (visium_memory_move_cost): Likewise.
8570 (prepare_move_operands): Likewise.
8571 (ok_for_simple_move_operands): Likewise.
8572 (ok_for_simple_move_strict_operands): Likewise.
8573 (ok_for_simple_arith_logic_operands): Likewise.
8574 (visium_function_value_1): Likewise.
8575 (rtx_ok_for_offset_p): Likewise.
8576 (visium_legitimize_reload_address): Likewise.
8577 (visium_split_double_move): Likewise.
8578 (visium_expand_copysign): Likewise.
8579 (visium_expand_int_cstore): Likewise.
8580 (visium_expand_fp_cstore): Likewise.
8581 (visium_split_cstore): Likewise.
8582 (visium_select_cc_mode): Likewise.
8583 (visium_split_cbranch): Likewise.
8584 (output_cbranch): Likewise.
8585 (visium_print_operand_address): Likewise.
8586 * expmed.c (flip_storage_order): Likewise.
8587 * expmed.h (emit_cstore): Likewise.
8588 (flip_storage_order): Likewise.
8589 * genrecog.c (validate_pattern): Likewise.
8590 * hsa-gen.c (gen_hsa_addr): Likewise.
8591 * internal-fn.c (expand_arith_overflow): Likewise.
8592 * ira-color.c (allocno_copy_cost_saving): Likewise.
8593 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8594 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
8595 (process_invariant_for_inheritance): Likewise.
8596 * lra-eliminations.c (move_plus_up): Likewise.
8597 * omp-low.c (lower_oacc_reductions): Likewise.
8598 * simplify-rtx.c (simplify_subreg): Likewise.
8599 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
8600 (TARGET_CHKP_BOUND_MODE): Likewise..
8601 * targhooks.c (default_chkp_bound_mode): Likewise.
8602 (default_setup_incoming_vararg_bounds): Likewise.
8603 * targhooks.h (default_chkp_bound_mode): Likewise.
8604 (default_setup_incoming_vararg_bounds): Likewise.
8605 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
8606 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
8607 (have_whole_vector_shift): Likewise.
8608 * tree-vect-stmts.c (vectorizable_load): Likewise.
8609 * doc/tm.texi: Regenerate.
8611 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
8613 Graceful degrade if Binutils PR21472 is not available.
8616 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
8617 .rodata in flash test fails.
8618 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
8619 * confgure: Regenerate.
8620 * config.in: Regenerate.
8621 * config/avr/avr.c (avr_asm_named_section)
8622 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
8623 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
8624 (avr_asm_init_sections): Same.
8626 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8628 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
8629 (fma<VH:mode>4_intrinsic): Likewise.
8630 (*fmsub<VCVTF:mode>4): Likewise.
8631 (*fmsub<VH:mode>4_intrinsic): Likewise.
8633 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
8636 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
8637 Don't depend on "optimize > 0".
8638 (out_movhi_r_mr, out_movqi_mr_r): Same.
8639 (out_movhi_mr_r, out_movqi_r_mr): Same.
8640 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
8641 io_address_operand on "optimize > 0".
8643 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8645 * tree-loop-distribution.c: Add general explanantion on the pass.
8646 (generate_loops_for_partition): Mark distributed loop.
8647 (pg_add_dependence_edges): New parameter. Handle alias data
8648 dependence specially and record it in the parameter if asked.
8649 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
8650 (init_partition_graph_vertices, add_partition_graph_edge): New.
8651 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
8652 (free_partition_graph_vdata, build_partition_graph): New.
8653 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
8654 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
8655 (data_ref_segment_size, latch_dominated_by_data_ref): New.
8656 (compute_alias_check_pairs, version_loop_by_alias_check): New.
8657 (version_for_distribution_p, finalize_partitions): New.
8658 (distribute_loop): Handle alias data dependence specially. Factor
8659 out loop fusion code as functions and call these functions.
8661 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8663 * tree-loop-distribution.c (classify_partition): New parameter and
8664 better handle reduction statement.
8665 (rdg_build_partitions): Revise comment.
8666 (distribute_loop): Compute statements in all partitions and pass it
8667 to classify_partition.
8669 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8671 * tree-loop-distribution.c (enum partition_type): New.
8672 (struct partition): New field type.
8673 (partition_merge_into): Add parameter. Update partition type.
8674 (data_dep_in_cycle_p, update_type_for_merge): New functions.
8675 (build_rdg_partition_for_vertex): Compute partition type.
8676 (rdg_build_partitions): Dump partition type.
8677 (distribute_loop): Update calls to partition_merge_into.
8679 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8681 * tree-loop-distribution.c (struct ddr_hasher): New.
8682 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
8684 (classify_partition): Call get_data_dependence.
8685 (pg_add_dependence_edges): Ditto.
8686 (distribute_loop): Release data dependence hash table.
8688 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8690 * tree-loop-distribution.c (ref_base_address): Delete.
8691 (similar_memory_accesses): Rename ...
8692 (share_memory_accesses): ... to this. Check if partitions access
8693 the same memory reference.
8694 (distribute_loop): Call share_memory_accesses.
8696 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8698 * tree-loop-distribution.c (struct partition): New field recording
8700 (partition_alloc, partition_free): Init and release data refs.
8701 (partition_merge_into): Merge data refs.
8702 (build_rdg_partition_for_vertex): Collect data refs for partition.
8703 (pg_add_dependence_edges): Change parameters from vector to bitmap.
8705 (distribute_loop): Remve data refs from vertice data of partition
8708 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8710 * tree-loop-distribution.c (params.h): Include header file.
8711 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
8712 (datarefs_vec): New global var.
8713 (create_rdg_vertices): Use datarefs_vec directly.
8714 (free_rdg): Don't free data references.
8715 (build_rdg): Update use. Don't free data references.
8716 (distribute_loop): Compute global variable for data references.
8717 Bail out if there are too many data references.
8719 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8721 * tree-loop-distribution.c (loop_nest): New global var.
8722 (build_rdg): Use loop directly, rather than loop nest.
8723 (pg_add_dependence_edges): Remove loop nest parameter. Use global
8725 (distribute_loop): Compute global variable loop nest. Update use.
8727 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8729 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
8730 (partition_merge_into): New parameter. Dump reason for fusion.
8731 (distribute_loop): Update use of partition_merge_into.
8733 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8735 * tree-loop-distribution.c (bb_top_order_index): New.
8736 (bb_top_order_index_size, bb_top_order_cmp): New.
8737 (stmts_from_loop): Use topological order.
8738 (pass_loop_distribution::execute): Compute and release topological
8739 order for basic blocks.
8741 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8743 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
8746 2017-07-05 Bin Cheng <bin.cheng@arm.com>
8748 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
8749 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
8750 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
8751 * internal-fn.def (LOOP_DIST_ALIAS): New.
8752 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
8753 (fold_loop_internal_call): ... this.
8754 (vect_loop_dist_alias_call): New function.
8755 (set_uid_loop_bbs): Call fold_loop_internal_call.
8756 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
8759 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
8762 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
8763 Require dead FLAGS_REG at the beginning of a peephole.
8765 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
8768 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
8769 arguments in the call to __builtin_ia32_sbb_u32.
8770 (_subborrow_u64): Swap _X and _Y arguments in the call to
8771 __builtin_ia32_sbb_u64.
8773 2017-07-04 Jakub Jelinek <jakub@redhat.com>
8776 * tree-vrp.c (compare_assert_loc): Turn into a function template
8777 with stable template parameter. Only test if a->e is NULL,
8778 !a->e == !b->e has been verified already. Use e == NULL or
8779 e != NULL instead of e or ! e tests. If stable is true, don't use
8780 iterative_hash_expr, on the other side allow a or b or both NULL
8781 and sort the NULLs last.
8782 (process_assert_insertions): Sort using compare_assert_loc<false>
8783 instead of compare_assert_loc, later sort using
8784 compare_assert_loc<true> before calling process_assert_insertions_for
8785 in a loop. Use break instead of continue once seen NULL pointer.
8787 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
8789 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
8790 Cortex-R7 and Cortex-R8 processors.
8792 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8794 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
8795 uninitialized while src is not.
8797 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
8799 * common/config/arm/arm-common.c: Adjust include path for
8801 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
8802 (arm-cpu.h): Create in build directory. Adjust dependency rules.
8803 (arm-cpu-data.h): Likewise.
8804 (arm-cpu-cdata.h): Likewise.
8805 * config/arm/arm-cpu.h: Delete.
8806 * config/arm/arm-cpu-cdata.h: Delete.
8807 * config/arm/arm-cpu-data.h: Delete.
8809 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
8811 * config/arm/arm-cpus.in (cortex-a55): New.
8812 (cortex-a75): Likewise.
8813 (cortex-a75.cortex-a55): Likewise.
8814 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
8816 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
8817 * config/arm/arm-cpu-cdata.h: Regenerate.
8818 * config/arm/arm-cpu-data.h: Regenerate.
8819 * config/arm/arm-cpu.h: Regenerate.
8820 * config/arm/arm-tables.opt: Regenerate.
8821 * config/arm/arm-tune.md: Regenerate.
8823 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8825 * haifa-sched.c (sched_create_recovery_edges): Update profile.
8827 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
8829 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
8832 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
8834 PR tree-optimization/81292
8835 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
8836 full_string_p, also call adjust_related_strinfos if the adjustment
8837 is simple, otherwise invalidate related strinfos.
8839 2017-07-04 Martin Liska <mliska@suse.cz>
8842 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
8843 newly created variable as DECL_IGNORED_P.
8845 2017-07-04 Martin Liska <mliska@suse.cz>
8848 * ipa-inline.c (inline_small_functions):
8849 Use xstrdup_for_dump.
8851 2017-07-04 Tom de Vries <tom@codesourcery.com>
8853 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
8855 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
8858 * config/darwin.c (darwin_function_switched_text_sections):
8861 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
8863 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
8865 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
8867 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
8869 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8871 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
8872 min_profitable_iters, and th as inclusive lower bounds.
8873 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
8874 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
8875 for min_profitable_iters and min_profitable_estimate.
8876 (vect_transform_loop): Treat th as an inclusive lower bound.
8877 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
8879 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
8882 * config/darwin.c (darwin_function_switched_text_sections):
8883 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
8884 in two pieces, and suppress the use of buf.
8886 2017-07-03 Nathan Sidwell <nathan@acm.org>
8888 * hash-table.h (hash_table_mod1): Fix indentation.
8890 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
8893 * predict.c (force_edge_cold): Be more careful about propagation
8895 * profile-count.h (profile_probability::guessed,
8896 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
8898 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
8900 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
8902 * doc/invoke.texi (rcpc architecture extension): Document it.
8904 2017-07-03 Richard Biener <rguenther@suse.de>
8906 PR tree-optimization/60510
8907 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
8908 the scalar reduction PHI and use it.
8909 (vectorizable_reduction): Properly guard the single_defuse_cycle
8910 path for non-SLP reduction chains where we cannot use it.
8911 Rework reduc_def/index and vector type deduction. Rework
8912 vector operand gathering during reduction op code-gen.
8913 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
8914 chains dissolve the chain and leave it to non-SLP reduction
8917 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8919 * tree-data-ref.h (dr_alignment): Declare.
8920 * tree-data-ref.c (dr_alignment): New function.
8921 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
8922 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
8924 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
8926 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8928 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
8929 and base_misalignment fields.
8930 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
8931 * tree-data-ref.c: Include builtins.h.
8932 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
8933 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
8934 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
8935 * tree-vect-data-refs.c: Include tree-cfg.h.
8936 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
8937 fields instead of calculating an alignment here.
8938 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
8939 innermost_loop_behavior fields.
8941 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8943 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
8945 (DR_STEP_ALIGNMENT): New macro.
8946 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
8947 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
8948 (create_data_ref): Print it.
8949 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
8950 to tell whether the step preserves vector (mis)alignment.
8951 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
8952 Move the check for an integer step and generalise to all INTEGER_CST.
8953 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
8954 Print the outer step alignment.
8956 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8958 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
8959 with offset_alignment.
8960 (DR_ALIGNED_TO): Delete.
8961 (DR_OFFSET_ALIGNMENT): New macro.
8962 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
8963 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
8964 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
8965 (create_data_ref): Likewise.
8966 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
8967 (vect_analyze_data_refs): Likewise.
8968 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
8969 creating dummy innermost behavior.
8971 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8973 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
8974 with a "innermost_loop_behavior *" and refeence tree.
8975 * tree-data-ref.c (dr_analyze_innermost): Likewise.
8976 (create_data_ref): Update call accordingly.
8977 * tree-predcom.c (find_looparound_phi): Likewise.
8979 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
8981 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
8982 fields with dr_wrt_vec_loop.
8983 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
8984 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
8985 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
8986 (vect_dr_behavior): New function.
8987 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
8988 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
8989 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
8990 track whether the step preserves the misalignment.
8991 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
8992 Use vect_dr_behavior.
8993 (vect_setup_realignment): Update call accordingly.
8994 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
8995 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
8996 call to vect_create_addr_base_for_vector_ref.
8997 (vect_create_cond_for_align_checks): Likewise.
8998 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
8999 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
9000 (vect_recog_mask_conversion_pattern): Likewise.
9001 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
9002 (new_stmt_vec_info): Remove redundant zeroing.
9004 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
9006 * common/config/arm/arm-common.c (arm_be8_option): New function.
9007 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
9008 (ISA_ARMv6): Add isa_bit_be8.
9009 * config/arm/arm.h (arm_be8_option): Add prototype.
9010 (BE8_SPEC_FUNCTION): New define.
9011 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
9012 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
9013 (mlittle-endian): Similarly.
9014 (mbe8, mbe32): New options.
9015 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
9016 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
9018 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9020 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
9022 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9024 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
9025 (cleanup_tree_cfg_bb): Use it.
9026 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
9028 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
9030 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9033 * loop-doloop.c (add_test): Update profile.
9035 2017-07-03 Martin Liska <mliska@suse.cz>
9038 * sanopt.c (rewrite_usage_of_param): New function.
9039 (sanitize_rewrite_addressable_params): Likewise.
9040 (pass_sanopt::execute): Call rewrite_usage_of_param.
9042 2017-07-03 Richard Biener <rguenther@suse.de>
9044 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
9045 back to using VIEW_CONVERT_EXPR.
9047 2017-07-03 Martin Liska <mliska@suse.cz>
9050 * doc/extend.texi: Document when a resolver function is
9051 generated for target_clones.
9053 2017-07-03 Martin Liska <mliska@suse.cz>
9055 * asan.c (asan_emit_stack_protection): Unpoison just red zones
9056 and shadow memory of auto variables which are subject of
9057 use-after-scope sanitization.
9058 (asan_expand_mark_ifn): Add do set only when is_poison.
9060 2016-07-03 Richard Biener <rguenther@suse.de>
9062 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
9064 (vect_force_simple_reduction): Record reduction def -> phi mapping.
9065 (vectorizable_reduction): Perform reduction PHI creation when
9066 visiting a reduction PHI and adjust and simplify code generation
9067 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
9068 (vect_transform_loop): Visit reduction PHIs.
9069 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
9070 defs into the SLP tree.
9071 (vect_build_slp_tree): Reduction defs terminate the recursion.
9072 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
9074 (vect_get_vec_defs_for_stmt_copy): Export.
9075 (vect_get_vec_defs): Likewise.
9076 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
9078 (vect_get_vec_defs_for_stmt_copy): Declare.
9079 (vect_get_vec_defs): Likewise.
9081 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
9083 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
9084 parameter with a "loop" parameter and use it instead of the
9085 loop containing DR_STMT. Don't check simple_iv when doing
9086 BB analysis. Describe the two analysis modes in the comment.
9088 2017-07-03 Tom de Vries <tom@codesourcery.com>
9090 PR tree-optimization/69468
9091 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
9092 (find_same_succ_bb): Handle ignore_edge_flags.
9094 2017-07-03 Tom de Vries <tom@codesourcery.com>
9096 PR tree-optimization/81192
9097 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
9099 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
9101 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
9103 2017-07-03 Tom de Vries <tom@codesourcery.com>
9105 PR tree-optimization/81192
9106 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
9107 BB_SAME_SUCC (bb) == NULL.
9109 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9111 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
9114 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9116 * dumpfile.c: Include profile-count.h
9117 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
9119 (insert_cond_bb): Update profile.
9120 * tree-cfg.h (insert_cond_bb): Update prototype.
9121 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
9122 * tree-dump.c: Do not include tree-cfg.
9124 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9126 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
9128 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9130 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
9133 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
9135 * tree-complex.c (expand_complex_div_wide): update profile.
9137 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9138 Alan Hayward <alan.hayward@arm.com>
9139 David Sherwood <david.sherwood@arm.com>
9141 * Makefile.in (MACHMODE_H): Remove insn-modes.h
9142 (CORETYPES_H): New define.
9143 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
9144 (insn-modes-inline.h, s-modes-inline-h): New rules.
9145 (generated_files): Add insn-modes-inline.h.
9146 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
9147 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
9148 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
9149 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
9150 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
9151 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
9152 (build/gencodes.o, build/genconditions.o): Likewise.
9153 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
9154 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
9155 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
9156 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
9157 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
9158 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
9159 * coretypes.h: Include everything up to real.h for generators.
9160 Include insn-modes.h first. Include wide-int-print.h after
9161 wide-int.h. Include insn-modes-inline.h and then machmode.h.
9162 * machmode.h: Don't include insn-modes.h here.
9163 * function-tests.c: Remove includes of signop.h, machmode.h,
9164 double-int.h and wide-int.h.
9166 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
9168 * optc-save-gen.awk: Likewise.
9169 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
9170 * godump.c: Remove include of wide-int-print.h.
9171 * pretty-print.h: Likewise.
9172 * wide-int-print.cc: Likewise.
9173 * wide-int.cc: Likewise.
9174 * hash-map-tests.c: Remove include of signop.h.
9175 * hash-set-tests.c: Likewise.
9176 * rtl-tests.c: Likewise.
9177 * mkconfig.sh: Remove include of machmode.h.
9178 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
9180 (emit_insn_modes_inline_h): ...this new function. Emit the code
9181 into an insn-modes-inline.h header file, adding appropriate
9182 include guards and end comments.
9183 (emit_insn_modes_c_header): Remove include of machmode.h.
9184 (emit_min_insn_modes_c_header): Include coretypes.h rather than
9186 (main): Handle -i flag and call emit_insn_modes_inline_h when
9189 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9191 * tree-ssa-strlen.c (strinfo): Rename the length field to
9192 nonzero_chars. Add a full_string_p field.
9193 (compare_nonzero_chars, zero_length_string_p): New functions.
9194 (get_addr_stridx): Add an offset_out parameter.
9195 Use compare_nonzero_chars.
9196 (get_stridx): Update accordingly. Use compare_nonzero_chars.
9197 (new_strinfo): Update after above changes to strinfo.
9198 (set_endptr_and_length): Set full_string_p.
9199 (get_string_length): Update after above changes to strinfo.
9200 (unshare_strinfo): Update call to new_strinfo.
9201 (maybe_invalidate): Likewise.
9202 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
9203 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
9204 as a uhwi instead of an shwi. Update after above changes to
9205 strinfo and new_strinfo.
9206 (zero_length_string): Assert that chainsi contains full strings.
9207 Use zero_length_string_p. Update call to new_strinfo.
9208 (adjust_related_strinfos): Update after above changes to strinfo.
9209 Copy full_string_p from origsi.
9210 (adjust_last_stmt): Use zero_length_string_p.
9211 (handle_builtin_strlen): Update after above changes to strinfo and
9212 new_strinfo. Install the lhs as the string length if the previous
9213 entry didn't describe a full string.
9214 (handle_builtin_strchr): Update after above changes to strinfo
9216 (handle_builtin_strcpy): Likewise.
9217 (handle_builtin_strcat): Likewise.
9218 (handle_builtin_malloc): Likewise.
9219 (handle_pointer_plus): Likewise.
9220 (handle_builtin_memcpy): Likewise. Track nonzero characters
9221 that aren't necessarily followed by a nul terminator.
9222 (handle_char_store): Likewise.
9224 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9226 PR tree-optimization/80769
9227 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
9228 for malloc and calloc. Document the new invariant that all related
9229 strinfos have delayed lengths or none do.
9230 (verify_related_strinfos): Move earlier in file.
9231 (set_endptr_and_length): New function, split out from...
9232 (get_string_length): ...here. Also set the lengths of related
9234 (zero_length_string): Assert that chainsi has known (rather than
9236 (adjust_related_strinfos): Likewise.
9238 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
9240 PR tree-optimization/81136
9241 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
9242 assert that two references with the same misalignment have the same
9243 compile-time misalignment if those compile-time misalignments
9246 2017-07-01 Andi Kleen <ak@linux.intel.com>
9248 * print-tree.c (print_node): Print all attributes.
9250 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9252 * cfg.c (scale_bbs_frequencies): New function.
9253 * cfg.h (scale_bbs_frequencies): Declare it.
9254 * cfgloopanal.c (single_likely_exit): Cleanup.
9255 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
9257 (scale_loop_profile): Likewise.
9258 (loop_version): Likewise.
9259 (create_empty_loop_on_edge): Update.
9260 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
9261 scale_loop_frequencies, scale_loop_profile, loopify,
9262 loop_version): Update prototypes.
9263 * modulo-sched.c (sms_schedule): Update.
9264 * predict.c (unlikely_executed_edge_p): Also check probability.
9265 (probably_never_executed_edge_p): Fix typo.
9266 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9267 * tree-parloops.c (gen_parallel_loop): Update.
9268 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
9269 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9270 * tree-ssa-loop-split.c (split_loop): Update.
9271 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9272 * tree-vect-loop-manip.c (vect_do_peeling): Update.
9273 (vect_loop_versioning): Update.
9274 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9276 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9278 * trans-mem.c (split_bb_make_tm_edge): Update profile.
9280 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9282 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
9283 to keep profile consistent.
9285 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9287 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
9288 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
9289 * profile-count.h (max_safe_multiplier): Make unsigned.
9290 (profile_count::guessed_zero): New.
9292 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9294 * bb-reorder.c (fix_up_crossing_landing_pad,
9295 fix_crossing_conditional_branches): Use make_single_succ_edge
9296 to keep profile consistent.
9298 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
9300 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
9303 2017-07-01 Jakub Jelinek <jakub@redhat.com>
9306 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
9307 the right scopes, make sure cond_jump isn't preserved between multiple
9308 iterations. Search for fallthru edge whenever there are 3+ edges and
9309 use find_fallthru_edge for it.
9311 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9313 Patch by Alexander Monakov <amonakov@ispras.ru>
9314 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
9315 probabilities consistently.
9317 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9319 * pa.c (pa_expand_compare_and_swap_loop): Update call of
9320 emit_cmp_and_jump_insns.
9322 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9325 * tree-inline.c (expand_call_inline): Combine profile statuses.
9327 2017-06-30 Andrew Pinski <apinski@cavium.com>
9329 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
9330 fold_stmt returned true.
9332 2017-06-30 Nathan Sidwell <nathan@acm.org>
9334 * ggc.h (empty_string): Delete.
9335 * cfgexpand.c (expand_asm_stmt): Use plain "".
9336 * optabs.c (expand_asm_memory_barrier): Likewise.
9337 * stringpool.c (empty_string): Delete.
9338 (digit_vector, digit_string): Delete.
9339 (ggc_alloc_string): Use plain "", don't optimize single digit
9340 strings. Use ggc_alloc_atomic.
9342 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
9344 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
9345 comparison set and one other set, use the cost of the non-comparison
9348 2017-06-30 Nathan Sidwell <nathan@acm.org>
9350 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
9353 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
9355 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
9356 loops. Remove now unneeded calls to gimple_switch_set_label() that
9357 just set removed labels to NULL_TREE.
9359 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
9361 * tree-ssanames.c (set_range_info_raw): Abstract from ...
9362 (set_range_info): ...here. Only call set_range_info_raw if domain
9364 (set_nonzero_bits): Call set_range_info_raw.
9365 * tree-ssanames.h (set_range_info_raw): New.
9367 2017-06-30 Jakub Jelinek <jakub@redhat.com>
9370 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
9371 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
9372 of nonimmediate_operand and <store_mask_constraint> instead of m
9373 for the input operand. For V8FI iterator, always split if input
9374 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
9375 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
9376 <store_mask_predicate> instead of register_operand and
9377 <store_mask_constraint> instead of v for the input operand. Make
9378 sure both operands aren't MEMs for if not <mask_applied>.
9380 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
9382 * lto-wrapper.c (copy_file) Close both file descriptors before
9385 2017-06-30 Martin Liska <mliska@suse.cz>
9388 * multiple_target.c (create_dispatcher_calls): Make ifunc
9389 also for function that don't have calls or are not referenced.
9391 2017-06-30 Richard Biener <rguenther@suse.de>
9393 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
9394 analyze the first scalar stmt. Move vector type computation
9395 for the BB case here from ...
9396 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
9397 live operation processing in the SLP case properly.
9399 2017-06-30 Richard Biener <rguenther@suse.de>
9401 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
9403 2017-06-30 Martin Liska <mliska@suse.cz>
9406 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
9407 before BUILT_IN_UNWIND_RESUME when ASAN is used.
9409 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
9411 * doc/invoke.texi (AArch64): Add missing options and remove redundant
9414 2017-06-30 Richard Biener <rguenther@suse.de>
9416 PR tree-optimization/81249
9417 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
9418 condition reduction result to original scalar type.
9420 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9422 * profile-count.h (enum profile_quality): Fix typos and whitespace
9425 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9427 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
9428 type for branch probabilities.
9430 2017-06-29 Julian Brown <julian@codesourcery.com>
9431 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9433 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
9434 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
9435 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
9436 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
9438 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9440 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
9441 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
9442 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
9443 CC usage from generic code to here.
9444 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
9445 CC usage into the target macros.
9447 2017-06-29 Maya Rashish <coypu@sdf.org>
9449 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
9452 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9454 * arm/arm-builtins.c: Include profile-count.h
9455 * except.c (sjlj_emit_function_enter): Use
9456 profile_probability::unlikely.
9458 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
9460 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
9461 and tocrel_offset be pointer args rather than implicitly using
9463 (legitimate_constant_pool_address_p, rs6000_emit_move,
9464 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
9465 tocrel_offset and use in toc_relative_expr_p call.
9466 (print_operand, print_operand_address): Use static tocrel_base_oac
9467 and tocrel_offset_oac.
9468 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
9471 2017-06-29 Maya Rashish <coypu@sdf.org>
9473 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
9475 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
9477 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
9478 objects, take into account only the alignment of 'op0' and 'mode1' if
9481 2017-06-29 Steve Ellcey <sellcey@cavium.com>
9483 * ccmp.c (ccmp_tree_comparison_p): New function.
9484 (ccmp_candidate_p): Update to use above function.
9485 (get_compare_parts): New function.
9486 (expand_ccmp_next): Update to use new functions.
9487 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
9489 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
9490 take mode as argument.
9491 * ccmp.h (expand_ccmp_expr): Add mode as argument.
9492 * expr.c (expand_expr_real_1): Pass mode as argument.
9494 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
9496 * combine.c (combine_instructions): Print insns to dump_file, together
9499 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
9501 * asan.c (asan_emit_stack_protection): Update.
9502 (create_cond_insert_point): Update.
9503 * auto-profile.c (afdo_propagate_circuit): Update.
9504 * basic-block.h (struct edge_def): Turn probability to
9505 profile_probability.
9506 (EDGE_FREQUENCY): Update.
9507 * bb-reorder.c (find_traces_1_round): Update.
9508 (better_edge_p): Update.
9509 (sanitize_hot_paths): Update.
9510 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
9511 (make_single_succ_edge): Update.
9512 (check_bb_profile): Update.
9513 (dump_edge_info): Update.
9514 (update_bb_profile_for_threading): Update.
9515 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
9516 probabilitycount to 0.
9517 * cfgbuild.c (compute_outgoing_frequencies): Update.
9518 * cfgcleanup.c (try_forward_edges): Update.
9519 (outgoing_edges_match): Update.
9520 (try_crossjump_to_edge): Update.
9521 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
9522 (expand_gimple_tailcall): Update.
9523 (construct_init_block): Use make_single_succ_edge.
9524 (construct_exit_block): Use make_single_succ_edge.
9525 * cfghooks.c (verify_flow_info): Update.
9526 (redirect_edge_succ_nodup): Update.
9527 (split_edge): Update.
9528 (account_profile_record): Update.
9529 * cfgloopanal.c (single_likely_exit): Update.
9530 * cfgloopmanip.c (scale_loop_profile): Update.
9531 (set_zero_probability): Remove.
9532 (duplicate_loop_to_header_edge): Update.
9533 * cfgloopmanip.h (loop_version): Update prototype.
9534 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
9535 (force_nonfallthru_and_redirect): Update.
9536 (update_br_prob_note): Update.
9537 (rtl_verify_edges): Update.
9538 (purge_dead_edges): Update.
9539 (rtl_lv_add_condition_to_bb): Update.
9540 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
9541 * cgraphunit.c (init_lowered_empty_function): Update.
9542 (cgraph_node::expand_thunk): Update.
9543 * cilk-common.c: Include profile-count.h
9544 * dojump.c (inv): Remove.
9545 (jumpifnot): Update.
9546 (jumpifnot_1): Update.
9547 (do_jump_1): Update.
9549 (do_jump_by_parts_greater_rtx): Update.
9550 (do_compare_rtx_and_jump): Update.
9551 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
9552 do_jump_1. do_compare_rtx_and_jump): Update prototype.
9553 * dwarf2cfi.c: Include profile-count.h
9554 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
9555 (sjlj_emit_dispatch_table): Likewise.
9556 * explow.c: Include profile-count.h
9557 * expmed.c (emit_store_flag_force): Update.
9558 (do_cmp_and_jump): Update.
9559 * expr.c (compare_by_pieces_d::generate): Update.
9560 (compare_by_pieces_d::finish_mode): Update.
9561 (emit_block_move_via_loop): Update.
9562 (store_expr_with_bounds): Update.
9563 (store_constructor): Update.
9564 (expand_expr_real_2): Update.
9565 (expand_expr_real_1): Update.
9566 * expr.h (try_casesi, try_tablejump): Update prototypes.
9567 * gimple-pretty-print.c (dump_probability): Update.
9568 (dump_profile): New.
9569 (dump_gimple_label): Update.
9570 (dump_gimple_bb_header): Update.
9571 * graph.c (draw_cfg_node_succ_edges): Update.
9572 * hsa-gen.c (convert_switch_statements): Update.
9573 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
9574 (find_if_case_1): Update.
9575 (find_if_case_2): Update.
9576 * internal-fn.c (expand_arith_overflow_result_store): Update.
9577 (expand_addsub_overflow): Update.
9578 (expand_neg_overflow): Update.
9579 (expand_mul_overflow): Update.
9580 (expand_vector_ubsan_overflow): Update.
9581 * ipa-cp.c (good_cloning_opportunity_p): Update.
9582 * ipa-split.c (split_function): Use make_single_succ_edge.
9583 * ipa-utils.c (ipa_merge_profiles): Update.
9584 * loop-doloop.c (add_test): Update.
9585 (doloop_modify): Update.
9586 * loop-unroll.c (compare_and_jump_seq): Update.
9587 (unroll_loop_runtime_iterations): Update.
9588 * lra-constraints.c (lra_inheritance): Update.
9589 * lto-streamer-in.c (input_cfg): Update.
9590 * lto-streamer-out.c (output_cfg): Update.
9591 * mcf.c (adjust_cfg_counts): Update.
9592 * modulo-sched.c (sms_schedule): Update.
9593 * omp-expand.c (expand_omp_for_init_counts): Update.
9594 (extract_omp_for_update_vars): Update.
9595 (expand_omp_ordered_sink): Update.
9596 (expand_omp_for_ordered_loops): Update.
9597 (expand_omp_for_generic): Update.
9598 (expand_omp_for_static_nochunk): Update.
9599 (expand_omp_for_static_chunk): Update.
9600 (expand_cilk_for): Update.
9601 (expand_omp_simd): Update.
9602 (expand_omp_taskloop_for_outer): Update.
9603 (expand_omp_taskloop_for_inner): Update.
9604 * omp-simd-clone.c (simd_clone_adjust): Update.
9605 * optabs.c (expand_doubleword_shift): Update.
9606 (expand_abs): Update.
9607 (emit_cmp_and_jump_insn_1): Update.
9608 (expand_compare_and_swap_loop): Update.
9609 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
9610 * predict.c (predictable_edge_p): Update.
9611 (edge_probability_reliable_p): Update.
9612 (set_even_probabilities): Update.
9613 (combine_predictions_for_insn): Update.
9614 (combine_predictions_for_bb): Update.
9615 (propagate_freq): Update.
9616 (estimate_bb_frequencies): Update.
9617 (force_edge_cold): Update.
9618 * profile-count.c (profile_count::dump): Add missing space into dump.
9619 (profile_count::debug): Add newline.
9620 (profile_count::differs_from_p): Explicitly convert to unsigned.
9621 (profile_count::stream_in): Update.
9622 (profile_probability::dump): New member function.
9623 (profile_probability::debug): New member function.
9624 (profile_probability::differs_from_p): New member function.
9625 (profile_probability::differs_lot_from_p): New member function.
9626 (profile_probability::stream_in): New member function.
9627 (profile_probability::stream_out): New member function.
9628 * profile-count.h (profile_count_quality): Rename to ...
9629 (profile_quality): ... this one.
9630 (profile_probability): New.
9631 (profile_count): Update.
9632 * profile.c (compute_branch_probabilities): Update.
9633 * recog.c (peep2_attempt): Update.
9634 * sched-ebb.c (schedule_ebbs): Update.
9635 * sched-rgn.c (find_single_block_region): Update.
9636 (compute_dom_prob_ps): Update.
9637 (schedule_region): Update.
9638 * sel-sched-ir.c (compute_succs_info): Update.
9639 * stmt.c (struct case_node): Update.
9640 (do_jump_if_equal): Update.
9641 (get_outgoing_edge_probs): Update.
9642 (conditional_probability): Update.
9643 (emit_case_dispatch_table): Update.
9644 (expand_case): Update.
9645 (expand_sjlj_dispatch_table): Update.
9646 (emit_case_nodes): Update.
9647 * targhooks.c: Update.
9648 * tracer.c (better_p): Update.
9649 (find_best_successor): Update.
9650 * trans-mem.c (expand_transaction): Update.
9651 * tree-call-cdce.c: Update.
9652 * tree-cfg.c (gimple_split_edge): Upate.
9653 (move_sese_region_to_fn): Upate.
9654 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
9655 * tree-eh.c (lower_resx): Upate.
9656 (cleanup_empty_eh_move_lp): Upate.
9657 * tree-if-conv.c (version_loop_for_if_conversion): Update.
9658 * tree-inline.c (copy_edges_for_bb): Update.
9659 (copy_cfg_body): Update.
9660 * tree-parloops.c (gen_parallel_loop): Update.
9661 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
9662 (gimple_gen_time_profiler): Update.
9663 * tree-ssa-dce.c (remove_dead_stmt): Update.
9664 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
9665 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
9666 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
9667 (unloop_loops): Update.
9668 (try_peel_loop): Update.
9669 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
9670 * tree-ssa-loop-split.c (connect_loops): Update.
9671 (split_loop): Update.
9672 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
9673 (hoist_guard): Update.
9674 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
9675 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
9676 (value_replacement): Update.
9677 * tree-ssa-reassoc.c (branch_fixup): Update.
9678 * tree-ssa-tail-merge.c (replace_block_by): Update.
9679 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
9680 (create_edge_and_update_destination_phis): Update.
9681 (compute_path_counts): Update.
9682 (recompute_probabilities): Update.
9683 (update_joiner_offpath_counts): Update.
9684 (freqs_to_counts_path): Update.
9685 (duplicate_thread_path): Update.
9686 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
9687 (struct switch_conv_info): Update.
9688 (gen_inbound_check): Update.
9689 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
9690 (vect_do_peeling): Update.
9691 (vect_loop_versioning): Update.
9692 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
9693 (optimize_mask_stores): Update.
9694 * ubsan.c (ubsan_expand_null_ifn): Update.
9695 * value-prof.c (gimple_divmod_fixed_value): Update.
9696 (gimple_divmod_fixed_value_transform): Update.
9697 (gimple_mod_pow2): Update.
9698 (gimple_mod_pow2_value_transform): Update.
9699 (gimple_mod_subtract): Update.
9700 (gimple_mod_subtract_transform): Update.
9701 (gimple_ic): Update.
9702 (gimple_stringop_fixed_value): Update.
9703 (gimple_stringops_transform): Update.
9704 * value-prof.h: Update.
9706 2017-06-29 Carl Love <cel@us.ibm.com>
9708 * config/rs6000/rs6000-c.c: Add support for built-in functions
9709 vector signed int vec_signed (vector float);
9710 vector signed long long vec_signed (vector double);
9711 vector signed int vec_signed2 (vector double, vector double);
9712 vector signed int vec_signede (vector double);
9713 vector signed int vec_signedo (vector double);
9714 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
9715 instruction generator.
9716 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
9717 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
9718 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
9720 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
9721 vunsignede_v2df): Add define_expands.
9722 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
9723 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
9724 VEC_UNSIGNEDO): Add definitions.
9725 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
9726 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
9727 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
9728 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
9729 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
9730 * config/rs6000/altivec.h (vec_signed, vec_signed2,
9731 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
9732 vec_unsignede, vec_unsignedo): Add builtin defines.
9733 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
9735 * doc/extend.texi: Update the built-in documentation file for the
9736 new built-in functions.
9738 2017-06-29 Richard Biener <rguenther@suse.de>
9740 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
9741 reduction chains to LOOP_VINFO_REDUCTIONS.
9742 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
9743 SLP reductions after processing reduction chains.
9745 2017-06-29 Nathan Sidwell <nathan@acm.org>
9747 * builtins.c (fold_builtin_FUNCTION): Use
9748 lang_hooks.decl_printable_name.
9750 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
9753 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
9754 with only one label.
9755 * stmt.c (expand_case): Assert NCASES is greater than one.
9757 2017-06-29 Richard Biener <rguenther@suse.de>
9759 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
9761 (group_case_labels): Likewise.
9762 (find_taken_edge): Push sanity checking on val to workers...
9763 (find_taken_edge_cond_expr): ... here
9764 (find_taken_edge_switch_expr): ... and here, handle cases
9765 with just a default label.
9766 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
9767 (group_case_labels): Likewise.
9768 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
9769 group_case_labels does anything cleanup the CFG again.
9771 2017-06-29 Bin Cheng <bin.cheng@arm.com>
9773 PR tree-optimization/81196
9774 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
9775 exit condition comparing two IVs.
9777 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
9779 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
9780 profile to the dummy entry at the end of the list of architectures.
9781 * config/arm/arm-cpu-cdata.h: Regenerated.
9783 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9784 Michael Collison <michael.collison@arm.com>
9787 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
9789 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
9790 (*aarch64_reg_<mode>3_minus_mask): New pattern.
9791 (*aarch64_<optab>_reg_di3_mask2): New pattern.
9792 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
9793 of shift when the shift amount is masked with constant equal to
9794 the size of the mode.
9795 * config/aarch64/predicates.md (subreg_lowpart_operator): New
9798 2017-06-29 Martin Liska <mliska@suse.cz>
9800 * config/i386/i386.opt: Change range from [1,5] to [0,5].
9802 2017-06-29 Yury Gribov <tetra2005@gmail.com>
9805 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
9808 (edge_growth_cache_entry::edge_growth_cache_entry): New
9810 (reset_edge_growth_cache): Update to use constructor.
9812 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9814 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
9815 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
9816 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
9818 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
9820 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
9821 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
9823 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
9825 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
9826 (*-linux-uclibc*): Add t-uclibc tmake_file.
9827 * config/t-musl: New.
9828 * config/t-uclibc: New.
9830 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
9832 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
9834 (gen_comm_data): Emit architectural setting of arch_prof.
9835 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
9837 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
9838 (armv8-m.base, armv8-m.main): Likewise.
9839 * arm-protos.h (arm_build_target): Add profile field.
9840 (arch_option): Likewise.
9841 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
9843 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
9844 arm_active_target.profile.
9846 2017-06-28 Richard Biener <rguenther@suse.de>
9849 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
9850 TYPE_OVERFLOW_WRAPS.
9851 * match.pd (negate_expr_p): Likewise.
9852 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
9853 fold_build2, not fold_binary.
9855 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
9857 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
9858 Convert memory address to Pmode.
9859 (aarch64_print_operand): Assert MEM operands are always Pmode.
9861 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
9864 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
9865 Remove redundant if.
9866 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
9867 * config/arm/aarch-common-protos.h
9868 (aarch_forward_to_shift_is_not_shifted_re): Remove.
9869 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
9871 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
9874 * multiple_target.c (create_dispatcher_calls): Set the default
9875 clone to be static, not public.
9877 2017-06-28 Richard Biener <rguenther@suse.de>
9879 * tree-vect-loop.c (vectorizable_reduction): Move special
9880 cond reduction IV var creation ...
9881 (vect_create_epilog_for_reduction): ... here. Remove induction_index
9882 parameter. Use STMT_VINFO_VECTYPE.
9883 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
9886 2017-06-28 Martin Liska <mliska@suse.cz>
9889 * ipa-visibility.c (non_local_p): Handle visibility.
9891 2017-06-28 Martin Liska <mliska@suse.cz>
9894 * common.opt: Add IntegerRange to various options.
9895 * opt-functions.awk (integer_range_info): New function.
9896 * optc-gen.awk: Add integer_range_info to cl_options struct.
9897 * opts-common.c (decode_cmdline_option): Handle
9898 CL_ERR_INT_RANGE_ARG.
9899 (cmdline_handle_error): Likewise.
9900 * opts.c (print_filtered_help): Show valid interval in
9901 when --help is provided.
9902 * opts.h (struct cl_option): Add range_min and range_max fields.
9903 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
9905 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
9907 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
9908 (x * C EQ/NE y * C): New transformation.
9910 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
9912 * genmultilib (combination_space): Accept '+' in option names.
9914 2017-06-28 Martin Liska <mliska@suse.cz>
9917 * asan.c (instrument_derefs): Bail out inner references
9918 that are hard register variables.
9920 2017-06-28 Jakub Jelinek <jakub@redhat.com>
9923 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
9924 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
9926 2017-06-28 Richard Biener <rguenther@suse.de>
9928 * tree-vectorizer.h (vect_get_vec_defs): Remove.
9929 (vect_get_slp_defs): Adjust.
9930 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
9932 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
9934 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
9935 get_initial_defs_for_reduction instead of vect_get_vec_defs.
9936 (vectorizable_reduction): Adjust.
9937 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
9939 (vect_get_slp_defs): Likewise.
9940 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
9941 (vectorizable_bswap): Adjust.
9942 (vectorizable_call): Likewise.
9943 (vectorizable_conversion): Likewise.
9944 (vectorizable_assignment): Likewise.
9945 (vectorizable_shift): Likewise.
9946 (vectorizable_operation): Likewise.
9947 (vectorizable_store): Likewise.
9948 (vectorizable_condition): Likewise.
9949 (vectorizable_comparison): Likewise.
9951 2017-06-28 Michael Collison <michael.collison@arm.com>
9954 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
9956 (arm_gen_movmemqi): Removed unused variable 'i'.
9957 Convert 'for' loop into 'while' loop.
9958 (arm_expand_prologue): Remove last unnecessary set of insn.
9959 (thumb_pop): Remove unused variable 'pushed_words'.
9960 (thumb_exit): Remove last unnecessary set of regs_to_pop.
9962 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9964 * config/s390/predicates.md: Use s390_rel_address_ok_p.
9965 * config/s390/s390-protos.h: Add prototype of
9966 s390_rel_address_ok_p.
9967 * config/s390/s390.c (s390_got_symbol): New function.
9968 (s390_rel_address_ok_p): New function.
9969 (legitimize_pic_address): Use s390_rel_address_ok_p.
9970 (s390_load_got): Use s390_got_symbol.
9971 (s390_option_override): Issue error if
9972 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
9973 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
9975 * config/s390/s390.opt: New option mpic-data-is-text-relative.
9977 2017-06-27 Andrew Pinski <apinski@cavium.com>
9979 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
9980 (X * copysign (1.0, X)): New pattern.
9981 (X * copysign (1.0, -X)): New pattern.
9982 (copysign (-1.0, CST)): New pattern.
9984 2017-06-27 Joseph Myers <joseph@codesourcery.com>
9986 * genmultilib (combination_space): Remove variable.
9987 Validate reuse rules against regular expression for any sequence
9988 of multilib options in any order.
9990 2017-06-27 Michael Collison <michael.collison@arm.com>
9992 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
9993 call aarch64_split_simd_combine.
9994 * (aarch64_combine_internal<mode>): Delete pattern.
9995 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
9996 Allow register and subreg operands.
9998 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10000 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
10001 specific need, just fallback on defaults.
10002 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
10004 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10005 Olivier Hainque <hainque@adacore.com>
10007 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
10009 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
10010 targets. Pick a default if no particular attempt applied.
10011 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
10014 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10016 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
10017 (x86_64-wrs-vxworks7): Likewise.
10019 2017-06-27 Marek Polacek <polacek@redhat.com>
10022 * ubsan.c (instrument_null): Check get_base_address's result for null.
10024 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10026 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
10028 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
10030 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
10031 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
10032 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
10033 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
10034 New function types.
10035 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
10036 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
10037 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
10038 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
10039 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
10040 BUILT_IN_FEUPDATEENV): New builtins.
10041 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
10042 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
10043 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
10044 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
10046 (builtin_structptr_types): Adjust size.
10047 * tree.c (builtin_structptr_types): Add four entries.
10049 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10050 Olivier Hainque <hainque@adacore.com>
10052 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
10053 (TLS_SYM): New local macro, forcing reference to __tls__ on
10054 link command lines for VxWorks 7 RTPs, triggering initialization
10056 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
10057 OS features TLS support, true for RTPs on VxWorks 7.
10058 * config/vxworks.c (vxworks_override_options): Setup emutls
10061 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10063 * predict.c (test_prediction_value_range): Use -1U instead of -1
10064 to avoid narrowing conversion warning.
10065 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
10066 to avoid narrowing conversion warning.
10067 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
10069 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
10071 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
10073 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
10074 64bit configurations.
10075 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
10076 (SIZE_TYPE): Likewise.
10077 * config/vxworks.c (vxworks_emutls_var_fields): Use
10078 long_unsigned_type_node instead of unsigned_type_node as the offset
10079 field type, which is "pointer" mode in emutls.c.
10081 2017-06-27 Jakub Jelinek <jakub@redhat.com>
10084 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
10086 PR middle-end/81207
10087 * gimple-fold.c (replace_call_with_call_and_fold): Handle
10088 gimple_vuse copying separately from gimple_vdef copying.
10090 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10092 * value-prof.c (free_hist): Remove call to memset and the enclosing if
10095 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10096 Olivier Hainque <hainque@adacore.com>
10098 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
10099 for all vxworks7 targets.
10100 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
10101 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
10102 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
10103 variations for VX6/VX7 and 32/64bits later on in ...
10104 (VXWORKS_LIB_SPEC): Leverage new macros.
10105 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
10106 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
10108 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
10110 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
10111 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
10113 2017-06-26 Carl Love <cel@us.ibm.com>
10115 * config/rs6000/rs6000-c.c: Add support for built-in functions
10116 vector bool char vec_reve (vector bool char);
10117 vector signed char vec_reve (vector signed char);
10118 vector unsigned char vec_reve (vector unsigned char);
10119 vector bool int vec_reve (vector bool int);
10120 vector signed int vec_reve (vector signed int);
10121 vector unsigned int vec_reve (vector unsigned int);
10122 vector bool long long vec_reve (vector bool long long);
10123 vector signed long long vec_reve (vector signed long long);
10124 vector unsigned long long vec_reve (vector unsigned long long);
10125 vector bool short vec_reve (vector bool short);
10126 vector signed short vec_reve (vector signed short);
10127 vector double vec_reve (vector double);
10128 vector float vec_reve (vector float);
10129 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
10130 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
10131 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
10132 (altivec_vreve): New pattern.
10133 * config/rs6000/altivec.h (vec_reve): New define.
10134 * doc/extend.texi (vec_rev): Update the built-in documentation file
10135 for the new built-in functions.
10137 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10139 PR tree-optimization/71815
10140 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
10142 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
10144 (slsr_process_phi): Likewise.
10145 (replace_uncond_cands_and_profitable_phis): Don't replace a
10146 multiply candidate with a stride of 1 (copy or cast).
10147 (phi_incr_cost): Call uses_consumed_by_stmt rather than
10149 (lowest_cost_path): Likewise.
10150 (total_savings): Likewise.
10152 2017-06-26 Richard Biener <rguenther@suse.de>
10155 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
10156 Use def_builtin_pure for all gather builtins.
10158 2017-06-26 Richard Biener <rguenther@suse.de>
10160 PR tree-optimization/81203
10161 * tree-tailcall.c (find_tail_calls): Do not move stmts into
10162 non-dominating BBs.
10164 2017-06-26 Marek Polacek <polacek@redhat.com>
10167 * doc/invoke.texi: Document -Wmultistatement-macros.
10169 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
10171 * doc/sourcebuild.texi (ARM-specific attributes): Document new
10172 arm_neon_ok_no_float_abi effective target.
10174 2017-06-26 Richard Biener <rguenther@suse.de>
10176 PR tree-optimization/80928
10177 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
10178 (copy_bbs): Set BB_DUPLICATED flag early.
10179 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
10181 (execute_on_shrinking_pred): Likewise.
10182 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
10183 BB_DUPLICATED blocks.
10184 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
10185 iterate over all PHIs considering removal of *gsi.
10187 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
10189 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
10192 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10194 * config/rs6000/rs6000-string.c: (expand_block_clear,
10195 do_load_for_compare, select_block_compare_mode,
10196 compute_current_alignment, expand_block_compare,
10197 expand_strncmp_align_check, expand_strn_compare,
10198 expand_block_move, rs6000_output_load_multiple)
10199 Move functions related to string/block move/compare
10200 to a separate file.
10201 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
10202 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
10203 for this function which is now used in two files.
10204 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
10205 * config.gcc: Add rs6000-string.o to extra_objs for
10206 targets powerpc*-*-* and rs6000*-*-*.
10208 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
10211 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
10212 32-bit, since indexed is not valid for DImode.
10213 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
10214 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
10215 (define_peephole2 for Altivec d-form load): Add 32-bit support.
10216 (define_peephole2 for Altivec d-form store): Likewise.
10219 * multiple_target.c (create_dispatcher_calls): Only create the
10220 dispatcher call if the function is the default clone of a
10221 versioned function.
10223 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
10225 PR middle-end/80902
10226 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
10227 a call, force the call to not be a tail call.
10229 2017-06-23 Jeff Law <law@redhat.com>
10231 * doc/contrib.texi: Add entry for Steven Pemberton's work on
10234 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
10236 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
10237 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
10238 handling for early expansion of vector shifts (sl,sr,sra,rl).
10239 (builtin_function_type): Add vector shift right instructions
10240 to the unsigned argument list.
10242 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
10244 rtl-optimizatoin/79286
10245 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
10246 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
10247 trap. PIC register plus a const unspec without offset can never trap.
10249 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
10251 * tree.h (builtin_structptr_type): New type.
10252 (builtin_structptr_types): Declare new array.
10253 * tree.c (builtin_structptr_types): New array.
10254 (free_lang_data, build_common_tree_nodes): Use it.
10256 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
10259 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
10262 2017-06-22 Matt Turner <mattst88@gmail.com>
10264 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
10265 Lake models to skylake case. Assume skylake for unknown
10266 models with clflushopt.
10268 2017-06-22 Jeff Law <law@redhat.com>
10270 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
10271 frame sizes that do not satisfy aarch64_uimm12_shift.
10273 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
10275 * profile-count.h (apply_probability,
10276 apply_scale, probability_in): Fix checks for zero.
10278 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10280 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
10281 * doc/cppdiropts.texi (-I @var{dir}): Document it.
10283 2016-06-22 Richard Biener <rguenther@suse.de>
10285 * tree-vect-loop.c (vect_model_reduction_cost): Handle
10286 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
10287 REDUC_MAX_EXPR support.
10288 (vectorizable_reduction): Likewise.
10289 (vect_create_epilog_for_reduction): Likewise.
10291 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
10293 * match.pd (A / (1 << B) -> A >> B): New.
10294 * generic-match-head.c: Include optabs-tree.h.
10295 * gimple-match-head.c: Likewise.
10296 * optabs-tree.h (target_supports_op_p): New.
10297 * optabs-tree.c (target_supports_op_p): New.
10299 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
10301 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
10302 $gcc_cv_ld --help output.
10303 (gcc_cv_ld_demangle): Likewise.
10304 (gcc_cv_ld_eh_frame_hdr): Likewise.
10305 (gcc_cv_ld_pie): Likewise.
10306 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
10307 (gcc_cv_ld_buildid): Likewise.
10308 (gcc_cv_ld_sysroot): Likewise.
10309 (ld_bndplt_support): Likewise.
10310 (ld_pushpopstate_support): Likewise.
10311 * configure: Regenerate.
10312 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
10314 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10317 * config/i386/sse.md (round<mode>2): Renumber match_dup and
10318 operands indexes to avoid gap between operands and match_dups.
10320 2017-06-21 Andrew Pinski <apinski@cavium.com>
10322 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
10323 Increment Arith_shift and Arith_shift_reg by 1.
10324 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
10326 * config/aarch64/aarch64.c (thunderx_tunings): Enable
10327 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
10328 (aarch64_strip_extend): Add new argument and test for it.
10329 (aarch64_cheap_mult_shift_p): New function.
10330 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
10331 add a cost if it is true.
10332 Update calls to aarch64_strip_extend.
10333 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
10335 2017-06-21 Andrew Pinski <apinski@cavium.com>
10337 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
10339 (thunderxt88): Likewise.
10340 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
10341 (thunderx_prefetch_tune): New variable.
10342 (thunderx2t99_prefetch_tune): Update for the correct values.
10343 (thunderxt88_tunings): New variable.
10344 (thunderx_tunings): Use thunderx_prefetch_tune instead of
10345 generic_prefetch_tune.
10346 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
10348 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10350 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
10351 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
10352 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
10353 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
10354 (aarch64_atomic_cas<mode>, GPI): Likewise.
10356 2017-06-21 Martin Liska <mliska@suse.cz>
10358 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
10359 statements on cold and hot labels.
10360 * predict.c (tree_estimate_probability_bb): Remove the
10361 prediction from this place.
10363 2017-06-21 Martin Liska <mliska@suse.cz>
10365 PR tree-optimization/79489
10366 * gimplify.c (maybe_add_early_return_predict_stmt): New
10368 (gimplify_return_expr): Call the function.
10369 * predict.c (tree_estimate_probability_bb): Remove handling
10371 * predict.def: Update comment about early return predictor.
10372 * gimple-predict.h (is_gimple_predict): New function.
10373 * predict.def: Change default value of early return to 66.
10374 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
10376 * passes.def: Put pass_strip_predict_hints to the beginning of
10379 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
10381 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
10382 FUNCTION_DECL declarations.
10383 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
10385 (dwaf2out_decl): Likewise.
10386 * godump.c (go_early_global_decl): Skip call to the real debug hook
10387 for FUNCTION_DECL declarations.
10388 * passes.c (rest_of_decl_compilation): Skip call to the
10389 early_global_decl debug hook for FUNCTION_DECL declarations, unless
10390 -fdump-go-spec is passed.
10392 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10394 * config/i386/i386.c (struct builtin_isa): New field pure_p.
10395 Reorder for compactness.
10396 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
10397 (def_builtin_pure, def_builtin_pure2): New functions.
10398 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
10400 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
10402 * match.pd (nop_convert): New predicate.
10403 ((A +- CST1) +- CST2): Allow some NOP conversions.
10405 2017-06-21 Jakub Jelinek <jakub@redhat.com>
10408 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
10409 with ctors/dtors if GOVD_SHARED is set.
10411 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10413 * config/aarch64/aarch64.md (movti_aarch64):
10414 Emit mov rather than orr.
10415 (movtf_aarch64): Likewise.
10416 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
10417 Emit mov rather than orr.
10419 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10421 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
10422 Swap alternatives, make integer dup more expensive.
10424 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
10426 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
10427 Return true for non-tls symbols.
10429 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
10431 * config/aarch64/aarch64-cores.def (cortex-a55): New.
10432 (cortex-a75): Likewise.
10433 (cortex-a75.cortex-a55): Likewise.
10434 * config/aarch64/aarch64-tune.md: Regenerate.
10435 * doc/invoke.texi (-mtune): Document new values for -mtune.
10437 2017-06-21 Tom de Vries <tom@codesourcery.com>
10439 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
10440 stack_size feature.
10441 (Effective-Target Keywords, Other attributes): Suggest using
10442 dg-add-options stack_size feature to get stack limit in stack_size
10443 effective target documentation.
10445 2017-06-21 Julian Brown <julian@codesourcery.com>
10446 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
10448 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
10449 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
10450 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
10452 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
10453 attribute type list for neon_multiply.
10454 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
10455 attribute type list for neon_multiply.
10456 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
10457 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
10458 attribute type list for neon_multiply.
10459 * config/arm/types.md (crypto_pmull): Add.
10460 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
10461 attribute type list.
10463 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
10465 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
10468 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10470 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
10471 to make sure not to dereference a NULL cost_classes_ptr pointer.
10473 2017-06-20 Carl Love <cel@us.ibm.com>
10475 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10476 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
10477 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
10478 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
10479 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
10480 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
10481 VMULOSW): New enum "unspec" values.
10482 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
10483 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
10484 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
10485 altivec_vmulosw): New patterns.
10486 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
10487 VMULOSW): Add definitions.
10489 2017-06-20 Julia Koval <julia.koval@intel.com>
10491 * config/i386/i386.c: Fix rounding expand for new pattern.
10492 * config/i386/subst.md: Fix pattern (parallel -> unspec).
10494 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10496 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
10497 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
10499 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10501 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
10504 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
10506 * config/aarch64/aarch64-cores.def: Rearrange to sort by
10507 architecture, then by implementer ID.
10508 * config/aarch64/aarch64-tune.md: Regenerate.
10510 2017-06-20 Richard Biener <rguenther@suse.de>
10512 PR middle-end/81097
10513 * fold-const.c (split_tree): Fold to type before negating.
10515 2017-06-20 David Malcolm <dmalcolm@redhat.com>
10517 * diagnostic-show-locus.c
10518 (selftest::test_fixit_deletion_affecting_newline): New function.
10519 (selftest::diagnostic_show_locus_c_tests): Call it.
10521 2017-06-20 Andreas Schwab <schwab@suse.de>
10524 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
10527 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
10529 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
10530 __ARM_FEATURE_COPROC according to support.
10532 2017-06-20 Jakub Jelinek <jakub@redhat.com>
10534 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10535 Rewritten to avoid overflow for > 32-bit pointers.
10538 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
10539 by removing enum keyword.
10540 (ubsan_type_descriptor): Likewise. Formatting fix.
10543 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
10544 splitter): Require TARGET_SSE2 in the condition.
10546 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
10549 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
10550 for doing vector set of SFmode on ISA 3.0.
10551 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
10552 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
10554 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
10555 SFmode value into a V4SF variable that was extracted from another
10556 V4SF variable without converting the element to double precision
10557 and back to single precision vector format.
10558 (vsx_insert_extract_v4sf_p9_2): Likewise.
10560 2017-06-19 Jakub Jelinek <jakub@redhat.com>
10562 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
10563 in UWHI to avoid undefined overflow.
10566 * ubsan.h (enum ubsan_encode_value_phase): New.
10567 (ubsan_encode_value): Change second argument to
10568 enum ubsan_encode_value_phase with default value of
10569 UBSAN_ENCODE_VALUE_GENERIC.
10570 * ubsan.c (ubsan_encode_value): Change second argument to
10571 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
10572 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
10573 create_tmp_var_raw instead of create_tmp_var and use a
10575 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
10576 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
10577 ubsan_encode_value callers.
10580 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
10581 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
10582 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
10584 2017-06-19 Richard Biener <rguenther@suse.de>
10586 PR middle-end/81118
10587 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
10588 estimates if we changed anything.
10590 2017-06-19 Richard Biener <rguenther@suse.de>
10592 PR tree-optimization/80887
10593 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
10594 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
10595 simplified lookups, then reset mprts_hook.
10596 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
10598 (try_to_simplify): Likewise.
10600 2017-06-19 Martin Liska <mliska@suse.cz>
10603 * gimplify.c (gimplify_switch_expr):
10604 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
10606 2017-06-19 Martin Liska <mliska@suse.cz>
10608 * doc/install.texi: Document that PGO runs in 4 stages.
10610 2017-06-19 Martin Liska <mliska@suse.cz>
10613 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
10614 to dispatcher function name.
10615 * multiple_target.c (replace_function_decl): New function.
10616 (create_dispatcher_calls): Redirect both edges and references.
10618 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
10620 * profile-count.c (profile_count::dump): Dump quality.
10621 (profile_count::differs_from_p): Update for unsigned val.
10622 * profile-count.h (profile_count_quality): New enum.
10623 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
10625 2017-06-19 Richard Biener <rguenther@suse.de>
10627 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
10628 struct function as arg.
10629 (estimate_numbers_of_iterations): Export overload with loop arg.
10630 (free_numbers_of_iterations_estimates_loop): Use an overload of
10631 free_numbers_of_iterations_estimates instead.
10632 * tree-cfg.c (remove_bb): Adjust.
10633 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
10634 * tree-parloops.c (gen_parallel_loop): Likewise.
10635 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
10637 (tree_unroll_loops_completely): Likewise.
10638 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
10639 Use an overload instead and export.
10640 (estimated_loop_iterations): Adjust.
10641 (max_loop_iterations): Likewise.
10642 (likely_max_loop_iterations): Likewise.
10643 (estimate_numbers_of_iterations): Take struct function as arg
10645 (loop_exits_before_overflow): Adjust.
10646 (free_numbers_of_iterations_estimates_loop): Use an overload.
10647 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
10648 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
10650 2017-06-19 Richard Biener <rguenther@suse.de>
10653 * ipa-prop.c (find_constructor_constant_at_offset): Handle
10654 RANGE_EXPR conservatively.
10656 2017-06-16 Carl Love <cel@us.ibm.com>
10658 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
10659 definitions for vec_float, vec_float2, vec_floato,
10660 vec_floate built-ins.
10661 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
10662 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
10664 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
10665 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
10666 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
10667 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
10668 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
10669 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
10670 vec_floato): Add builtin defines.
10671 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
10672 Update the built-in documentation file for the new built-in
10675 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10677 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
10678 (mthumb): Mark as the negative of -marm.
10680 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10682 * doc/invoke.texi (ARM Options, -mcpu): Document supported
10684 (ARM Options, -mtune): Document that this accepts the same
10685 extension options as -mcpu.
10686 (ARM Options, -mfpu): Document addition of -mfpu=auto.
10688 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10690 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
10691 permitted extensions.
10693 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10695 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
10696 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
10697 (armv8-m.main): Add option +nodsp.
10698 * config/arm/arm-cpu-cdata.h: Regenerated.
10700 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10702 * config/arm/t-fuchsia: New file.
10703 * config.gcc (arm*-*-fuchsia*): Use it.
10705 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10707 * config/arm/t-symbian: Rewrite for new option infrastructure.
10709 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10711 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
10712 (MULTILIB_REQUIRED): Likewise.
10714 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10716 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
10717 (MULTILIB_RESUE): Likewise.
10718 (MULTILIB_MATCHES): Likewise.
10719 (MULTLIB_REQUIRED): Likewise.
10721 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10723 * config/arm/t-rtems: Rewrite for new option framework.
10725 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10727 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
10728 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
10729 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
10730 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
10731 * config/arm/t-multilib: ... here.
10732 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
10733 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
10734 armv7-a and armv8*-a when A-profile libraries have not been built.
10735 * config/arm/t-rmprofile: Rewrite.
10737 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10739 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
10740 with a backslash. Remove the backslash after substituting unescaped
10742 * doc/fragments.texi (MULTILIB_REUSE): Document it.
10744 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10746 * config.gcc: (arm*-*-*): When building a-profile libraries, force
10747 the driver to pass through the default setting of -mfloat-abi.
10748 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
10750 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
10751 (all_feat_combs): New rule.
10752 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
10754 * config/arm/t-aprofile: Rewrite.
10756 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10758 * config/arm/arm.h (FPUTYPE_AUTO): Define.
10759 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
10760 fpu is not specified by the user/command-line.
10761 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
10762 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
10763 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
10764 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
10765 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
10766 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
10768 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10770 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
10771 * config/arm/t-arm-elf: Rewritten.
10773 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10775 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
10776 have some floating-point instructions.
10777 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
10778 (TARGET_MAYBE_HARD_FLOAT): New macro.
10779 * config/arm/arm-builtins.c (arm_init_builtins): Use
10780 TARGET_MAYBE_HARD_FLOAT.
10781 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
10783 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10785 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
10786 (configargs.h): Include it.
10787 (arm_print_hint_for_fpu_option): New function.
10788 (arm_parse_fpu_option): New function.
10789 (candidate_extension): New class.
10790 (arm_canon_for_multilib): New function.
10791 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
10792 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
10793 (ARCH_CANONICAL_SPECS): New macro.
10794 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
10796 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10798 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
10799 are set after handling multilib fragments. Set target_cpu_default2
10802 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10804 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
10806 (arm*-*-*): Set target_cpu_default2 to a quoted string.
10807 * config/arm/parsecpu.awk (check_cpu): Validate any extension
10809 (check_arch): Likewise.
10810 * config/arm/arm.c (arm_configure_build_target): Handle
10811 TARGET_CPU_DEFAULT being a string constant. Scan any feature
10812 options in the default.
10814 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10816 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
10817 when an option is an alias of another.
10818 * config/arm/parsecpu.awk (optalias): New parser token.
10819 (gen_comm_data): Mark non-alias options as such. Emit entries
10820 for extension aliases.
10821 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
10822 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
10823 (armv6kz, armv6zk, armv6t2): Likewise.
10824 (armv7): Make vfpv3-d16 an alias.
10825 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
10827 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
10828 Sort in canonical order.
10829 (armv8-a): Sort in canonical order.
10830 (armv8.1-a, armv8.2-a): Likewise.
10831 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
10833 (cortex-a9): Sort in canonical order.
10834 * config/arm/arm.c (selftests.h): Include it.
10835 (arm_test_cpu_arch_data): New function.
10836 (arm_run_self_tests): New function.
10837 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
10838 (targetm): Move declaration to the end of the file.
10839 * arm-cpu-cdata.h: Regenerated.
10841 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10843 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
10844 call to target_mode_check describing the type of option passed.
10845 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
10846 (arm_target_thumb_only): Use arm_parse_arch_option_name or
10847 arm_parse_cpu_option_name to match parameters against list of
10849 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
10850 arm_arch_core_flags data structure.
10851 * config/arm/arm-cpu_cdata.h: Regenerated.
10853 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10855 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
10857 (arm_print_hint_for_cpu_option): Likewise.
10858 (arm_print_hint_for_arch_option): Likewise.
10859 (arm_parse_cpu_option_name): Likewise.
10860 (arm_parse_arch_option_name): Likewise.
10861 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
10862 of entries in the all_fpus list.
10863 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
10864 (arm_parse_cpu_option_name): Declare.
10865 (arm_parse_arch_option_name): Declare.
10866 (arm_parse_option_features): Declare.
10867 (arm_intialize_isa): Declare.
10868 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
10870 (gen_comm_data): ... here. Make definitions non-static.
10871 * config/arm/arm-cpu-data.h: Regenerated.
10872 * config/arm/arm-cpu-cdata.h: Regenerated.
10874 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10876 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
10877 (cpu_arch_extension): New structure.
10878 (cpu_arch_option, arch_option, cpu_option): New structures.
10879 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
10880 architecture types.
10881 (gen_data): Generate new format data tables.
10882 * config/arm/arm.c (cpu_tune): New structure.
10883 (cpu_option, processors): Delete.
10884 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
10885 (arm_print_hint_for_cpu_option): ... this and ...
10886 (arm_print_hint_for_arch_option): ... this.
10887 (arm_parse_arch_cpu_name): Delete. Replace with ...
10888 (arm_parse_cpu_option_name): ... this and ...
10889 (arm_parse_arch_option_name): ... this.
10890 (arm_unrecognized_feature): Change type of target parameter to
10892 (arm_parse_arch_cpu_features): Delete. Replace with ...
10893 (arm_parse_option_features): ... this.
10894 (arm_configure_build_target): Rework to use new configuration data
10896 (arm_print_tune_info): Rework for new configuration data tables.
10897 * config/arm/arm-cpu-data.h: Regenerated.
10898 * config/arm/arm-cpu.h: Regenerated.
10900 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10902 * Makefile.in (OBJS): Move sbitmap.o from here ...
10903 (OBJS-libcommon): ... to here.
10905 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10907 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
10908 (ISA_ALL_CRYPTO): New macro.
10909 (ISA_ALL_SIMD): New macro
10910 (ISA_ALL_FP): New macro.
10911 * config/arm/arm.c (fpu_bitlist): Update initializer.
10912 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
10914 (arm9e): Add fpu. Add option for nofp
10915 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
10916 (arm926ej-s, arm1026ej-s): Likewise.
10917 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
10918 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
10919 neon-fp16, neon-vfpv4, nofp and nosimd.
10920 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
10921 (cortex-a8): Add fpu. Add option for nofp.
10922 (cortex-a9): Add fpu. Add options for nosimd and nofp.
10923 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
10924 (cortex-r4f): Add fpu.
10925 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
10926 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
10928 (cortex-r8): Likewise.
10929 (cortex-m4): Add fpu. Add option for nofp.
10930 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
10931 (cortex-a17.cortex-a7): Likewise.
10932 (cortex-a32): Add fpu. Add options for crypto and nofp.
10933 (cortex-a35, cortex-a53): Likewise.
10934 (cortex-a57): Add fpu. Add option for crypto.
10935 (cortex-a72, cortex-a73): Likewise.
10936 (exynos-m1): Likewise.
10937 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
10938 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
10939 (cortex-m33): Add fpu. Add option for nofp.
10940 * config/arm/arm-cpu-cdata.h: Regenerated
10941 * config/arm/arm-cpu-data.h: Regenerated.
10943 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10945 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
10946 (armv5te, armv5tej): Likewise.
10947 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
10948 (armv7): Add options fp and vfpv3-d16.
10949 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
10950 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
10952 (armv7ve): Likewise.
10953 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
10954 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
10955 (armv8-a): Add nocrypto option.
10956 (armv8.1-a, armv8.2-a): Likewise.
10957 (armv8-m.main): add options fp, fp.dp and nofp.
10959 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10961 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
10963 (armv8-a+crc): Delete.
10964 (armv8.1-a): Add options simd, crypto and nofp.
10965 (armv8.2-a): Add options fp16, simd, crypto and nofp.
10966 (armv8.2-a+fp16): Delete.
10967 (armv8-m.main): Add option dsp.
10968 (armv8-m.main+dsp): Delete.
10969 (cortex-a8): Add fpu. Add nofp option.
10970 (cortex-a9): Add fpu. Add nofp and nosimd options.
10971 * config/arm/parsecpu.awk (gen_data): Generate option tables and
10972 link to main cpu and architecture data structures.
10973 (gen_comm_data): Only put isa attributes from the main architecture
10975 (option): New statement for architecture and CPU entries.
10976 * arm.c (struct cpu_option): New structure.
10977 (struct processors): Add entry for options.
10978 (arm_unrecognized_feature): New function.
10979 (arm_parse_arch_cpu_name): Ignore any characters after the first
10981 (arm_parse_arch_cpu_feature): New function.
10982 (arm_configure_build_target): Separate out any CPU and architecture
10983 features and parse separately. Don't error out if -mfpu=auto is
10984 used with only an architecture string.
10985 (arm_print_asm_arch_directives): New function.
10986 (arm_file_start): Call it.
10987 * config/arm/arm-cpu-cdata.h: Regenerated.
10988 * config/arm/arm-cpu-data.h: Likewise.
10989 * config/arm/arm-tables.opt: Likewise.
10991 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10993 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
10994 assembler when it is not -mfpu=auto.
10996 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
10998 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
10999 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
11000 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
11001 (ASM_CPU_SPEC): Rewrite.
11002 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
11003 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
11004 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
11005 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
11006 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
11007 copied string is NUL-terminated. Also strip any characters prefixed
11009 (arm_rewrite_selected_arch): New function.
11010 (arm_rewrite_march): New function.
11012 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
11014 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
11015 (x_arm_cpu_string, x_arm_tune_string): Likewise.
11016 (march, mcpu, mtune): Convert to string-based options.
11017 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
11018 (arm_parse_arch_cpu_name): New function.
11019 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
11020 identify selected architecture or CPU.
11021 (arm_option_save): New function.
11022 (TARGET_OPTION_SAVE): Redefine.
11023 (arm_option_restore): Restore string options.
11024 (arm_option_print): Print string options.
11026 2017-06-16 Martin Sebor <msebor@redhat.com>
11028 PR tree-optimization/80933
11029 PR tree-optimization/80934
11030 * builtins.c (fold_builtin_3): Do not handle bcmp here.
11031 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
11032 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
11033 (gimple_fold_builtin): Call them.
11035 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11037 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
11038 as unlikely; update profile.
11040 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11042 * predict.c (force_edge_cold): Handle declaring edges impossible
11045 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11047 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
11049 (try_unroll_loop_completely): Fix reporting.
11051 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11053 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
11055 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
11058 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
11059 if given a non-constant argument for an intrinsic which requires a
11062 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11064 * profile.c (compare_freqs): New function.
11065 (branch_prob): Sort edge list.
11066 (find_spanning_tree): Assume that the list is priority sorted.
11068 2017-06-16 Richard Biener <rguenther@suse.de>
11070 PR tree-optimization/81090
11071 * passes.def (pass_record_bounds): Remove.
11072 * tree-pass.h (make_pass_record_bounds): Likewise.
11073 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
11074 make_pass_record_bounds): Likewise.
11075 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
11076 not free niter estimates at the beginning but at the end.
11077 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
11079 2017-06-16 Richard Biener <rguenther@suse.de>
11081 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
11082 initializer to workaround ICE in host GCC 4.8.
11084 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
11086 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
11088 (clone_inlined_nodes): Update.
11090 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11092 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
11093 prefetch settings, and enable prefetching by default at -O3.
11095 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11097 * config/aarch64/aarch64.c (aarch64_override_options_internal):
11098 Set flag_prefetch_loop_arrays according to tuning data.
11100 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
11102 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
11103 New tune structure.
11104 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
11105 [Unrelated to main purpose of the patch] Place the pointer field last
11106 to enable type checking errors when tune structure are wrongly merged.
11107 * config/aarch64/aarch64.c (generic_prefetch_tune,)
11108 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
11109 (thunderx2t99_prefetch_tune): New tune constants.
11110 (tune_params *_tunings): Update all tunings (no functional change).
11111 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
11112 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
11113 from tunings structures.
11115 2017-06-16 Jakub Jelinek <jakub@redhat.com>
11118 * ubsan.c (instrument_null): Add T argument, use it instead
11119 of computing it based on IS_LHS.
11120 (instrument_object_size): Likewise.
11121 (pass_ubsan::execute): Adjust instrument_null and
11122 instrument_object_size callers to pass gimple_get_lhs or
11123 gimple_assign_rhs1 result to it. Use instrument_null instead of
11124 calling get_base_address and instrument_mem_ref. Handle
11125 aggregate call arguments for object-size sanitization.
11127 2017-06-16 Yury Gribov <tetra2005@gmail.com>
11129 PR tree-optimization/81089
11130 * tree-vrp.c (is_masked_range_test): Validate operands of
11133 2017-06-15 Martin Sebor <msebor@redhat.com>
11136 * dumpfile.c (dump_register): Avoid calling memset to initialize
11137 a class with a default ctor.
11138 * gcc.c (struct compiler): Remove const qualification.
11139 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
11140 * hash-table.h: Ditto.
11141 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
11143 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
11144 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
11146 * params.h (struct param_info): Make struct members non-const.
11147 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
11148 with default initialization.
11149 * vec.h (vec_copy_construct, vec_default_construct): New helper
11151 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
11152 with vec_copy_construct.
11153 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
11154 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
11155 * doc/invoke.texi (-Wclass-memaccess): Document.
11157 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
11159 * emit-rtl.h (is_leaf): Update comment about local
11160 register allocator.
11162 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11165 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
11166 for a variable to have a section before checking if the section has a
11168 Set section to.persistent if persistent attribute is set.
11169 Warn if .persistent attribute is used on an automatic variable.
11171 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
11173 PR rtl-optimization/80474
11174 * reorg.c (update_block): Do not ignore instructions in a delay slot.
11176 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
11178 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
11181 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
11183 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
11184 (casesi): Emit bounds checking as RTL.
11185 (casesi_internal_mips16_<mode>): Remove bounds checking.
11187 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
11189 * config/xtensa/xtensa.c (xtensa_option_override): Append
11190 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
11191 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
11192 xtensa_doloop_hooks): Define unconditionally.
11193 (xtensa_reorg_loops): Only call reorg_loops in the presence of
11195 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
11196 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
11197 for it in xtensa_option_override.
11198 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
11199 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
11201 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
11203 * doc/cppopts.texi: Document '-' special value to -MF.
11205 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
11207 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
11208 (cortex_a53_fconst): Likewise.
11209 (cortex_a53_fpmul): Likewise.
11210 (cortex_a53_f_load_64): Likewise.
11211 (cortex_a53_f_load_many): Likewise.
11212 (cortex_a53_advsimd_alu): Likewise.
11213 (cortex_a53_advsimd_alu_q): Likewise.
11214 (cortex_a53_advsimd_mul): Likewise.
11215 (cortex_a53_advsimd_mul_q): Likewise.
11216 (fpmac bypass): Add new bypass for fpmac-fpmac case.
11217 Add missing fmul, r2f_cvt and fconst cases.
11219 2017-06-14 Richard Biener <rguenther@suse.de>
11221 PR middle-end/81088
11222 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
11224 (fold_binary_loc): When associating do not treat pre-existing
11225 TREE_OVERFLOW on literal constants as a reason to allow
11226 TREE_OVERFLOW on associated literal constants.
11228 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11230 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
11231 (MASK_FEATURES): New macro.
11232 * config/sparc/sparc.c (sparc_option_override): Remove the special
11233 handling of -mfpu and generalize it to all MASK_FEATURES switches.
11235 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11237 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
11238 a division of 0 if non-call exceptions are enabled.
11240 2017-06-14 Andrew Pinski <apinski@cavium.com>
11241 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11244 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
11245 Improve vector initialization code gen for only variable case.
11247 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
11249 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
11251 2017-06-14 Richard Biener <rguenther@suse.de>
11253 PR tree-optimization/81083
11254 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
11257 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11259 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
11260 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
11261 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
11262 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
11263 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
11264 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
11266 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11268 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
11269 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
11271 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11273 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
11275 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11277 * config/rs6000/t-rtems: Don't handle SPE.
11279 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11281 * config/rs6000/t-linux: Don't handle SPE.
11283 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11285 * config/rs6000/eabispe.h: Delete file.
11287 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11289 * config/rs6000/t-spe: Delete file.
11291 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11293 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
11294 (rs6000_legitimate_offset_address_p): Return false for anything in
11295 V2SImode or V2SFmode.
11297 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
11299 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
11300 except V2SF and V2SI. Rearrange the vector modes, and add comments.
11301 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
11303 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
11304 (rs6000_legitimate_offset_address_p): Ditto.
11305 (rs6000_emit_move): Ditto.
11306 (rs6000_init_builtins): Remove V4HI_type_node.
11308 2017-06-13 Martin Liska <mliska@suse.cz>
11311 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
11312 (gate_asan): Likewise.
11313 * asan.h (asan_no_sanitize_address_p): Remove the function.
11314 (sanitize_flags_p): New function.
11315 * builtins.def: Fix coding style.
11316 * common.opt: Use renamed enum value.
11317 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
11318 * doc/extend.texi: Document no_sanitize attribute.
11319 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
11320 to SANITIZE_UNDEFINED_NONDEFAULT.
11321 * gcc.c (sanitize_spec_function): Use the renamed enum value.
11322 * gimple-fold.c (optimize_atomic_compare_exchange_p):
11323 Use sanitize_flags_p.
11324 * gimplify.c (gimplify_function_tree): Likewise.
11325 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
11326 * opts.c (parse_no_sanitize_attribute): New function.
11327 (common_handle_option): Use renamed enum value.
11328 * opts.h (parse_no_sanitize_attribute): Declare.
11329 * tree.c (sanitize_flags_p): New function.
11330 * tree.h: Declared here.
11331 * tsan.c: Use sanitize_flags_p.
11332 * ubsan.c (ubsan_expand_null_ifn): Likewise.
11333 (instrument_mem_ref): Likewise.
11334 (instrument_bool_enum_load): Likewise.
11335 (do_ubsan_in_current_function): Remove the function.
11336 (pass_ubsan::execute): Use sanitize_flags_p.
11337 * ubsan.h: Remove do_ubsan_in_current_function
11338 * tree-cfg.c (print_no_sanitize_attr_value): New function.
11339 (dump_function_to_file): Use it here.
11341 2017-06-13 Martin Jambor <mjambor@suse.cz>
11343 PR tree-optimization/80803
11344 PR tree-optimization/81063
11345 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
11346 (propagate_subaccesses_across_link): Enqueue subtree whenever
11347 necessary instead of relying on the caller.
11349 2017-06-13 Martin Jambor <mjambor@suse.cz>
11351 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
11352 that have a first_link.
11353 (sort_and_splice_var_accesses): Do not check first_link before
11355 (subtree_mark_written_and_enqueue): Likewise.
11356 (propagate_all_subaccesses): Likewise and do not stop at first
11357 parent with a first_link.
11359 2017-06-13 Martin Jambor <mjambor@suse.cz>
11361 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
11364 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11366 * match.pd: New pattern.
11368 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11370 * tree-vrp.c (is_masked_range_test): New function.
11371 (register_edge_assert_for): Determine ranges for
11374 2017-06-13 Yury Gribov <tetra2005@gmail.com>
11376 PR tree-optimization/67328
11377 * fold-const.c (maskable_range_p): New function.
11378 (build_range_check): Generate bittests if possible.
11380 2017-06-13 Martin Liska <mliska@suse.cz>
11382 * gimple-pretty-print.c (dump_probability): Add new argument.
11383 (dump_edge_probability): Dump both probability and count.
11384 (dump_gimple_label): Likewise.
11385 (dump_gimple_bb_header): Likewise.
11387 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
11390 * config/avr/avr-devices.c: Fix indentation.
11391 * config/avr/gen-avr-mmcu-specs.c: Dito.
11393 2017-06-13 Richard Biener <rguenther@suse.de>
11395 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
11396 instead get vector type from stmt_info.
11397 (vectorizable_reduction): Adjust. Remove dead code.
11399 2017-06-13 Richard Biener <rguenther@suse.de>
11401 PR middle-end/81065
11402 * fold-const.c (extract_muldiv_1): Remove bogus distribution
11403 case of C * (x * C2 + C3).
11404 (fold_addr_of_array_ref_difference): Properly fold index difference.
11406 2017-06-12 David S. Miller <davem@davemloft.net>
11409 * config/sparc/sparc.md (return expander): Emit frame blockage if
11410 function uses alloca.
11412 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
11414 * combine.c (make_field_assignment): Check len rather than the mode
11415 precision when calling force_to_mode.
11417 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
11419 Support multilibs and devices that see flash in RAM address range.
11422 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
11423 (avr_mcu_t) <flash_pm_offset>: New field.
11424 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
11425 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
11426 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
11427 (AVR_TINY_PM_OFFSET): Remove macro.
11428 * config/avr/avr.opt (-mshort-calls): New option.
11429 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
11430 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
11431 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
11432 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
11433 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
11434 instead of avr_arch->have_jmp_call.
11435 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
11436 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
11437 avr_arch->flash_pm_offset to define.
11438 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
11439 new field flash_pm_offset. Add entry for avrxmega3.
11440 (avr_texinfo): Add entry for avrxmega3.
11441 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
11442 attiny212, attiny214,
11443 attiny412, attiny414, attiny416, attiny417,
11444 attiny814, attiny816, attiny817,
11445 attiny1614, attiny1616, attiny1617,
11446 attiny3214, attiny3216, attiny3217.
11447 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
11448 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
11449 (avr_print_operand_address) [AVR_TINY]: Same.
11450 (avr_asm_init_sections) <readonly_data_section>: Only patch
11451 callback if avr_arch->flash_pm_offset = 0.
11452 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
11453 for rodata if avr_arch->flash_pm_offset != 0.
11454 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
11455 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
11456 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
11457 (m_options): Append opt_rcall.
11458 (m_dirnames): Append dir_rcall.
11459 * config/avr/t-multilib: Regenerate.
11461 * configure.ac [target=avr]: Check whether avrxmega3 default
11462 linker description file works as needed.
11463 * configure: Regenerate.
11464 * doc/avr-mmcu.texi: Regenerate.
11465 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
11466 <__AVR_ARCH__>: Document avrxmega3 and 103.
11467 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
11468 <__AVR_SHORT_CALLS__>: Document it.
11469 <__AVR_PM_BASE_ADDRESS__>: Document it.
11470 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
11471 (AVR Variable Attributes) <progmem>: Document this is
11472 not needed for avrxmega3.
11473 (AVR Named Address Spaces) <__flash>: Dito.
11475 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
11477 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
11479 2017-06-12 Doug Rupp <rupp@adacore.com>
11481 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
11482 Append vxworks-stdint.h to the tm_file list.
11483 * config/vxworks-stdint.h: New file.
11485 2017-06-12 Martin Liska <mliska@suse.cz>
11487 PR tree-optimization/81041
11488 * tree-profile.c (gimple_gen_ic_func_profiler):
11489 Create an extra BB in profile-generate
11490 (gimple_gen_time_profiler): Likewise.
11492 2017-06-12 Jakub Jelinek <jakub@redhat.com>
11494 PR tree-optimization/81003
11495 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
11496 (update_range_test): Use it instead of force_gimple_operand_gsi.
11498 2017-06-12 Richard Biener <rguenther@suse.de>
11500 PR tree-optimization/81053
11501 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
11502 with backedge value not defined in loop. Simplify def stmt
11505 2017-06-11 Tom de Vries <tom@codesourcery.com>
11508 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
11510 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
11511 nvptx_cannot_force_const_mem.
11513 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11515 * opts.c (finish_options): Move test for flag_split_stack after
11516 it has been initialized.
11518 2017-06-11 Jason Merrill <jason@redhat.com>
11520 * tree.h (id_equal): New.
11521 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
11522 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
11523 instead of strcmp of IDENTIFIER_POINTER.
11525 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11527 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
11528 (mark_all_inlined_calls_cdtor): Fix formating.
11529 (inline_transform): Rescale profile before inlining.
11531 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11533 * cgraph.h (cgraph_edge::clone): Update prototype.
11534 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
11535 (cgraph_node::create_clone): Update.
11536 (cgraph_node::create_version_clone): Update.
11537 * tree-inline.c (copy_bb): Update.
11538 (expand_call_inline): Update.
11540 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11542 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
11543 factored out from ...
11544 (rs6000_emit_prologue): ... here.
11546 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
11548 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
11549 factored out from ...
11550 (rs6000_emit_prologue): ... here.
11552 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11554 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
11556 (handle_missing_profiles): Fix computation of tp_first_run.
11557 (counts_to_freqs): Do not touch freqs when count is 0.
11559 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
11561 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
11564 2017-06-10 Tom de Vries <tom@codesourcery.com>
11566 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
11567 attributes): Document signal effective target.
11569 2017-06-10 Tom de Vries <tom@codesourcery.com>
11571 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
11572 Document effective target stack_size.
11574 2017-06-09 David Malcolm <dmalcolm@redhat.com>
11576 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
11577 to the edit_context if they can be auto-applied.
11579 2017-06-9 Ian Lance Taylor <iant@golang.org>
11581 * opts.c (finish_options): If -fsplit-stack, disable implicit
11582 -forder-blocks-and-partition.
11583 * doc/invoke.texi (Optimize Options): Document that when using
11584 -fsplit-stack -forder-blocks-and-partition is not implicitly
11587 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11589 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
11590 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
11591 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
11592 * builtins.def (abort, trap, unreachable): Declare cold.
11593 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
11594 * tree-core.h (ECF_COLD): New.
11595 * tree.c (set_call_expr_flags): Handle ECF_COLD.
11596 (build_common_builtin_nodes): Mark unreachable and abort as cold.
11598 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11600 * predict.c (unlikely_executed_stmt_p): Cleanup.
11602 2017-06-09 Richard Biener <rguenther@suse.de>
11604 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
11605 model if the ref is always written to.
11607 2017-06-09 Tamar Christina <tamar.christina@arm.com>
11609 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
11611 2017-06-09 Tamar Christina <tamar.christina@arm.com>
11613 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
11616 2017-06-09 Tom de Vries <tom@codesourcery.com>
11619 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
11620 "target cannot support label values" when encountering LABEL_REF.
11622 2017-06-09 Martin Liska <mliska@suse.cz>
11624 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
11625 (gimple_gen_ic_func_profiler): Emit direct comparison
11626 of __gcov_indirect_call_callee with NULL.
11627 (gimple_gen_time_profiler): Change probability from
11628 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
11630 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11632 * profile.c (edge_gcov_counts): Turn to pointer.
11633 (compute_branch_probabilities, compute_branch_probabilities): Update.
11634 (branch_prob): Do not clear edge_gcov_count.
11635 * profile.h (edge_gcov_counts): Turn to pointer.
11636 (edge_gcov_count): Update.
11638 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
11640 * gimple.h (gimple_check_failed): Mark cold.
11642 2017-06-09 Richard Biener <rguenther@suse.de>
11644 PR tree-optimization/66623
11645 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
11646 refactor check_reduction into two parts, properly computing
11647 whether we have to check reduction validity for outer loop
11650 2017-06-09 Richard Biener <rguenther@suse.de>
11652 PR tree-optimization/79483
11653 * graphite-scop-detection.c (order): New global.
11654 (get_order): Compute bb to order mapping that satisfies code
11655 generation constraints.
11656 (cmp_pbbs): New helper.
11657 (build_scops): Start domwalk at entry block, sort generated
11660 2017-06-09 Richard Biener <rguenther@suse.de>
11662 PR middle-end/81007
11663 * ipa-polymorphic-call.c
11664 (ipa_polymorphic_call_context::restrict_to_inner_class):
11665 Skip FIELD_DECLs with error_mark_node type.
11666 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
11669 2017-06-09 Martin Liska <mliska@suse.cz>
11671 * predict.c (struct branch_predictor): New struct.
11672 (test_prediction_value_range): New test.
11673 (predict_c_tests): New function.
11674 * selftest-run-tests.c (selftest::run_tests): Run the function.
11675 * selftest.h: Declare new tests.
11677 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
11680 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
11681 gen_add3_insn did not fail.
11682 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
11683 r0, construct that number in a temporary reg and add that reg to r0.
11684 If asked to put the result in r0 as well, fail.
11686 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
11688 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11689 for early expansion of vec_eqv.
11691 2017-06-08 Jakub Jelinek <jakub@redhat.com>
11693 PR middle-end/81005
11694 * ubsan.c (instrument_null): Avoid pointless code temporary.
11695 (pass_ubsan::execute): Instrument aggregate arguments of calls.
11697 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
11701 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
11704 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
11707 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11709 * predict.c (unlikely_executed_edge_p): Move ahead.
11710 (probably_never_executed_edge_p): Use it.
11712 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
11714 PR middle-end/79988
11715 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
11716 gimple_call_builtin_p call.
11718 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11720 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
11721 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
11722 rtl_check_failed_type2, rtl_check_failed_code1,
11723 rtl_check_failed_code2, rtl_check_failed_code_mode,
11724 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
11725 rtvec_check_failed_bounds, rtl_check_failed_flag,
11726 _fatal_insn_not_found, _fatal_insn): Likewise.
11727 * tree.h (tree_contains_struct_check_failed,
11728 tree_check_failed, tree_not_check_failed,
11729 tree_class_check_failed, tree_range_check_failed,
11730 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
11731 tree_vec_elt_check_failed, phi_node_elt_check_failed,
11732 tree_operand_check_failed, omp_clause_check_failed,
11733 omp_clause_operand_check_failed, omp_clause_range_check_failed):
11736 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11738 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
11739 flag_branch_probabilities.
11740 * ipa-inline.c (edge_badness): Likewise.
11741 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
11742 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
11743 * predict.c (maybe_hot_frequency_p): Likewise.
11744 (probably_never_executed): Likewise.
11745 * sched-ebb.c (schedule_ebbs): Likewise.
11746 * sched-rgn.c (find_single_block_region): Likewise.
11747 * tracer.c (tail_duplicate): Likewise.
11749 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11751 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
11752 longer requires x_flag_profile_use.
11754 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11756 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
11757 instead of flag_reorder_blocks_and_partition.
11758 * dbxout.c (dbxout_function_end): Likewise.
11759 * dwarf2out.c (gen_subprogram_die): Likewise.
11760 * haifa-sched.c (sched_create_recovery_edges): Likewise.
11761 * hw-doloop.c (reorg_loops): Likewise.
11762 * varasm.c (assemble_start_function,
11763 assemble_end_function): Likewise.
11764 (decide_function_section): Do not check for
11765 flag_reorder_blocks_and_partition.
11767 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
11769 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
11771 (chkp_get_hard_register_fake_addr_expr): Ditto.
11772 (chkp_build_addr_expr): Add check for hard reg case.
11773 (chkp_parse_array_and_component_ref): Ditto.
11774 (chkp_find_bounds_1): Ditto.
11775 (chkp_process_stmt): Don't generate bounds store for
11778 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11780 * predict.c (maybe_hot_bb_p): Do not check profile status.
11781 (maybe_hot_edge_p): Likewise.
11782 (probably_never_executed): Check for zero counts even if profile
11784 (unlikely_executed_edge_p): New function.
11785 (unlikely_executed_stmt_p): New function.
11786 (unlikely_executed_bb_p): New function.
11787 (set_even_probabilities): Use unlikely predicates.
11788 (combine_predictions_for_bb): Likewise.
11789 (predict_paths_for_bb): Likewise.
11790 (predict_paths_leading_to_edge): Likewise.
11791 (determine_unlikely_bbs): New function.
11792 (estimate_bb_frequencies): Use it.
11793 (compute_function_frequency): Use zero counts even if profile is
11795 * profile-count.h: Fix typo.
11797 2017-08-08 Julia Koval <julia.koval@intel.com>
11799 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
11800 _mm512_mask_cvtsepi16_storeu_epi8,
11801 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
11802 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
11803 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
11804 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
11805 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
11806 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
11807 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
11808 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
11809 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
11810 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
11811 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
11812 __builtin_ia32_pmovuswb256mem_mask,
11813 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
11814 __builtin_ia32_pmovwb512mem_mask): New builtins.
11816 2017-08-08 Julia Koval <julia.koval@intel.com>
11818 PR target/73350,80862
11819 * config/i386/subst.md (round): Fix round pattern.
11820 * config/i386/i386.c (ix86_erase_embedded_rounding):
11821 Fix erasing rounding for the fixed pattern.
11823 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
11825 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
11827 2017-06-08 Martin Liska <mliska@suse.cz>
11829 PR gcov-profile/80911
11830 * gcov.c (block_info::block_info): New constructor.
11832 2017-06-07 Carl Love <cel@us.ibm.com>
11834 * config/rs6000/rs6000-c: The return type of the following
11835 built-in functions was implemented as int not long long. Fix sign
11836 of return value for the unsigned version of vec_mulo and vec_mule.
11837 vector unsigned long long vec_bperm (vector unsigned long long,
11838 vector unsigned char)
11839 vector signed long long vec_mule (vector signed int,
11841 vector unsigned long long vec_mule (vector unsigned int,
11842 vector unsigned int)
11843 vector signed long long vec_mulo (vector signed int,
11845 vector unsigned long long vec_mulo (vector unsigned int,
11846 vector unsigned int)
11847 * doc/extend.texi: Fix the documentation for the built-in
11850 2017-06-07 Carl Love <cel@us.ibm.com>
11853 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
11856 2017-06-07 Carl Love <cel@us.ibm.com>
11858 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
11859 support, Generate doublehv for signed int/float for BE case only.
11861 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
11863 * doc/invoke.texi (mcx16): Rewrite.
11865 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11867 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
11868 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
11869 *mov<mode>_softfloat, and an anonymous splitter): Use
11870 nonimmediate_operand instead of rs6000_nonimmediate_operand.
11872 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11874 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
11876 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
11877 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
11878 (rs6000_debug_reg_global): Adjust.
11879 (rs6000_init_hard_regno_mode_ok): Adjust.
11880 (rs6000_dbx_register_number): Adjust.
11881 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
11882 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
11883 Remove SPE_ACC and SPEFSCR.
11884 (REG_ALLOC_ORDER): Ditto.
11885 (FRAME_POINTER_REGNUM): Change to 111.
11886 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
11887 (REG_CLASS_NAMES): Ditto.
11888 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
11889 (REGISTER_NAMES): Ditto.
11890 (ADDITIONAL_REG_NAMES): Ditto.
11891 (rs6000_reg_names): Ditto.
11892 * config/rs6000/rs6000.md: Renumber some register number
11895 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11897 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
11899 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
11900 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
11902 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
11903 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
11904 Delete the SPE high registers.
11905 (REG_ALLOC_ORDER): Ditto.
11906 (enum reg_class): Remove SPE_HIGH_REGS.
11907 (REG_CLASS_NAMES): Ditto.
11908 (REG_CLASS_CONTENTS): Delete the SPE high registers.
11909 (REGISTER_NAMES): Ditto.
11910 (rs6000_reg_names): Ditto.
11911 * doc/tm.texi.in: Remove SPE as example.
11912 * doc/tm.texi: Regenerate.
11914 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11916 * config/rs6000/8540.md (ppc8540_brinc): Delete.
11917 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
11918 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
11919 * config/rs6000/rs6000.md (type): Remove "brinc".
11921 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11923 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
11924 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
11925 * config/rs6000/linuxspe.h: Delete file.
11926 * config/rs6000/rs6000.md: Don't include spe.md.
11927 * config/rs6000/spe.h: Delete file.
11928 * config/rs6000/spe.md: Delete file.
11929 * config/rs6000/t-rs6000: Remove spe.md.
11931 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11933 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
11934 (reg_or_none500mem_operand): Delete.
11935 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
11936 instead of reg_or_none500mem_operand.
11938 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11940 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
11941 handling of SPE flags.
11942 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
11944 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
11946 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
11948 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
11949 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
11950 paired_divv2sf3): Similar.
11951 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
11952 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
11953 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
11955 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
11956 Rename the paired_* instruction patterns.
11957 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
11959 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
11960 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
11961 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
11962 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
11963 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
11964 PAIRED_VECTOR_MODE.
11965 (struct machine_function): Delete field spe_insn_chain_scanned_p.
11966 (spe_func_has_64bit_regs_p): Delete.
11967 (spe_expand_predicate_builtin): Delete.
11968 (spe_expand_evsel_builtin): Delete.
11969 (TARGET_DWARF_REGISTER_SPAN): Do not define.
11970 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
11971 (invalid_e500_subreg): Delete.
11972 (rs6000_legitimize_address): Always force_reg op2 as well, for
11973 paired single memory accesses.
11974 (rs6000_member_type_forces_blk): Delete.
11975 (rs6000_spe_function_arg): Delete.
11976 (rs6000_expand_unop_builtin): Delete SPE handling.
11977 (rs6000_expand_binop_builtin): Ditto.
11978 (spe_expand_stv_builtin): Delete.
11979 (bdesc_2arg_spe): Delete.
11980 (spe_expand_builtin): Delete.
11981 (spe_expand_predicate_builtin): Delete.
11982 (spe_expand_evsel_builtin): Delete.
11983 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
11984 (spe_init_builtins): Delete.
11985 (spe_func_has_64bit_regs_p): Delete.
11986 (savres_routine_name): Delete "info" parameter. Adjust callers.
11987 (rs6000_emit_stack_reset): Ditto.
11988 (rs6000_dwarf_register_span): Delete.
11989 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
11990 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
11991 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
11993 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
11995 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
11996 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
11997 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
11998 mulv2sf3, divv2sf3): Delete expanders.
12000 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12002 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
12004 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12006 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
12007 * config/rs6000/rs6000.c: Ditto.
12009 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12011 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
12012 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
12013 comparison_operator.
12015 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12017 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
12018 * config/rs6000/rs6000.opt: Ditto.
12019 * config/rs6000/t-rtems: Ditto.
12021 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12023 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
12024 TARGET_E500_SINGLE by 0, simplify.
12025 * config/rs6000/rs6000.c: Ditto.
12026 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
12027 (spe_build_register_parallel): Delete.
12028 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
12029 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
12030 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
12031 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
12032 (E500_CONVERT): Delete.
12033 * config/rs6000/spe.md: Remove many patterns and all define_constants.
12035 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
12037 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
12038 * config/rs6000/dfp.md: Ditto.
12039 (negdd2, *negdd2_fpr): Merge.
12040 (absdd2, *absdd2_fpr): Merge.
12041 (negtd2, *negtd2_fpr): Merge.
12042 (abstd2, *abstd2_fpr): Merge.
12043 * config/rs6000/e500.h: Delete file.
12044 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
12045 TARGET_FPRS by 1 and simplify.
12046 * config/rs6000/rs6000-c.c: Ditto.
12047 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
12048 TARGET_DF_SPE by 0.
12049 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
12051 * config/rs6000/rs6000.md: Ditto.
12052 (floatdidf2, *floatdidf2_fpr): Merge.
12053 (move_from_CR_gt_bit): Delete.
12054 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
12055 (E500_CR_IOR_COMPARE): Delete.
12056 (All patterns that require !TARGET_FPRS): Delete.
12057 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
12059 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12061 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
12063 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12065 * graphds.c (add_edge): Intitialize edge's attached data.
12066 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
12067 pointer parameter. Call pointed function on each edge during
12068 graph traversing. Skip traversing the edge when the function
12070 (graphds_dfs, graphds_scc): Ditto.
12071 (for_each_edge): New parameter. Pass the new parameter to callback
12073 * graphds.h (skip_edge_callback): New function pointer type.
12074 (graphds_dfs, graphds_scc): New function pointer parameter.
12075 (graphds_edge_callback, for_each_edge): New parameter.
12077 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12079 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
12080 out code checking if runtime alias check is possible to below ...
12081 Call the new function.
12082 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
12083 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
12085 2017-06-07 Marek Polacek <polacek@redhat.com>
12088 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
12089 TYPE_OVERFLOW_WRAPS check.
12091 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12093 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
12094 if versioning is required.
12095 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
12096 peeling with the check for versioning.
12098 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12100 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
12101 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
12102 Set true to new parameter if new ssa variable is defined.
12103 (vect_gen_vector_loop_niters): Refactor. Set range information
12104 for the new vector loop bound variable.
12105 (vect_do_peeling): Ditto.
12107 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12109 * tree-affine.c (ssa.h): Include header file.
12110 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
12111 has wrapping overflow behavior.
12113 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12115 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
12117 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12119 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
12120 (tree_to_aff_combination): ... here.
12122 2017-06-07 Bin Cheng <bin.cheng@arm.com>
12124 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
12125 reg_pressure model function.
12126 (ivopts_global_cost_for_size): Delete.
12127 (determine_set_costs, iv_ca_recount_cost): Call new model function
12128 ivopts_estimate_reg_pressure.
12130 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12132 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
12133 expensive than udiv. Remove floating point cases from mod.
12135 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12137 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
12138 Increase idiv cost.
12140 2017-06-07 Tamar Christina <tamar.christina@arm.com>
12142 * config/aarch64/aarch64.md
12143 (copysignsf3): Fix mask generation.
12145 2017-06-07 Jakub Jelinek <jakub@redhat.com>
12147 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
12149 (class dump_manager): Add register_dumps method.
12150 * dumpfile.c: Include langhooks.h.
12151 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
12152 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
12153 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
12154 (dump_manager::dump_register): Start with 512 entries instead of 32.
12155 (dump_manager::register_dumps): New method.
12156 * toplev.c (general_init): Instead of invoking register_dumps
12157 langhook, invoke register_dumps method on the dump manager.
12158 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
12161 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
12163 * doc/md.texi: Clarify the restrictions on a define_insn condition.
12164 Say that # requires an associated define_split to exist, and that
12165 the define_split must be suitable for use after register allocation.
12167 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12169 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
12170 (compute_outgoing_frequencies): Also initialize zero counts.
12171 (find_many_sub_basic_blocks): Do not produce uninitialized profile
12172 around loops; preserve more of profile when nothing changes.
12174 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
12176 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
12178 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
12179 * config/arm/arm-cpu-cdata.h: Regenerate.
12180 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
12181 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
12182 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
12183 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
12184 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
12186 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
12188 * config/arm/t-rmprofile: Likewise.
12189 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
12191 2017-06-06 David S. Miller <davem@davemloft.net>
12194 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
12195 blockage if function uses alloca.
12197 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12199 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
12200 New "uid" fields to hold pretty-print IDs of group and ref.
12201 Memory references are now identified as <group_id>:<ref_id>
12202 instead of using [random] addresses.
12203 (dump_mem_details): Simplify, no functional change.
12204 (dump_mem_ref): Simplify and make output more concise.
12205 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
12206 (find_or_create_group): Initialize group uid.
12207 (record_ref): Initialize ref uid. Improve debug output.
12208 (prune_group_by_reuse, should_issue_prefetch_p,)
12209 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
12210 (mark_nontemporal_store, determine_loop_nest_reuse):
12211 Improve debug output.
12213 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
12215 * dbgcnt.def (prefetch): New debug counter.
12216 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
12217 (schedule_prefetches): Stop issueing prefetches if debug counter
12220 2017-06-06 Tom de Vries <tom@codesourcery.com>
12222 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
12223 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
12225 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12227 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
12228 Use aarch64_reg_or_zero predicate for operand 4.
12229 (aarch64_compare_and_swap<mode> define_insn_and_split):
12230 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
12231 (aarch64_store_exclusive<mode>): Likewise for operand 2.
12233 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
12235 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
12236 (arm_compute_save_core_reg_mask): This.
12237 (thumb1_compute_save_reg_mask): Rename into ...
12238 (thumb1_compute_save_core_reg_mask): This.
12239 (arm_compute_save_reg0_reg12_mask): Adapt comment.
12240 (arm_compute_frame_layout): Likewise.
12242 2017-06-06 Richard Biener <rguenther@suse.de>
12244 PR tree-optimization/80974
12245 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
12246 keep or clear leaders SSA info.
12248 2017-06-06 Tom de Vries <tom@codesourcery.com>
12250 * config/nvptx/nvptx.c (split_mode_p): New function.
12251 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
12253 2017-06-06 Tom de Vries <tom@codesourcery.com>
12255 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
12257 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12260 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
12263 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12265 * shrink-wrap.c (handle_simple_exit): Update profile.
12266 (try_shrink_wrapping): Upate profile.
12268 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12270 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
12271 (tree_guess_outgoing_edge_probabilities): New.
12272 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
12273 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
12275 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12277 * ipa-split.c (split_function): Initialize return bb profile.
12279 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
12281 * profile.c (compute_branch_probabilities): Also initialize
12282 EXIT_BLOCK profile.
12284 2017-06-06 Richard Biener <rguenther@suse.de>
12286 PR tree-optimization/80928
12287 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
12288 (vect_analyze_loop_operations): Properly guard analysis for
12290 (vect_transform_loop): Likewise.
12291 (vect_analyze_loop_2): Also reset SLP type on PHIs.
12292 (vect_model_induction_cost): Do not cost for pure SLP.
12293 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
12294 of induction in inner loop vectorization.
12295 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
12296 (vect_get_and_check_slp_defs): Handle vect_induction_def.
12297 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
12299 (vect_analyze_slp_cost_1): Cost induction.
12300 (vect_detect_hybrid_slp_stmts): Handle PHIs.
12301 (vect_get_slp_vect_defs): Likewise.
12302 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
12303 (vect_transform_stmt): Handle SLP reductions.
12304 * tree-vectorizer.h (vectorizable_induction): Adjust.
12306 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12308 * config/rs6000/rs6000.c (make_resolver_func): Update
12309 init_lowered_empty_function call.
12311 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
12313 * doc/invoke.texi: Document the -fprofile-abs-path option.
12314 * common.opt (fprofile-abs-path): New option.
12315 * gcov-io.h (gcov_write_filename): Declare.
12316 * gcov-io.c (gcov_write_filename): New function.
12317 * coverage.c (coverage_begin_function): Use gcov_write_filename.
12318 * profile.c (output_location): Likewise.
12320 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12322 * shring-wrap.c: Revert accidental commit.
12324 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
12326 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
12328 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12330 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
12332 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
12333 profile in callgraph edge.
12334 * profile-count.h (apply_probability): If THIS is 0, then result is 0
12335 (apply_scale): Likewise.
12336 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
12337 Also scale profile when inlining function with zero profile.
12338 (initialize_cfun): Update exit block profile even when it is zero.
12339 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
12340 when profile is read.
12342 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
12344 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
12345 (CLONE_*): New constants to define the processors we can generate
12346 code for with the target_clone attribute.
12347 (rs6000_clone_map): New array to identify which clone processors
12348 the current program is running on.
12349 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
12350 target_clone attribute.
12351 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
12352 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
12353 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
12354 (cpu_expand_builtin): Add support for target_clone attribute.
12355 (rs6000_valid_attribute_p): Allow "default" attribute.
12356 (get_decl_name): New debug function to simplify printing the
12357 current function name in debugging statements.
12358 (rs6000_clone_priority): New functions to support the target_clone
12359 attribute, and be able to generate code to switch between ISA 2.05
12360 through ISA 3.0 (power6 through power9).
12361 (rs6000_compare_version_priority): Likewise.
12362 (rs6000_get_function_versions_dispatcher): Likewise.
12363 (make_resolver_func): Likewise.
12364 (add_condition_to_bb): Likewise.
12365 (dispatch_function_versions): Likewise.
12366 (rs6000_generate_version_dispatcher_body): Likewise.
12367 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
12368 (fusion_gpr_load_p): Fix a spacing issue.
12369 * doc/extend.texi (Common Function Attributes): Document that the
12370 PowerPC supports the target_clone attribute.
12372 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
12374 * config/arm/arm.h: explain F symbol found in description of ARM
12375 register allocation in its legend.
12377 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12379 * config/mips/frame-header-opt.c: Include profile-count.h.
12380 * config/riscv/riscv.c: Include profile-count.h
12382 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
12384 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
12386 (sm_set_flag_if_changed): Add bbs field.
12387 (execute_sm_if_changed_flag_set): Pass BBS.
12388 (execute_sm): Update.
12390 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12392 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
12395 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12397 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
12398 (peephole2): New peephole2 to emit the above.
12399 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
12401 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12403 * config/aarch64/aarch64.c (define_peephole2 above
12404 *sub_<shift>_<mode>): New peephole.
12406 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12408 * config/i386/i386.c (make_resolver_func): Update.
12409 * Makefile.in: Add profile-count.h and profile-count.o
12410 * auto-profile.c (afdo_indirect_call): Update to new API.
12411 (afdo_set_bb_count): Update.
12412 (afdo_propagate_edge): Update.
12413 (afdo_propagate_circuit): Update.
12414 (afdo_calculate_branch_prob): Update.
12415 (afdo_annotate_cfg): Update.
12416 * basic-block.h: Include profile-count.h
12417 (struct edge_def): Turn count to profile_count.
12418 (struct basic_block_def): Likewie.
12419 (REG_BR_PROB_BASE): Move to profile-count.h
12420 (RDIV): Move to profile-count.h
12421 * bb-reorder.c (max_entry_count): Turn to profile_count.
12422 (find_traces): Update.
12423 (rotate_loop):Update.
12424 (connect_traces):Update.
12425 (sanitize_hot_paths):Update.
12426 * bt-load.c (migrate_btr_defs): Update.
12427 * cfg.c (RDIV): Remove.
12428 (init_flow): Use alloc_block.
12429 (alloc_block): Uninitialize count.
12430 (unchecked_make_edge): Uninitialize count.
12431 (check_bb_profile): Update.
12432 (dump_edge_info): Update.
12433 (dump_bb_info): Update.
12434 (update_bb_profile_for_threading): Update.
12435 (scale_bbs_frequencies_int): Update.
12436 (scale_bbs_frequencies_gcov_type): Update.
12437 (scale_bbs_frequencies_profile_count): New.
12438 * cfg.h (update_bb_profile_for_threading): Update.
12439 (scale_bbs_frequencies_profile_count): Declare.
12440 * cfgbuild.c (compute_outgoing_frequencies): Update.
12441 (find_many_sub_basic_blocks): Update.
12442 * cfgcleanup.c (try_forward_edges): Update.
12443 (try_crossjump_to_edge): Update.
12444 * cfgexpand.c (expand_gimple_tailcall): Update.
12445 (construct_exit_block): Update.
12446 * cfghooks.c (verify_flow_info): Update.
12447 (dump_bb_for_graph): Update.
12448 (split_edge): Update.
12449 (make_forwarder_block): Update.
12450 (duplicate_block): Update.
12451 (account_profile_record): Update.
12452 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
12453 (get_estimated_loop_iterations): Update.
12454 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
12455 (single_likely_exit): Update.
12456 * cfgloopmanip.c (scale_loop_profile): Update.
12458 (set_zero_probability): Update.
12459 (lv_adjust_loop_entry_edge): Update.
12460 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12461 (purge_dead_edges): Update.
12462 (rtl_account_profile_record): Update.
12463 * cgraph.c (cgraph_node::create): Uninitialize count.
12464 (symbol_table::create_edge): Uninitialize count.
12465 (cgraph_update_edges_for_call_stmt_node): Update.
12466 (cgraph_edge::dump_edge_flags): Update.
12467 (cgraph_node::dump): Update.
12468 (cgraph_edge::maybe_hot_p): Update.
12469 * cgraph.h: Include profile-count.h
12470 (create_clone), create_edge, create_indirect_edge): Update.
12471 (cgraph_node): Turn count to profile_count.
12472 (cgraph_edge0: Likewise.
12473 (make_speculative, clone): Update.
12474 (create_edge): Update.
12475 (init_lowered_empty_function): Update.
12476 * cgraphclones.c (cgraph_edge::clone): Update.
12477 (duplicate_thunk_for_node): Update.
12478 (cgraph_node::create_clone): Update.
12479 * cgraphunit.c (cgraph_node::analyze): Update.
12480 (cgraph_node::expand_thunk): Update.
12481 * final.c (dump_basic_block_info): Update.
12482 * gimple-streamer-in.c (input_bb): Update.
12483 * gimple-streamer-out.c (output_bb): Update.
12484 * graphite.c (print_global_statistics): Update.
12485 (print_graphite_scop_statistics): Update.
12486 * hsa-brig.c: Include basic-block.h.
12487 * hsa-dump.c: Include basic-block.h.
12488 * hsa-gen.c (T sum_slice): Update.
12489 (convert_switch_statements):Update.
12490 * hsa-regalloc.c: Include basic-block.h.
12491 * ipa-chkp.c (chkp_produce_thunks): Update.
12492 * ipa-cp.c (struct caller_statistics): Update.
12493 (init_caller_stats): Update.
12494 (gather_caller_stats): Update.
12495 (ipcp_cloning_candidate_p): Update.
12496 (good_cloning_opportunity_p): Update.
12497 (get_info_about_necessary_edges): Update.
12498 (dump_profile_updates): Update.
12499 (update_profiling_info): Update.
12500 (update_specialized_profile): Update.
12501 (perhaps_add_new_callers): Update.
12502 (decide_about_value): Update.
12503 (ipa_cp_c_finalize): Update.
12504 * ipa-devirt.c (struct odr_type_warn_count): Update.
12505 (struct decl_warn_count): Update.
12506 (struct final_warning_record): Update.
12507 (possible_polymorphic_call_targets): Update.
12508 (ipa_devirt): Update.
12509 * ipa-fnsummary.c (redirect_to_unreachable): Update.
12510 * ipa-icf.c (sem_function::merge): Update.
12511 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
12512 * ipa-inline.c (compute_uninlined_call_time): Update.
12513 (compute_inlined_call_time): Update.
12514 (want_inline_small_function_p): Update.
12515 (want_inline_self_recursive_call_p): Update.
12516 (edge_badness): Update.
12517 (lookup_recursive_calls): Update.
12518 (recursive_inlining): Update.
12519 (inline_small_functions): Update.
12520 (dump_overall_stats): Update.
12521 (dump_inline_stats): Update.
12522 * ipa-profile.c (ipa_profile_generate_summary): Update.
12523 (ipa_propagate_frequency): Update.
12524 (ipa_profile): Update.
12525 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12526 * ipa-utils.c (ipa_merge_profiles): Update.
12527 * loop-doloop.c (doloop_modify): Update.
12528 * loop-unroll.c (report_unroll): Update.
12529 (unroll_loop_runtime_iterations): Update.
12530 * lto-cgraph.c (lto_output_edge): Update.
12531 (lto_output_node): Update.
12532 (input_node): Update.
12533 (input_edge): Update.
12534 (merge_profile_summaries): Update.
12535 * lto-streamer-in.c (input_cfg): Update.
12536 * lto-streamer-out.c (output_cfg): Update.
12537 * mcf.c (create_fixup_graph): Update.
12538 (adjust_cfg_counts): Update.
12539 (sum_edge_counts): Update.
12540 * modulo-sched.c (sms_schedule): Update.
12541 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
12542 * predict.c (maybe_hot_count_p): Update.
12543 (probably_never_executed): Update.
12544 (dump_prediction): Update.
12545 (combine_predictions_for_bb): Update.
12546 (propagate_freq): Update.
12547 (handle_missing_profiles): Update.
12548 (counts_to_freqs): Update.
12549 (rebuild_frequencies): Update.
12550 (force_edge_cold): Update.
12551 * predict.h: Include profile-count.h
12552 (maybe_hot_count_p, counts_to_freqs): UPdate.
12553 * print-rtl-function.c: Do not include cfg.h
12554 * print-rtl.c: Include basic-block.h
12555 * profile-count.c: New file.
12556 * profile-count.h: New file.
12557 * profile.c (is_edge_inconsistent): Update.
12558 (correct_negative_edge_counts): Update.
12559 (is_inconsistent): Update.
12560 (set_bb_counts): Update.
12561 (read_profile_edge_counts): Update.
12562 (compute_frequency_overlap): Update.
12563 (compute_branch_probabilities): Update; Initialize and deinitialize
12565 (branch_prob): Update.
12566 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
12567 (edge_gcov_count): New.
12568 (bb_gcov_count): New.
12569 * shrink-wrap.c (try_shrink_wrapping): Update.
12570 * tracer.c (better_p): Update.
12571 * trans-mem.c (expand_transaction): Update.
12572 (ipa_tm_insert_irr_call): Update.
12573 (ipa_tm_insert_gettmclone_call): Update.
12574 * tree-call-cdce.c: Update.
12575 * tree-cfg.c (gimple_duplicate_sese_region): Update.
12576 (gimple_duplicate_sese_tail): Update.
12577 (gimple_account_profile_record): Update.
12578 (execute_fixup_cfg): Update.
12579 * tree-inline.c (copy_bb): Update.
12580 (copy_edges_for_bb): Update.
12581 (initialize_cfun): Update.
12582 (freqs_to_counts): Update.
12583 (copy_cfg_body): Update.
12584 (expand_call_inline): Update.
12585 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
12586 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
12587 (try_unroll_loop_completely): Update.
12588 (try_peel_loop): Update.
12589 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
12590 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
12591 * tree-ssa-loop-split.c (connect_loops): Update.
12592 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
12593 * tree-ssa-reassoc.c (branch_fixup): Update.
12594 * tree-ssa-tail-merge.c (replace_block_by): Update.
12595 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
12596 (compute_path_counts): Update.
12597 (update_profile): Update.
12598 (recompute_probabilities): Update.
12599 (update_joiner_offpath_counts): Update.
12600 (estimated_freqs_path): Update.
12601 (freqs_to_counts_path): Update.
12602 (clear_counts_path): Update.
12603 (ssa_fix_duplicate_block_edges): Update.
12604 (duplicate_thread_path): Update.
12605 * tree-switch-conversion.c (case_bit_test_cmp): Update.
12606 (struct switch_conv_info): Update.
12607 * tree-tailcall.c (decrease_profile): Update.
12608 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
12609 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
12610 * value-prof.c (check_counter): Update.
12611 (gimple_divmod_fixed_value): Update.
12612 (gimple_mod_pow2): Update.
12613 (gimple_mod_subtract): Update.
12614 (gimple_ic_transform): Update.
12615 (gimple_stringop_fixed_value): Update.
12616 * value-prof.h (gimple_ic): Update.
12618 2017-06-02 Carl Love <cel@us.ibm.com>
12620 * config/rs6000/rs6000-c: Add support for built-in functions
12621 vector double vec_doublee (vector signed int);
12622 vector double vec_doublee (vector unsigned int);
12623 vector double vec_doublee (vector float);
12624 vector double vec_doubleh (vector signed int);
12625 vector double vec_doubleh (vector unsigned int);
12626 vector double vec_doubleh (vector float);
12627 vector double vec_doublel (vector signed int);
12628 vector double vec_doublel (vector unsigned int);
12629 vector double vec_doublel (vector float);
12630 vector double vec_doubleo (vector signed int);
12631 vector double vec_doubleo (vector unsigned int);
12632 vector double vec_doubleo (vector float);.
12633 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
12634 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
12636 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
12637 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
12638 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
12640 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
12641 vec_doublel, vec_doubleh.
12642 * doc/extend.texi: Update the built-in documentation file for the
12643 new built-in functions.
12645 2017-06-02 David Malcolm <dmalcolm@redhat.com>
12648 * ipa-inline-analysis.c (free_growth_caches): Set
12649 edge_removal_hook_holder to NULL after removing it.
12651 2017-06-02 Sudakshina Das <sudi.das@arm.com>
12653 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
12654 comparision with zero.
12656 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
12657 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12658 for early expansion of vec_min and vec_max builtins.
12659 (builtin_function_type): Add min/max unsigned variants to those
12660 identified as having unsigned arguments.
12662 2017-06-02 Olivier Hainque <hainque@adacore.com>
12664 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
12666 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12668 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
12669 Use VALL_F16 iterator rather than VALL.
12671 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12673 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
12674 Emit CBNZ inside loop when doing a strong exchange and comparing
12675 against zero. Generate the CC flags after the loop.
12677 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
12679 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
12680 (dl_section_ref): New.
12681 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
12682 On AIX, append an expression to subtract the size of the
12683 section length to dl_section_ref.
12685 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
12687 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
12688 for early expansion of vector absolute builtins.
12690 2017-06-02 Richard Biener <rguenther@suse.de>
12692 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
12693 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
12695 2017-06-02 Richard Biener <rguenther@suse.de>
12697 PR tree-optimization/80948
12698 * tree-tailcall.c (find_tail_calls): Track stmts to move in
12699 stmt order as well.
12701 2017-06-02 Richard Biener <rguenther@suse.de>
12703 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
12705 * tree-vect-stmts.c (process_use): Do not mark backedge defs
12706 for inductions as relevant.
12708 2017-06-02 Richard Biener <rguenther@suse.de>
12710 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
12711 (vectorizable_induction): ... this. Remove dead code.
12713 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
12715 * builtins. (expand_builtin_alloca): Remove second parameter and
12716 infer its value from the first parameter instead.
12717 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
12719 2017-06-02 Jakub Jelinek <jakub@redhat.com>
12721 PR rtl-optimization/80903
12722 * loop-doloop.c (add_test): Unshare sequence.
12724 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
12726 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
12728 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
12730 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
12732 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
12733 xlogue_layout::get_instance, logue_layout::xlogue_layout,
12734 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
12735 (xlogue_layout::get_stub_rtx): Make static.
12736 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
12737 (xlogue_layout::compute_stub_managed_regs): Rename to...
12738 (xlogue_layout::count_stub_managed_regs): ...this.
12739 (xlogue_layout::is_stub_managed_reg): New function.
12740 (xlogue_layout::m_stub_names): Rename to...
12741 (xlogue_layout::s_stub_names): ...this, make static.
12742 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
12743 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
12744 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
12745 xlogue_layout::s_stub_names): Instantiate statics.
12746 (stub_managed_regs): Remove.
12747 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
12748 (disable_call_ms2sysv_xlogues): Rename to...
12749 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
12750 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
12752 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
12753 change after reload_completed.
12754 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
12756 (ix86_expand_prologue): Likewise.
12757 (ix86_expand_epilogue): Likewise.
12758 (ix86_expand_split_stack_prologue): Likewise.
12759 (ix86_compute_frame_layout): Remove frame parameter ...
12760 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
12761 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
12763 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
12764 (ix86_frame): Move from here ...
12765 * config/i386/i386.h (ix86_frame): ... to here.
12766 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
12767 complete ix86_frame data structure instead. Remove some_ld_name.
12769 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
12771 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
12772 symbols that hold a DECL_VALUE_EXPR.
12774 2017-06-01 Martin Jambor <mjambor@suse.cz>
12776 PR tree-optimization/80898
12777 * tree-sra.c (process_subtree_disqualification): Removed.
12778 (disqualify_candidate): Do not acll
12779 process_subtree_disqualification.
12780 (subtree_mark_written_and_enqueue): New function.
12781 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
12782 RHS has been disqualified and re-queue LHS if necessary. Apart
12783 from that, ignore disqualified RHS.
12785 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12787 * config/s390/s390.c (s390_emit_epilogue): Disable early return
12788 address fetch for z10 or later.
12790 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12792 * config/arc/arc.md (tst_movb): Add guard when splitting.
12794 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12796 * config/arc/arc.c (arc_can_eliminate): Test against
12797 arc_frame_pointer_needed.
12799 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12801 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
12802 to prevent store reordering.
12803 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
12804 (type): Add block type.
12805 (stack_tie): Define special instruction to be used in
12808 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12810 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
12811 constraint. It is not valid for the pattern.
12812 (noncommutative_binary_comparison): Likewise.
12814 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12816 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
12819 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12821 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
12822 be used by the reg-alloc.
12824 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12826 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
12827 reg-alloc when having mul64 or mul32x16 instructions.
12828 (mulsidi3): Likewise.
12829 (umulsidi3): Likewise.
12830 (mulsi32x16): New pattern.
12831 (mulsi64): Likewise.
12832 (mulsidi64): Likewise.
12833 (umulsidi64): Likewise.
12834 (MUL32x16_REG): Define.
12835 (mul64_600): Use MUL32x16_REG.
12836 (mac64_600): Likewise.
12837 (umul64_600): Likewise.
12838 (umac64_600): Likewise.
12840 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
12842 * config/arc/arc.md (mulsi3_700): Make it commutative.
12844 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
12846 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
12848 (*sign_extendsidi2_insn): Likewise for movstosw.
12850 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
12852 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
12853 the type of the input discriminant value. Convert the
12854 discriminant value of signedness vary.
12856 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
12858 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
12859 Add to -Wall section.
12861 2017-06-01 Richard Biener <rguenther@suse.de>
12863 PR middle-end/66313
12864 * fold-const.c (fold_plusminus_mult_expr): If the factored
12865 factor may be zero use a wrapping type for the inner operation.
12866 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
12867 and handle moved defs.
12868 (process_assignment): Properly guard the unary op case. Return a
12869 tri-state indicating that moving the stmt before the call may allow
12870 to continue. Pass through to_move.
12871 (find_tail_calls): Handle moving unrelated defs before
12874 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
12877 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
12878 splitter result in the canonical way.
12880 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
12882 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
12883 also for 32bit target. Update insn attributes.
12884 (zero-extendsidi2 splitter): Allow all registers for operand 1.
12886 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
12888 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
12889 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
12890 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
12891 (_mm_maskz_min_ss): New intrinsics.
12893 2017-05-31 Martin Liska <mliska@suse.cz>
12895 * tree-vect-loop.c (vect_create_epilog_for_reduction):
12896 Change comment style to one we normally use.
12897 (vectorizable_reduction): Likewise.
12898 (vectorizable_induction): Likewise.
12899 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12900 (vectorizable_call): Likewise.
12901 (vectorizable_simd_clone_call): Likewise.
12902 (vectorizable_conversion): Likewise.
12903 (vectorizable_assignment): Likewise.
12904 (vectorizable_shift): Likewise.
12905 (vectorizable_operation): Likewise.
12906 (vectorizable_store): Likewise.
12907 (vectorizable_load): Likewise.
12908 * tree-vectorizer.h: Likewise.
12910 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
12912 * passes.c (emergency_dump_function): New.
12913 * tree-pass.h (emergency_dump_function): Declare.
12914 * plugin.c (plugins_internal_error_function): Remove.
12915 * plugin.h (plugins_internal_error_function): Remove declaration.
12916 * toplev.c (internal_error_function): New static function. Use it...
12917 (general_init): ...here.
12919 2017-05-31 Graham Markall <graham.markall@embecosm.com>
12921 * config/arc/arc.c (arc_print_operand): Handle constant operands.
12922 (arc_rtx_costs): Add costs for new patterns.
12923 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
12924 * config/arc/predicates.md: Add _1_2_3_operand predicate.
12926 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
12928 * tree-ssa-strlen.c (get_next_strinfo): New function.
12929 (get_stridx_plus_constant): Use it.
12930 (zero_length_string): Likewise.
12931 (adjust_related_strinfos): Likewise.
12932 (adjust_last_stmt): Likewise.
12934 2017-05-31 Richard Biener <rguenther@suse.de>
12937 * config/i386/i386.c (ix86_expand_builtin): Remove assert
12938 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
12940 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
12942 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
12943 loop_vinfo argument and use of dependence distance vectors.
12944 Check instead whether the two references differ only in their
12945 initial value and assume that they have the same alignment if the
12946 difference is a multiple of the vector alignment.
12947 (vect_analyze_data_refs_alignment): Update call accordingly.
12949 2017-05-31 Martin Liska <mliska@suse.cz>
12952 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
12954 2017-05-31 Bin Cheng <bin.cheng@arm.com>
12956 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
12957 (create_intersect_range_checks): Move from ...
12958 * tree-data-ref.c (create_intersect_range_checks_index)
12959 (create_intersect_range_checks): ... to here.
12960 (create_runtime_alias_checks): New function factored from ...
12961 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
12962 here. Call above function.
12963 * tree-data-ref.h (create_runtime_alias_checks): New function.
12965 2017-05-31 Bin Cheng <bin.cheng@arm.com>
12967 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
12968 segment length for dr_b and compute it in wide_int.
12970 2017-05-31 Richard Biener <rguenther@suse.de>
12972 PR tree-optimization/80906
12973 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
12974 and pass through iv_map.
12975 (copy_bb_and_scalar_dependences): Adjust.
12976 (translate_pending_phi_nodes): Likewise.
12977 (copy_loop_close_phi_args): Handle code-generating IVs instead
12980 2017-05-30 David Malcolm <dmalcolm@redhat.com>
12982 * diagnostic-color.c (color_dict): Add "type-diff".
12983 (parse_gcc_colors): Update comment.
12984 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
12985 -fdiagnostics-show-template-tree and -fno-elide-type.
12986 (GCC_COLORS): Add type-diff to example.
12988 (-fdiagnostics-show-template-tree): New.
12989 (-fno-elide-type): New.
12990 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
12991 the pp_format_decoder callback. Call any m_format_postprocessor's
12993 (pretty_printer::pretty_printer): Initialize
12994 m_format_postprocessor.
12995 (pretty_printer::~pretty_printer): Delete any
12996 m_format_postprocessor.
12997 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
12998 (class format_postprocessor): New class.
12999 (struct pretty_printer::format_decoder): Document the new parameters.
13000 (struct pretty_printer::m_format_postprocessor): New field.
13001 * tree-diagnostic.c (default_tree_printer): Update for new
13002 bool and const char ** params.
13003 * tree-diagnostic.h (default_tree_printer): Likewise.
13005 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
13007 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
13008 (lwa_operand): Delete rs6000_gen_cell_microcode test.
13009 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
13010 rs6000_gen_cell_microcode code.
13011 (rs6000_final_prescan_insn): Delete.
13012 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
13013 "warn-cell-microcode" entries.
13014 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
13015 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
13016 throughout. Change cc_reg_not_micro_cr0_operand to
13017 cc_reg_not_cr0_operand throughout.
13018 (*extendhi<mode>2_noload): Delete.
13019 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
13020 (mwarn-cell-microcode): Delete.
13021 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
13022 -mgen-cell-microcode and -mwarn-cell-microcode.
13024 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
13027 * config/i386/constraints.md (Yd): New constraint.
13029 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
13030 and (?Yd, r) alternatives. Update insn attributes.
13031 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
13032 and (?*Yd, r) alternatives. Update insn attributes.
13033 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
13035 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
13037 * gimplify.c (gimplify_modify_expr): Don't create a
13038 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
13041 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
13043 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
13045 2017-05-30 Richard Biener <rguenther@suse.de>
13047 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
13048 and reduc_def fields.
13049 (STMT_VINFO_REDUC_TYPE): New define.
13050 (STMT_VINFO_REDUC_DEF): Likewise.
13051 (vect_force_simple_reduction): Adjust prototype.
13052 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
13053 (vect_is_simple_reduction): Remove check_reduction argument.
13054 (vect_force_simple_reduction): Adjust and set
13055 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13056 (vectorizable_reduction): Do not re-do reduction analysis
13057 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
13058 * tree-parloops.c (gather_scalar_reductions): Adjust.
13060 2017-05-30 Richard Biener <rguenther@suse.de>
13062 PR middle-end/80901
13063 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
13066 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
13068 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
13069 Introduce unknown_misalignment parameter and remove vf.
13070 (vect_peeling_hash_get_lowest_cost):
13071 Pass unknown_misalignment parameter.
13072 (vect_enhance_data_refs_alignment):
13073 Fix unsupportable data ref treatment.
13075 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13077 * tree-vect-data-refs.c (vect_get_data_access_cost):
13078 Workaround for SLP handling.
13079 (vect_enhance_data_refs_alignment):
13080 Compute costs for doing no peeling at all, compare to the best
13081 peeling costs so far and avoid peeling if cheaper.
13083 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13085 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
13086 Return peeling info and set costs to zero for unlimited cost
13088 (vect_enhance_data_refs_alignment): Also inspect all datarefs
13089 with unknown misalignment. Compute and costs for unknown
13090 misalignment, compare them to the costs for known misalignment
13091 and choose the cheapest for peeling.
13093 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13095 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
13096 (vect_get_peeling_costs_all_drs): Create function.
13097 (vect_peeling_hash_get_lowest_cost):
13098 Use vect_get_peeling_costs_all_drs.
13099 (vect_peeling_supportable): Create function.
13100 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
13102 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
13104 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
13105 DR_HAS_NEGATIVE_STEP.
13106 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
13107 (vect_enhance_data_refs_alignment): Use.
13108 (vect_duplicate_ssa_name_ptr_info): Use.
13109 * tree-vectorizer.h (dr_misalignment): Use.
13110 (known_alignment_for_access_p): Use.
13112 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13115 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
13117 (has_section_name): New function.
13119 2017-05-30 Martin Liska <mliska@suse.cz>
13122 * auto-profile.c (get_function_decl_from_block): Fix
13125 2017-05-30 Richard Biener <rguenther@suse.de>
13127 PR middle-end/80876
13128 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
13130 2017-05-30 Martin Liska <mliska@suse.cz>
13132 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
13133 * dumpfile.h (struct dump_file_info): Remove ctors.
13135 2017-05-30 Martin Liska <mliska@suse.cz>
13137 * predict.def: Fix GNU coding style.
13139 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13141 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13142 Mark 'to' argument with ATTRIBUTE_UNUSED.
13144 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
13146 * config/xtensa/xtensa.c (xtensa_emit_call): Use
13147 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
13148 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
13151 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
13153 * doc/install.texi (Options specification): Restore entry of
13154 --enable-sjlj-exceptions.
13156 2017-05-27 Michael Eager <eager@eagercon.com>
13159 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
13161 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
13163 * config/microblaze/microblaze.h
13164 (FIXED_REGISTERS): Update in macro.
13165 (CALL_USED_REGISTERS): Update in macro.
13167 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
13169 * doc/install.texi: Add links to macOS binary distributions.
13171 2017-05-27 Jakub Jelinek <jakub@redhat.com>
13175 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13177 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13179 2017-05-26 Martin Liska <mliska@suse.cz>
13181 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
13183 2017-05-26 Martin Liska <mliska@suse.cz>
13185 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
13186 always leading ';; '.
13187 (dump_bb_info): Likewise.
13188 (brief_dump_cfg): Likewise.
13189 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
13190 * dumpfile.c: Remove usage of TDF_VERBOSE.
13191 * dumpfile.h (enum dump_kind): Likewise.
13192 (dump_gimple_bb_header): Do not use TDF_COMMENT.
13193 * print-tree.c (debug_verbose): Remove.
13194 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
13195 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
13196 * tree-diagnostic.c (default_tree_printer): Replace
13197 TDF_DIAGNOSTIC with TDF_SLIM.
13199 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13201 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
13202 in parameter loop, rather than loop_vinfo.
13203 (create_intersect_range_checks): Ditto.
13204 (vect_create_cond_for_alias_checks): Update call to above functions.
13206 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13208 PR tree-optimization/80815
13209 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
13210 for merging runtime alias checks. Handle negative DR_STEPs.
13212 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13214 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
13216 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
13217 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
13218 out code pruning runtime alias checks.
13219 * tree-data-ref.c (prune_runtime_alias_test_list): New function
13220 factored out from above.
13221 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13223 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
13225 (prune_runtime_alias_test_list): New decalaration.
13227 2017-05-26 Bin Cheng <bin.cheng@arm.com>
13229 * tree-vect-data-refs.c (compare_tree): Rename and move ...
13230 * tree-data-ref.c (data_ref_compare_tree): ... to here.
13231 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
13232 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
13233 (operator==, comp_dr_with_seg_len_pair): Ditto.
13234 (vect_prune_runtime_alias_test_list): Ditto.
13236 2017-05-26 Martin Liska <mliska@suse.cz>
13239 * params.def: Bound partial-inlining-entry-probability param.
13241 2017-05-26 Marek Polacek <polacek@redhat.com>
13244 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
13247 2017-05-26 Richard Biener <rguenther@suse.de>
13249 PR tree-optimization/80842
13250 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
13253 2017-05-26 Richard Biener <rguenther@suse.de>
13255 PR tree-optimization/80844
13256 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
13258 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
13260 * doc/md.texi (Machine Constraints): Update x86 family
13261 machine constraints section to match 'config/i386/constraints.md'.
13263 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
13265 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
13267 2017-05-25 Nathan Sidwell <nathan@acm.org>
13269 * doc/invoke.texi (--enable-languages): Update documentation.
13271 2017-05-25 Martin Liska <mliska@suse.cz>
13273 * dumpfile.c: Add TDF_FOLDING.
13274 * dumpfile.h (enum dump_kind): Likewise.
13275 * genmatch.c (dt_simplify::gen_1): Use it.
13277 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13279 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
13281 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13283 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
13284 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
13286 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
13288 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
13289 * match.pd (X == C): Rewrite it here.
13290 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
13291 with_certain_nonzero_bits2): New predicates.
13292 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
13294 2017-05-24 Nathan Sidwell <nathan@acm.org>
13296 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
13299 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
13302 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13304 * config/powerpcspe: New port. Files are copied from the rs6000
13305 port, with "rs6000" in filenames replaced by "powerpcspe".
13307 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
13309 PR rtl-optimization/80754
13310 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
13312 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
13314 * config/sparc/sparc.md (length): Return the correct value for -mflat
13315 sibcalls to match output_sibcall.
13317 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
13321 * config/rs6000/rs6000.c (struct machine_function): Add new field
13323 (rs6000_get_separate_components): Init that field, use it.
13324 (rs6000_components_for_bb): Use the field.
13326 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
13328 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
13330 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
13332 PR middle-end/80823
13333 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
13335 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13338 * config/s390/s390.c (s390_check_qrst_address): Check incoming
13339 address against address_operand predicate.
13340 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
13342 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
13344 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
13345 parameters passed indirectly.
13347 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
13349 * config/i386/i386.md (*movdi_internal): Remove SSE4
13350 alternative 18 (?r, *v). Update insn attributes.
13351 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
13352 Update insn attributes.
13353 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
13354 Update insn attributes.
13355 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
13356 alternative 1 (r, v). Remove isa attribute.
13357 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
13358 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
13359 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
13361 2017-05-23 Tom de Vries <tom@codesourcery.com>
13363 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
13366 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13368 * cgraphunit.c (symbol_table::process_new_functions): Update.
13369 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
13370 (inline_generate_summary): Rename to ...
13371 (ipa_fn_summary_generate): ... this one.
13372 (inline_read_summary): Rename to ...
13373 (ipa_fn_summary_read): ... this one.
13374 (inline_write_summary): Rename to ...
13375 (ipa_fn_summary_write): ... this one.
13376 (inline_free_summary): Rename to ...
13377 (ipa_free_fn_summary): ... this one.
13378 (pass_data_local_fn_summary, pass_local_fn_summary,
13379 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
13380 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
13381 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
13382 make_pass_ipa_fn_summary): New.
13383 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
13384 inline_write_summary, inline_free_summary): Remove.
13385 (ipa_free_fn_summary) : New.
13386 * ipa-inline.c (ipa_inline): Update.
13387 (pass_ipa_inline): Do not generate summaries.
13388 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
13390 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
13391 and add pass_ipa_fn_summary.
13392 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
13394 (make_pass_inline_parameters): Remove.
13396 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
13398 * omp-low.c (struct omp_context): Remove "default_kind" member.
13401 * omp-offload.c (execute_oacc_device_lower): Remove the
13402 parallelism dimensions function attributes for unparallelized
13403 OpenACC kernels constructs.
13405 2017-05-23 Martin Liska <mliska@suse.cz>
13407 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
13409 (cgraph_edge::make_speculative): Likewise.
13410 (cgraph_edge::resolve_speculation): Likewise.
13411 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
13412 (cgraph_node::dump): Likewise.
13413 * cgraph.h: Likewise.
13414 * cgraphunit.c (analyze_functions): Likewise.
13415 (symbol_table::compile): Likewise.
13416 * ipa-cp.c (print_all_lattices): Likewise.
13417 (determine_versionability): Likewise.
13418 (initialize_node_lattices): Likewise.
13419 (ipcp_verify_propagated_values): Likewise.
13420 (estimate_local_effects): Likewise.
13421 (update_profiling_info): Likewise.
13422 (create_specialized_node): Likewise.
13423 (perhaps_add_new_callers): Likewise.
13424 (decide_about_value): Likewise.
13425 (decide_whether_version_node): Likewise.
13426 (identify_dead_nodes): Likewise.
13427 (ipcp_store_bits_results): Likewise.
13428 * ipa-devirt.c (dump_targets): Likewise.
13429 (ipa_devirt): Likewise.
13430 * ipa-icf.c (sem_item::dump): Likewise.
13431 (sem_function::equals): Likewise.
13432 (sem_variable::equals): Likewise.
13433 (sem_item_optimizer::read_section): Likewise.
13434 (sem_item_optimizer::execute): Likewise.
13435 (congruence_class::dump): Likewise.
13436 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
13437 (dump_inline_summary): Likewise.
13438 (estimate_node_size_and_time): Likewise.
13439 (inline_analyze_function): Likewise.
13440 * ipa-inline-transform.c (inline_call): Likewise.
13441 * ipa-inline.c (report_inline_failed_reason): Likewise.
13442 (want_early_inline_function_p): Likewise.
13443 (edge_badness): Likewise.
13444 (update_edge_key): Likewise.
13445 (inline_small_functions): Likewise.
13446 * ipa-profile.c (ipa_profile): Likewise.
13447 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
13448 (ipa_make_edge_direct_to_target): Likewise.
13449 (remove_described_reference): Likewise.
13450 (ipa_impossible_devirt_target): Likewise.
13451 (propagate_controlled_uses): Likewise.
13452 (ipa_print_node_params): Likewise.
13453 (ipcp_transform_function): Likewise.
13454 * ipa-pure-const.c (pure_const_read_summary): Likewise.
13455 (propagate_pure_const): Likewise.
13456 * ipa-reference.c (generate_summary): Likewise.
13457 (read_write_all_from_decl): Likewise.
13458 (propagate): Likewise.
13459 (ipa_reference_read_optimization_summary): Likewise.
13460 * ipa-utils.c (ipa_merge_profiles): Likewise.
13461 * ipa.c (walk_polymorphic_call_targets): Likewise.
13462 (symbol_table::remove_unreachable_nodes): Likewise.
13463 (ipa_single_use): Likewise.
13464 * passes.c (execute_todo): Likewise.
13465 * predict.c (drop_profile): Likewise.
13466 * symtab.c (symtab_node::get_dump_name): New function.
13467 (symtab_node::dump_name): Likewise.
13468 (symtab_node::dump_asm_name): Likewise.
13469 (symtab_node::dump_references): Likewise.
13470 (symtab_node::dump_referring): Likewise.
13471 (symtab_node::dump_base): Likewise.
13472 (symtab_node::debug_symtab): Likewise.
13473 * tree-sra.c (convert_callers_for_node): Likewise.
13474 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
13475 * value-prof.c (init_node_map): Likewise.
13477 2017-05-23 Martin Liska <mliska@suse.cz>
13479 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
13480 and symtab_node::debug_symtab to symbol_table::debug.
13481 * cgraphunit.c (analyze_functions): Use the renamed function.
13482 (symbol_table::compile): Likewise.
13483 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
13484 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
13485 * passes.c (execute_todo): Likewise.
13486 * symtab.c (symbol_table::dump): New function.
13487 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
13489 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13491 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
13492 that nonconst implies exec.
13494 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
13496 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
13497 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
13498 (inline_edge_summary_vec): Turn into ...
13499 (ipa_call_summaries): ... this one.
13500 (redirect_to_unreachable, edge_set_predicate,
13501 evaluate_properties_for_edge, inline_summary_alloc,
13502 reset_ipa_call_summary, reset_inline_summary,
13503 inline_summary_t::duplicate): Update.
13504 (inline_edge_duplication_hook): Turn to ...
13505 (ipa_call_summary_t::duplicate): ... this one.
13506 (inline_edge_removal_hook): Turn to ...
13507 (ipa_call_summary_t::remove): ... this one.
13508 (dump_inline_edge_summary): Turn to ...
13509 (dump_ipa_call_summary): ... this one.
13510 (estimate_function_body_sizes): Update.
13511 (inline_update_callee_summaries): Update.
13512 (remap_edge_change_prob): Update.
13513 (remap_edge_summaries): Update.
13514 (inline_merge_summary): Update.
13515 (do_estimate_edge_time): Update.
13516 (inline_generate_summary): Update.
13517 (inline_read_section): Update.
13518 (inline_read_summary): Update.
13519 (inline_free_summary): Update.
13520 * ipa-inline.c (can_inline_edge_p): Update.
13521 (compute_inlined_call_time): Update.
13522 (want_inline_small_function_p): Update.
13523 (edge_badness): Update.
13524 (early_inliner): Update.
13525 * ipa-inline.h (inline_edge_summary): Turn to ...
13526 (ipa_call_summary): ... this one.
13527 (ipa_call_summary_t): New class.
13528 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
13529 (ipa_call_summaries): New.
13530 (inline_edge_summary): Remove.
13531 (estimate_edge_growth): Update.
13532 * ipa-profile.c (ipa_propagate_frequency_1): Update.
13533 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13534 * ipa-split.c (execute_split_functions): Update.
13535 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
13537 2017-05-23 Tom de Vries <tom@codesourcery.com>
13539 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13540 attributes): Document rdrand effective target.
13542 2017-05-23 Tom de Vries <tom@codesourcery.com>
13544 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
13545 attributes): Sort alphabetically.
13547 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
13549 * config/avr/genmultilib.awk: Use gsub instead of gensub.
13551 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
13554 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
13555 V2DF/V2DI splat into two separate patterns, one that handles
13556 registers, and the other that only handles memory. Drop support
13557 for splatting from a GPR on ISA 2.07 and then splitting the
13558 splat into direct move and splat.
13559 (vsx_splat_<mode>_reg): Likewise.
13560 (vsx_splat_<mode>_mem): Likewise.
13562 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
13564 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
13566 2017-05-22 Jakub Jelinek <jakub@redhat.com>
13568 PR middle-end/80809
13569 * omp-low.c (finish_taskreg_remap): New function.
13570 (finish_taskreg_scan): If unit size of ctx->record_type
13571 is non-constant, unshare the size expression and replace
13572 decls in it with possible outer var refs.
13574 PR middle-end/80809
13575 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
13576 GOVD_SHARED rather than GOVD_PRIVATE with it.
13577 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
13578 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
13580 PR middle-end/80853
13581 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
13582 as last argument to build_outer_var_ref for pointer bases of array
13583 section reductions.
13585 2017-05-19 Martin Sebor <msebor@redhat.com>
13587 * print-tree.c (print_node): Print DECL_READ_P flag.
13589 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13591 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
13592 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
13593 * cgraph.c: Likewise.
13594 * cgraphunit.c: Likewise.
13595 * gengtype.c: Likewise.
13596 * ipa-cp.c: Likewise.
13597 * ipa-devirt.c: Likewise.
13598 * ipa-icf.c: Likewise.
13599 * ipa-predicate.c: Likewise.
13600 * ipa-profile.c: Likewise.
13601 * ipa-prop.c: Likewise.
13602 * ipa-split.c: Likewise.
13604 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
13605 edge_predicate_pool, dump_inline_hints,
13606 inline_summary::account_size_time, redirect_to_unreachable,
13607 edge_set_predicate, set_hint_predicate,
13608 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
13609 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
13610 inline_summary_t::remove, remap_hint_predicate_after_duplication,
13611 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
13612 ipa_call_summary_t::remove, initialize_growth_caches,
13613 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
13614 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
13615 mark_modified, unmodified_parm_1, unmodified_parm,
13616 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
13617 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
13618 compute_bb_predicates, will_be_nonconstant_expr_predicate,
13619 will_be_nonconstant_predicate, record_modified_bb_info,
13620 get_minimal_bb, record_modified, param_change_prob,
13621 phi_result_unknown_predicate, predicate_for_phi_result,
13622 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
13623 estimate_function_body_sizes, compute_inline_parameters,
13624 compute_inline_parameters_for_curren, pass_data_inline_parameters,
13625 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
13626 inline_update_callee_summaries, remap_edge_change_prob,
13627 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
13628 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
13629 inline_analyze_function, inline_summary_t::insert,
13630 inline_generate_summary, read_ipa_call_summary, inline_read_section,
13631 inline_read_summary, write_ipa_call_summary, inline_write_summary,
13632 inline_free_summary): Move to ipa-fnsummary.h
13633 (predicate_t): Remove.
13634 * ipa-fnsummary.c: New file.
13635 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
13636 (enum inline_hints_vals, inline_hints, agg_position_info,
13637 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
13638 inline_summaries, ipa_call_summary, ipa_call_summary_t,
13639 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
13640 dump_inline_summary, dump_inline_hints, inline_generate_summary,
13641 inline_read_summary, inline_write_summary, inline_free_summary,
13642 inline_analyze_function, initialize_inline_failed,
13643 inline_merge_summary, inline_update_overall_summary,
13644 compute_inline_parameters): Move to ipa-fnsummary.h
13645 * ipa-fnsummary.h: New file.
13646 * ipa-inline-transform.h: Include ipa-inline.h.
13647 * ipa-inline.c: LIkewise.
13649 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13651 * ipa-inline.c (edge_badness): Use inlined_time instead of
13652 inline_summaries->get.
13654 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13656 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
13658 2017-05-22 Nathan Sidwell <nathan@acm.org>
13660 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
13661 (fdump-lang): Document 'raw' option.
13662 * dumpfile.h (TDI_tu): Delete.
13663 * dumpfile.c (dump_files): Remove translation-unit.
13664 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
13666 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
13668 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
13669 command option from $(AWK) call.
13670 * config/avr/genmultilib.awk: Simplify and rewrite so that it
13671 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
13672 [FORMAT]: Remove handling of variable.
13673 * config/avr/t-multilib: Regenerate.
13675 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13677 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
13679 (dump_inline_summary): Do not print self_time.
13680 (estimate_function_body_sizes): Do not set self_time.
13681 (compute_inline_parameters): Likewise.
13682 (inline_read_section, inline_write_summary): Do not stream self_time.
13683 * ipa-inline.h (inline_summary): Drop self_time.
13685 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
13687 * ipa-inline-analysis.c (account_size_time): Rename to ...
13688 (inline_summary::account_size_time): ... this one.
13689 (reset_ipa_call_summary): Turn to ...
13690 (ipa_call_summary::reset): ... this one.
13691 (reset_inline_summary): Turn to ...
13692 (inline_summary::reset): ... this one.
13693 (inline_summary_t::remove): Update.
13694 (inline_summary_t::duplicate): Update.
13695 (ipa_call_summary_t::remove): Update.
13696 (dump_inline_summary): Update.
13697 (estimate_function_body_sizes): Update.
13698 (compute_inline_parameters): Update.
13699 (estimate_node_size_and_time): Update.
13700 (inline_merge_summary): Update.
13701 (inline_update_overall_summary): Update.
13702 (inline_read_section): Update.
13703 (inline_write_summary): Update.
13704 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
13705 add account_size_time and reset member functions.
13706 (ipa_call_summary): Add reset function.
13707 * ipa-predicate.h (predicate::operator &): Constify.
13709 2017-05-22 Richard Biener <rguenther@suse.de>
13711 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
13713 2017-05-19 Jason Merrill <jason@redhat.com>
13715 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
13717 2017-05-19 Marek Polacek <polacek@redhat.com>
13720 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
13721 TYPE_OVERFLOW_WRAPS checks.
13723 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
13725 * tree-core.h (enum omp_clause_default_kind): Add
13726 "OMP_CLAUSE_DEFAULT_PRESENT".
13727 * tree-pretty-print.c (dump_omp_clause): Handle it.
13728 * gimplify.c (enum gimplify_omp_var_data): Add
13729 "GOVD_MAP_FORCE_PRESENT".
13730 (gimplify_adjust_omp_clauses_1): Map it to
13731 "GOMP_MAP_FORCE_PRESENT".
13732 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
13734 * gimplify.c (oacc_default_clause): Clarify.
13736 2017-05-19 Nathan Sidwell <nathan@acm.org>
13738 LANG_HOOK_REGISTER_DUMPS
13739 * toplev.c (general_init): Call register dump lang hook.
13740 * doc/invoke.texi: Document -fdump-lang option family.
13741 * dumpfile.c (dump_files): Remove class dump here.
13742 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
13743 * dumpfile.h (tree_dump_index): Remove TDI_class.
13744 * langhooks-def.h (lhd_register_dumps): Declare.
13745 (LANG_HOOKS_REGISTER_DUMPS): Define.
13746 (LANG_HOOKS_INITIALIZER): Add it.
13747 * langhooks.c (lhd_register_dumps): Define.
13748 * langhooks.h (struct lang_hooks): Add register_dumps.
13750 2017-05-19 Nathan Sidwell <nathan@acm.org>
13752 * context.h (context::set_passes): New.
13753 * context.c (context::context): Do not create pass manager.
13754 * toplev.c (general_init): Create pass manager here.
13756 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
13758 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
13759 use this splitter if two add or or instructions would also work for
13760 the constant we want to generate.
13762 2017-05-19 Richard Biener <rguenther@suse.de>
13765 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
13766 predicate evaluation.
13768 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
13770 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
13772 * ipa-inline.c (want_inline_small_function_p): Do not cast to
13775 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
13777 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
13778 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
13779 (inline_edge_summary_vec): Turn into ...
13780 (ipa_call_summaries): ... this one.
13781 (redirect_to_unreachable, edge_set_predicate,
13782 evaluate_properties_for_edge, inline_summary_alloc,
13783 reset_ipa_call_summary, reset_inline_summary,
13784 inline_summary_t::duplicate): Update.
13785 (inline_edge_duplication_hook): Turn to ...
13786 (ipa_call_summary_t::duplicate): ... this one.
13787 (inline_edge_removal_hook): Turn to ...
13788 (ipa_call_summary_t::remove): ... this one.
13789 (dump_inline_edge_summary): Turn to ...
13790 (dump_ipa_call_summary): ... this one.
13791 (estimate_function_body_sizes): Update.
13792 (inline_update_callee_summaries): Update.
13793 (remap_edge_change_prob): Update.
13794 (remap_edge_summaries): Update.
13795 (inline_merge_summary): Update.
13796 (do_estimate_edge_time): Update.
13797 (inline_generate_summary): Update.
13798 (inline_read_section): Update.
13799 (inline_read_summary): Update.
13800 (inline_free_summary): Update.
13801 * ipa-inline.c (can_inline_edge_p): Update.
13802 (compute_inlined_call_time): Update.
13803 (want_inline_small_function_p): Update.
13804 (edge_badness): Update.
13805 (early_inliner): Update.
13806 * ipa-inline.h (inline_edge_summary): Turn to ...
13807 (ipa_call_summary): ... this one.
13808 (ipa_call_summary_t): New class.
13809 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
13810 (ipa_call_summaries): New.
13811 (inline_edge_summary): Remove.
13812 (estimate_edge_growth): Update.
13813 * ipa-profile.c (ipa_propagate_frequency_1): Update.
13814 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
13815 * ipa-split.c (execute_split_functions): Update.
13816 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
13818 2017-05-19 Richard Biener <rguenther@suse.de>
13820 PR middle-end/80764
13821 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
13823 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
13825 * config/rs6000/rs6000.c (struct machine_function): Add field
13826 fpr_is_wrapped_separately.
13827 (rs6000_get_separate_components): Use 64 components. Handle the
13828 new FPR components.
13829 (rs6000_components_for_bb): Handle the FPR components.
13830 (rs6000_emit_prologue_components): Handle the FPR components.
13831 (rs6000_emit_epilogue_components): Handle the FPR components.
13832 (rs6000_set_handled_components): Handle the FPR components.
13833 (rs6000_emit_prologue): Don't output prologue code for those FPRs
13834 that are already separately shrink-wrapped.
13835 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
13836 that are already separately shrink-wrapped.
13838 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
13841 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
13844 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
13845 (define_peephole2 for Altivec d-form load): Add peepholes to catch
13846 cases where the register allocator uses a move and an offsettable
13847 memory operation to/from a FPR register on ISA 2.06/2.07.
13848 (define_peephole2 for Altivec d-form store): Likewise.
13850 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
13853 * config/i386/mmx.md (*mov<mode>_internal): Enable
13854 alternatives 11, 12, 13 and 14 also for 32bit targets.
13855 Remove alternatives 15, 16, 17 and 18.
13856 * config/i386/sse.md (vec_concatv2di): Change
13857 alternative (!x, *y) to (x, ?!*Yn).
13859 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
13861 * dumpfile.h (enum dump_kind): Remove stray comma.
13863 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
13865 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
13866 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
13867 predicate::num_conditions
13868 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
13869 (CHANGED): turn into predicate::changed.
13870 (agg_position_info): Move to ipa-predicate.h
13871 (add_condition, predicate::add_clause, predicate::operator &=,
13872 predicate::or_with, predicate::evaluate, predicate::probability,
13873 dump_condition, dump_clause, predicate::dump,
13874 predicate::remap_after_duplication, predicate::remap_after_inlining,
13875 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
13876 (evaluate_conditions_for_known_args): Update.
13877 (set_cond_stmt_execution_predicate): Update.
13878 * ipa-inline.h: Include ipa-predicate.h
13879 (condition, inline_param_summary, conditions, agg_position_info,
13880 predicate): Move to ipa-predicate.h
13881 * ipa-predicate.c: New file.
13882 * ipa-predicate.h: New file.
13884 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
13886 * final.c (leaf_function_p): Check we are not in a sequence.
13888 2017-05-18 Martin Liska <mliska@suse.cz>
13890 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
13891 * dumpfile.c (dump_register): Use new enum dump_kind.
13892 (get_dump_file_name): Likewise.
13893 (dump_enable_all): Likewise.
13894 (dump_switch_p_1): Likewise.
13895 (enable_rtl_dump_file): Remove usage of TDF_RTL.
13896 * dumpfile.h (enum dump_kind): New enum type.
13897 (struct dump_file_info): Create constructor and
13898 format fields and comments.
13899 * passes.c (pass_manager::register_one_dump_file):
13901 * statistics.c (statistics_early_init): Likewise.
13902 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
13903 TDF_TREE with TDF_SLIM.
13904 (gather_memory_references_ref): Likewise.
13906 2017-05-18 Martin Liska <mliska@suse.cz>
13908 * vec.h (struct vnull): Use it.
13910 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
13912 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
13913 (true_predicate, false_predicate, true_predicate_p,
13914 false_predicate_p): Remove.
13915 (single_cond_predicate, not_inlined_predicate): Turn to member function
13917 (add_condition): Update.
13918 (add_clause): Turn to...
13919 (predicate::add_clause): ... this one; update; allow passing NULL
13921 (and_predicates): Turn to ...
13922 (predicate::operator &=): ... this one.
13923 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
13924 (or_predicates): Turn to ...
13925 (predicate::or_with): ... this one.
13926 (evaluate_predicate): Turn to ...
13927 (predicate::evaluate): ... this one.
13928 (predicate_probability): Turn to ...
13929 (predicate::probability): ... this one.
13930 (dump_condition): Update.
13931 (dump_predicate): Turn to ...
13932 (predicate::dump): ... this one.
13933 (account_size_time): Update.
13934 (edge_set_predicate): Update.
13935 (set_hint_predicate): UPdate.
13936 (evaluate_conditions_for_known_args): Update.
13937 (evaluate_properties_for_edge): Update.
13938 (remap_predicate_after_duplication): Turn to...
13939 (predicate::remap_after_duplication): ... this one.
13940 (remap_hint_predicate_after_duplication): Update.
13941 (inline_summary_t::duplicate): UPdate.
13942 (dump_inline_edge_summary): Update.
13943 (dump_inline_summary): Update.
13944 (set_cond_stmt_execution_predicate): Update.
13945 (set_switch_stmt_execution_predicate): Update.
13946 (compute_bb_predicates): Update.
13947 (will_be_nonconstant_expr_predicate): Update.
13948 (will_be_nonconstant_predicate): Update.
13949 (phi_result_unknown_predicate): Update.
13950 (predicate_for_phi_result): Update.
13951 (array_index_predicate): Update.
13952 (estimate_function_body_sizes): Update.
13953 (estimate_node_size_and_time): Update.
13954 (estimate_ipcp_clone_size_and_time): Update.
13955 (remap_predicate): Rename to ...
13956 (predicate::remap_after_inlining): ... this one.
13957 (remap_hint_predicate): Update.
13958 (inline_merge_summary): Update.
13959 (inline_update_overall_summary): Update.
13960 (estimate_size_after_inlining): Update.
13961 (read_predicate): Rename to ...
13962 (predicate::stream_in): ... this one.
13963 (read_inline_edge_summary): Update.
13964 (write_predicate): Rename to ...
13965 (predicate::stream_out): ... this one.
13966 (write_inline_edge_summary): Update.
13967 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
13968 (clause_t): Turn to uint32_t
13969 (predicate): Turn to class; implement constructor and operators
13971 (size_time_entry): Update.
13972 (inline_summary): Update.
13973 (inline_edge_summary): Update.
13975 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
13977 * fold-const.c (fold_binary_loc): Move transformation...
13978 * match.pd (C - X CMP X): ... here.
13980 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
13982 * config/sparc/sparc.c (sparc_option_override): Set function
13983 alignment for -mcpu=niagara7 to 64 to match the I$ line.
13984 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
13986 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
13988 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
13990 2017-05-18 Marek Polacek <polacek@redhat.com>
13993 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
13994 (pass_ubsan::execute): Call gimple_assign_single_p instead of
13995 gimple_assign_load_p.
13997 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
13999 PR middle-end/80692
14000 * real.c (do_compare): Give decimal_do_compare preference over
14001 comparing just the signs.
14003 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
14005 * doc/md.texi (Canonicalization of Instructions): Describe the
14006 canonical form of instructions that inherently set a condition
14009 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
14011 PR middle-end/80775
14012 * tree-cfg.c: Move deletion of unreachable case statements to after
14013 the merging of consecutive case labels.
14015 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
14017 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
14018 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
14019 restoring of callee-saved registers.
14021 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
14023 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
14024 * config/visium/visium.c (single_set_and_flags): Likewise.
14025 * config/visium/visium.md (Substitutions): Likewise.
14027 2017-05-17 Martin Liska <mliska@suse.cz>
14029 * cfg.c: Introduce dump_flags_t type and
14030 use it instead of int type.
14032 * cfghooks.c: Likewise.
14033 * cfghooks.h (struct cfg_hooks): Likewise.
14034 * cfgrtl.c: Likewise.
14035 * cfgrtl.h: Likewise.
14036 * cgraph.c (cgraph_node::get_body): Likewise.
14037 * coretypes.h: Likewise.
14038 * domwalk.c: Likewise.
14039 * domwalk.h: Likewise.
14040 * dumpfile.c (struct dump_option_value_info): Likewise.
14041 (dump_enable_all): Likewise.
14042 (dump_switch_p_1): Likewise.
14043 (opt_info_switch_p): Likewise.
14044 * dumpfile.h (enum tree_dump_index): Likewise.
14045 (struct dump_file_info): Likewise.
14046 * genemit.c: Likewise.
14047 * generic-match-head.c: Likewise.
14048 * gengtype.c (open_base_files): Likewise.
14049 * gimple-pretty-print.c: Likewise.
14050 * gimple-pretty-print.h: Likewise.
14051 * graph.c (print_graph_cfg): Likewise.
14052 * graphite-scop-detection.c (dot_all_sese): Likewise.
14053 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
14054 * loop-unroll.c (report_unroll): Likewise.
14055 * passes.c (pass_manager::register_one_dump_file): Likewise.
14056 * print-tree.c: Likewise.
14057 * statistics.c: Likewise.
14058 * tree-cfg.c: Likewise.
14059 * tree-cfg.h: Likewise.
14060 * tree-dfa.c: Likewise.
14061 * tree-dfa.h: Likewise.
14062 * tree-dump.c (dump_function): Likewise.
14063 * tree-dump.h (struct dump_info): Likewise.
14064 * tree-pretty-print.c: Likewise.
14065 * tree-pretty-print.h: Likewise.
14066 * tree-ssa-live.c: Likewise.
14067 * tree-ssa-live.h: Likewise.
14068 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
14069 * tree-vect-loop.c: Likewise.
14070 * tree-vect-slp.c: Likewise.
14072 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
14073 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14075 PR tree-optimization/80457
14076 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
14077 of all arguments to a statement as scalar_to_vec operations.
14078 (vectorizable_call): Adjust call to vect_model_simple_cost for
14080 (vectorizable_conversion): Likewise.
14081 (vectorizable_assignment): Likewise.
14082 (vectorizable_shift): Likewise.
14083 (vectorizable_operation): Likewise.
14084 (vectorizable_comparison): Likewise.
14085 (vect_is_simple_cond): Record the def types for operands.
14086 (vectorizable_condition): Likewise, call vect_model_simple_cost.
14087 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
14088 for statement argument count.
14090 2017-05-16 Carl Love <cel@us.ibm.com>
14092 * config/rs6000/rs6000-c: Add support for built-in functions
14093 vector unsigned long long vec_bperm (vector unsigned long long,
14094 vector unsigned char)
14095 vector signed long long vec_mule (vector signed int,
14097 vector unsigned long long vec_mule (vector unsigned int,
14098 vector unsigned int)
14099 vector signed long long vec_mulo (vector signed int,
14101 vector unsigned long long vec_mulo (vector unsigned int,
14102 vector unsigned int)
14103 vector signed char vec_sldw (vector signed char,
14104 vector signed char,
14106 vector unsigned char vec_sldw (vector unsigned char,
14107 vector unsigned char,
14109 vector signed short vec_sldw (vector signed short,
14110 vector signed short,
14112 vector unsigned short vec_sldw (vector unsigned short,
14113 vector unsigned short,
14115 vector signed int vec_sldw (vector signed int,
14118 vector unsigned int vec_sldw (vector unsigned int,
14119 vector unsigned int,
14121 vector signed long long vec_sldw (vector signed long long,
14122 vector signed long long,
14124 vector unsigned long long vec_sldw (vector unsigned long long,
14125 vector unsigned long long,
14127 * config/rs6000/rs6000-c: Add support for built-in functions
14128 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
14129 * config/rs6000/altivec.h: Add defintion for vec_sldw.
14130 * doc/extend.texi: Update the built-in documentation for the
14131 new built-in functions.
14133 2017-05-16 Marek Polacek <polacek@redhat.com>
14137 * tree.c (save_expr): Don't fold the expression.
14139 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
14141 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
14142 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
14143 and (?*y,m). Update insn attributes.
14145 2017-05-16 Martin Liska <mliska@suse.cz>
14147 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
14148 flags argument of print_gimple_stmt, print_gimple_expr,
14149 print_generic_stmt and print_generic_expr.
14150 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
14151 * coretypes.h: Likewise.
14152 * except.c (dump_eh_tree): Likewise.
14153 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
14154 * gimple-pretty-print.h: Likewise.
14155 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
14156 (backprop::push_to_worklist): Likewise.
14157 (backprop::pop_from_worklist): Likewise.
14158 (backprop::process_use): Likewise.
14159 (backprop::intersect_uses): Likewise.
14160 (note_replacement): Likewise.
14161 * gimple-ssa-store-merging.c
14162 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
14163 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
14164 (pass_store_merging::execute): Likewise.
14165 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
14166 (ssa_base_cand_dump_callback): Likewise.
14167 (dump_incr_vec): Likewise.
14168 (replace_refs): Likewise.
14169 (replace_mult_candidate): Likewise.
14170 (create_add_on_incoming_edge): Likewise.
14171 (create_phi_basis): Likewise.
14172 (insert_initializers): Likewise.
14173 (all_phi_incrs_profitable): Likewise.
14174 (introduce_cast_before_cand): Likewise.
14175 (replace_one_candidate): Likewise.
14176 * gimplify.c (gimplify_expr): Likewise.
14177 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
14178 (set_rename): Likewise.
14179 (rename_uses): Likewise.
14180 (copy_loop_phi_nodes): Likewise.
14181 (add_close_phis_to_merge_points): Likewise.
14182 (copy_loop_close_phi_args): Likewise.
14183 (copy_cond_phi_args): Likewise.
14184 (graphite_copy_stmts_from_block): Likewise.
14185 (translate_pending_phi_nodes): Likewise.
14186 * graphite-poly.c (print_pdr): Likewise.
14187 (dump_gbb_cases): Likewise.
14188 (dump_gbb_conditions): Likewise.
14189 (print_scop_params): Likewise.
14190 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
14191 (build_cross_bb_scalars_use): Likewise.
14192 (gather_bbs::before_dom_children): Likewise.
14193 * hsa-dump.c (dump_hsa_immed): Likewise.
14194 * ipa-cp.c (print_ipcp_constant_value): Likewise.
14195 (get_replacement_map): Likewise.
14196 * ipa-inline-analysis.c (dump_condition): Likewise.
14197 (estimate_function_body_sizes): Likewise.
14198 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
14199 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
14200 * ipa-prop.c (ipa_dump_param): Likewise.
14201 (ipa_print_node_jump_functions_for_edge): Likewise.
14202 (ipa_modify_call_arguments): Likewise.
14203 (ipa_modify_expr): Likewise.
14204 (ipa_dump_param_adjustments): Likewise.
14205 (ipa_dump_agg_replacement_values): Likewise.
14206 (ipcp_modif_dom_walker::before_dom_children): Likewise.
14207 * ipa-pure-const.c (check_stmt): Likewise.
14208 (pass_nothrow::execute): Likewise.
14209 * ipa-split.c (execute_split_functions): Likewise.
14210 * omp-offload.c (dump_oacc_loop_part): Likewise.
14211 (dump_oacc_loop): Likewise.
14212 * trans-mem.c (tm_log_emit): Likewise.
14213 (tm_memopt_accumulate_memops): Likewise.
14214 (dump_tm_memopt_set): Likewise.
14215 (dump_tm_memopt_transform): Likewise.
14216 * tree-cfg.c (gimple_verify_flow_info): Likewise.
14217 (print_loop): Likewise.
14218 * tree-chkp-opt.c (chkp_print_addr): Likewise.
14219 (chkp_gather_checks_info): Likewise.
14220 (chkp_get_check_result): Likewise.
14221 (chkp_remove_check_if_pass): Likewise.
14222 (chkp_use_outer_bounds_if_possible): Likewise.
14223 (chkp_reduce_bounds_lifetime): Likewise.
14224 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
14225 (chkp_mark_completed_bounds): Likewise.
14226 (chkp_register_incomplete_bounds): Likewise.
14227 (chkp_mark_invalid_bounds): Likewise.
14228 (chkp_maybe_copy_and_register_bounds): Likewise.
14229 (chkp_build_returned_bound): Likewise.
14230 (chkp_get_bound_for_parm): Likewise.
14231 (chkp_build_bndldx): Likewise.
14232 (chkp_get_bounds_by_definition): Likewise.
14233 (chkp_generate_extern_var_bounds): Likewise.
14234 (chkp_get_bounds_for_decl_addr): Likewise.
14235 * tree-chrec.c (chrec_apply): Likewise.
14236 * tree-data-ref.c (dump_data_reference): Likewise.
14237 (dump_subscript): Likewise.
14238 (dump_data_dependence_relation): Likewise.
14239 (analyze_overlapping_iterations): Likewise.
14240 * tree-inline.c (expand_call_inline): Likewise.
14241 (tree_function_versioning): Likewise.
14242 * tree-into-ssa.c (dump_defs_stack): Likewise.
14243 (dump_currdefs): Likewise.
14244 (dump_names_replaced_by): Likewise.
14245 (dump_update_ssa): Likewise.
14246 (update_ssa): Likewise.
14247 * tree-object-size.c (pass_object_sizes::execute): Likewise.
14248 * tree-parloops.c (build_new_reduction): Likewise.
14249 (try_create_reduction_list): Likewise.
14250 (ref_conflicts_with_region): Likewise.
14251 (oacc_entry_exit_ok_1): Likewise.
14252 (oacc_entry_exit_single_gang): Likewise.
14253 * tree-pretty-print.h: Likewise.
14254 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
14255 (get_scalar_evolution): Likewise.
14256 (add_to_evolution): Likewise.
14257 (get_loop_exit_condition): Likewise.
14258 (analyze_evolution_in_loop): Likewise.
14259 (analyze_initial_condition): Likewise.
14260 (analyze_scalar_evolution): Likewise.
14261 (instantiate_scev): Likewise.
14262 (number_of_latch_executions): Likewise.
14263 (gather_chrec_stats): Likewise.
14264 (final_value_replacement_loop): Likewise.
14265 (scev_const_prop): Likewise.
14266 * tree-sra.c (dump_access): Likewise.
14267 (disqualify_candidate): Likewise.
14268 (create_access): Likewise.
14269 (reject): Likewise.
14270 (maybe_add_sra_candidate): Likewise.
14271 (create_access_replacement): Likewise.
14272 (analyze_access_subtree): Likewise.
14273 (analyze_all_variable_accesses): Likewise.
14274 (sra_modify_assign): Likewise.
14275 (initialize_constant_pool_replacements): Likewise.
14276 (find_param_candidates): Likewise.
14277 (decide_one_param_reduction): Likewise.
14278 (replace_removed_params_ssa_names): Likewise.
14279 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
14280 * tree-ssa-copy.c (dump_copy_of): Likewise.
14281 (copy_prop_visit_cond_stmt): Likewise.
14282 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
14283 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
14284 (record_equivalences_from_stmt): Likewise.
14285 * tree-ssa-dse.c (compute_trims): Likewise.
14286 (delete_dead_call): Likewise.
14287 (delete_dead_assignment): Likewise.
14288 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
14289 (forward_propagate_into_cond): Likewise.
14290 (pass_forwprop::execute): Likewise.
14291 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
14292 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
14294 (move_computations_worker): Likewise.
14295 (execute_sm): Likewise.
14296 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
14297 (remove_exits_and_undefined_stmts): Likewise.
14298 (remove_redundant_iv_tests): Likewise.
14299 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
14300 (adjust_iv_update_pos): Likewise.
14301 * tree-ssa-math-opts.c (bswap_replace): Likewise.
14302 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
14303 (value_replacement): Likewise.
14304 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
14305 * tree-ssa-pre.c (print_pre_expr): Likewise.
14306 (get_representative_for): Likewise.
14307 (create_expression_by_pieces): Likewise.
14308 (insert_into_preds_of_block): Likewise.
14309 (eliminate_insert): Likewise.
14310 (eliminate_dom_walker::before_dom_children): Likewise.
14311 (eliminate): Likewise.
14312 (remove_dead_inserted_code): Likewise.
14313 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
14314 * tree-ssa-reassoc.c (get_rank): Likewise.
14315 (eliminate_duplicate_pair): Likewise.
14316 (eliminate_plus_minus_pair): Likewise.
14317 (eliminate_not_pairs): Likewise.
14318 (undistribute_ops_list): Likewise.
14319 (eliminate_redundant_comparison): Likewise.
14320 (update_range_test): Likewise.
14321 (optimize_range_tests_var_bound): Likewise.
14322 (optimize_vec_cond_expr): Likewise.
14323 (rewrite_expr_tree): Likewise.
14324 (rewrite_expr_tree_parallel): Likewise.
14325 (linearize_expr): Likewise.
14326 (break_up_subtract): Likewise.
14327 (linearize_expr_tree): Likewise.
14328 (attempt_builtin_powi): Likewise.
14329 (attempt_builtin_copysign): Likewise.
14330 (transform_stmt_to_copy): Likewise.
14331 (transform_stmt_to_multiply): Likewise.
14332 (dump_ops_vector): Likewise.
14333 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
14334 (print_scc): Likewise.
14335 (set_ssa_val_to): Likewise.
14336 (visit_reference_op_store): Likewise.
14337 (visit_use): Likewise.
14338 (sccvn_dom_walker::before_dom_children): Likewise.
14339 (run_scc_vn): Likewise.
14340 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
14342 (expr_hash_elt::print): Likewise.
14343 (const_and_copies::pop_to_marker): Likewise.
14344 (const_and_copies::record_const_or_copy_raw): Likewise.
14345 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
14346 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
14347 (dump_predicates): Likewise.
14348 (find_uninit_use): Likewise.
14349 (warn_uninitialized_phi): Likewise.
14350 (pass_late_warn_uninitialized::execute): Likewise.
14351 * tree-ssa.c (verify_vssa): Likewise.
14352 (verify_ssa): Likewise.
14353 (maybe_optimize_var): Likewise.
14354 * tree-vrp.c (dump_value_range): Likewise.
14355 (dump_all_value_ranges): Likewise.
14356 (dump_asserts_for): Likewise.
14357 (register_edge_assert_for_2): Likewise.
14358 (vrp_visit_cond_stmt): Likewise.
14359 (vrp_visit_switch_stmt): Likewise.
14360 (vrp_visit_stmt): Likewise.
14361 (vrp_visit_phi_node): Likewise.
14362 (simplify_cond_using_ranges_1): Likewise.
14363 (fold_predicate_in): Likewise.
14364 (evrp_dom_walker::before_dom_children): Likewise.
14365 (evrp_dom_walker::push_value_range): Likewise.
14366 (evrp_dom_walker::pop_value_range): Likewise.
14367 (execute_early_vrp): Likewise.
14369 2017-05-16 Richard Biener <rguenther@suse.de>
14371 * dwarf2out.c (loc_list_from_tree_1): Do not create
14372 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
14374 2017-05-16 Richard Biener <rguenther@suse.de>
14376 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
14378 (note_variable_value_in_expr): If we resolved the decl ref
14379 do not push to the stack.
14381 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
14383 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
14384 operations in fast-math mode.
14385 (vaddq_f16): Likewise.
14386 (vmul_f16): Likewise.
14387 (vmulq_f16): Likewise.
14388 (vsub_f16): Likewise.
14389 (vsubq_f16): Likewise.
14390 * config/arm/neon.md (add<mode>3): New.
14392 (fma:<VH:mode>3): New. Also remove outdated comment.
14395 2017-05-16 Martin Liska <mliska@suse.cz>
14399 * ipa-devirt.c (warn_types_mismatch): Fix typo.
14400 (odr_types_equivalent_p): Likewise.
14402 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
14404 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
14406 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
14409 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
14410 non-interunit SSE move alternatives with '?'.
14411 (zero-extendsidi peephole2): New peephole to skip intermediate
14412 general register in SSE zero-extend sequence.
14414 2017-05-15 Jeff Law <law@redhat.com>
14416 * reorg.c (relax_delay_slots): Create a new variable to hold
14417 the temporary target rather than clobbering TARGET_LABEL.
14419 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
14420 missing argument to extract_bit_field call.
14421 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
14423 2017-05-15 Martin Liska <mliska@suse.cz>
14426 * gcc.c (process_command): Do not allow empty argument of -o option.
14428 2017-05-15 Renlin Li <renlin.li@arm.com>
14430 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
14431 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
14432 * config/aarch64/constraints.md (Usf): Add long call check.
14433 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
14434 (call_value): Likewise.
14435 (sibcall): Likewise.
14436 (sibcall_value): Likewise.
14438 (call_value_insn): New.
14439 (sibcall_insn): Update rtx pattern.
14440 (sibcall_value_insn): Likewise.
14441 (call_internal): Remove.
14442 (call_value_internal): Likewise.
14443 (sibcall_internal): Likewise.
14444 (sibcall_value_internal): Likewise.
14445 (call_reg): Likewise.
14446 (call_symbol): Likewise.
14447 (call_value_reg): Likewise.
14448 (call_value_symbol): Likewise.
14450 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
14453 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
14455 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
14457 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
14458 compatible with CCGOCmode and with CCZmode.
14460 2017-05-14 Martin Sebor <msebor@redhat.com>
14462 PR middle-end/77671
14463 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
14464 (gimple_fold_builtin_snprintf): Same.
14465 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
14466 (gimple_fold_builtin_snprintf): Same.
14467 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
14468 of character types.
14469 (is_call_safe): New function.
14470 (try_substitute_return_value): Call it.
14471 (try_simplify_call): New function.
14472 (pass_sprintf_length::handle_gimple_call): Call it.
14474 2017-05-14 Martin Sebor <msebor@redhat.com>
14476 PR middle-end/80669
14477 * builtins.c (expand_builtin_stpncpy): Simplify.
14479 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
14481 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
14482 * config/i386/i386.h
14483 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
14484 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
14485 (struct machine_function): Add new members call_ms2sysv,
14486 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
14487 (struct machine_frame_state): New fields sp_realigned and
14488 sp_realigned_offset.
14489 * config/i386/i386.c
14490 (enum xlogue_stub): New enum.
14491 (enum xlogue_stub_sets): New enum.
14492 (class xlogue_layout): New class.
14493 (struct ix86_frame): New fields stack_realign_allocate_offset,
14494 stack_realign_offset and outlined_save_offset. Modify comments to
14495 detail stack layout when using out-of-line stubs.
14496 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
14497 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
14498 -mcall-ms2sysv-xlogues.
14499 (stub_managed_regs): New static variable.
14500 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
14501 registers managed by out-of-line stub.
14502 (disable_call_ms2sysv_xlogues): New function.
14503 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
14504 m->call_ms2sysv when appropriate and compute frame layout for
14506 (sp_valid_at, fp_valid_at): New inline functions.
14507 (choose_basereg): New function.
14508 (choose_baseaddr): Add align parameter, use choose_basereg and modify
14510 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
14511 Use align parameter of choose_baseaddr to generated aligned SSE movs
14513 (pro_epilogue_adjust_stack): Modify to track
14514 machine_frame_state::sp_realigned.
14515 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
14516 (ix86_nsaved_sseregs): Likewise.
14517 (ix86_emit_save_regs): Likewise.
14518 (ix86_emit_save_regs_using_mov): Likewise.
14519 (ix86_emit_save_sse_regs_using_mov): Likewise.
14520 (get_scratch_register_on_entry): Likewise.
14521 (gen_frame_set): New function.
14522 (gen_frame_load): Likewise.
14523 (gen_frame_store): Likewise.
14524 (emit_outlined_ms2sysv_save): Likewise.
14525 (emit_outlined_ms2sysv_restore): Likewise.
14526 (ix86_expand_prologue): Modify stack re-alignment code and call
14527 emit_outlined_ms2sysv_save when appropriate.
14528 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
14529 parameter rtx_insn *insn, which allows the function to be used to only
14530 generate the notes.
14531 (ix86_expand_epilogue): Modify validity checks of frame and stack
14532 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
14533 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
14534 * config/i386/predicates.md
14535 (save_multiple): New predicate.
14536 (restore_multiple): Likewise.
14537 * config/i386/sse.md
14538 (save_multiple<mode>): New pattern.
14539 (save_multiple_realign<mode>): Likewise.
14540 (restore_multiple<mode>): Likewise.
14541 (restore_multiple_and_return<mode>): Likewise.
14542 (restore_multiple_leave_return<mode>): Likewise.
14543 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
14545 2017-05-14 Julia Koval <julia.koval@intel.com>
14547 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
14548 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
14549 (__builtin_ia32_xsetbv): New builtins.
14550 * config/i386/i386.c (ix86_expand_special_args_builtin):
14552 (ix86_expand_builtin): Special expand for new intrinsics.
14553 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
14554 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
14555 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
14557 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14559 * cfganal.c (inverted_post_order_compute): Change argument type
14561 * cfganal.h (inverted_post_order_compute): Adjust prototype.
14562 * df-core.c (rest_of_handle_df_initialize): Adjust.
14563 (rest_of_handle_df_finish): Likewise.
14564 (df_analyze_1): Likewise.
14565 (df_analyze): Likewise.
14566 (loop_inverted_post_order_compute): Change argument to be a vec *.
14567 (df_analyze_loop): Adjust.
14568 (df_get_n_blocks): Likewise.
14569 (df_get_postorder): Likewise.
14570 * df.h (struct df_d): Change field to be a vec.
14571 * lcm.c (compute_laterin): Adjust.
14572 (compute_available): Likewise.
14573 * lra-lives.c (lra_create_live_ranges_1): Likewise.
14574 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
14575 * tree-ssa-pre.c (compute_antic): Likewise.
14577 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14579 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
14580 (depth_first_search::depth_first_search): Change structure init
14581 function to this constructor.
14582 (depth_first_search::add_bb): Rename function to this member.
14583 (depth_first_search::execute): Likewise.
14584 (flow_dfs_compute_reverse_finish): Adjust.
14586 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14588 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
14589 (longest_simple_path): Likewise.
14590 * shrink-wrap.c (spread_components): Likewise.
14591 (disqualify_problematic_components): Likewise.
14592 (emit_common_heads_for_components): Likewise.
14593 (emit_common_tails_for_components): Likewise.
14594 (insert_prologue_epilogue_for_components): Likewise.
14596 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14598 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
14601 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14603 * df-core.c (df_set_blocks): Start using auto_bitmap.
14604 (df_compact_blocks): Likewise.
14605 * df-problems.c (df_rd_confluence_n): Likewise.
14606 * df-scan.c (df_insn_rescan_all): Likewise.
14607 (df_process_deferred_rescans): Likewise.
14608 (df_update_entry_block_defs): Likewise.
14609 (df_update_exit_block_uses): Likewise.
14610 (df_entry_block_bitmap_verify): Likewise.
14611 (df_exit_block_bitmap_verify): Likewise.
14612 (df_scan_verify): Likewise.
14613 * lra-constraints.c (lra_constraints): Likewise.
14614 (undo_optional_reloads): Likewise.
14615 (lra_undo_inheritance): Likewise.
14616 * lra-remat.c (calculate_gen_cands): Likewise.
14617 (do_remat): Likewise.
14618 * lra-spills.c (assign_spill_hard_regs): Likewise.
14619 (spill_pseudos): Likewise.
14620 * tree-ssa-pre.c (bitmap_set_and): Likewise.
14621 (bitmap_set_subtract_values): Likewise.
14623 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14625 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
14626 management with auto_bitmap.
14627 (fix_inter_tick): Likewise.
14628 (fix_recovery_deps): Likewise.
14629 * ira.c (add_store_equivs): Likewise.
14630 (find_moveable_pseudos): Likewise.
14631 (split_live_ranges_for_shrink_wrap): Likewise.
14632 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
14633 (rtx_reuse_manager::seen_def_p): Likewise.
14634 (rtx_reuse_manager::set_seen_def): Likewise.
14635 * print-rtl.h (class rtx_reuse_manager): Likewise.
14637 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14639 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
14641 (migrate_btr_def): Likewise.
14642 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
14643 * df-core.c (loop_post_order_compute): Likewise.
14644 (loop_inverted_post_order_compute): Likewise.
14645 * hsa-common.h: Likewise.
14646 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
14647 * init-regs.c (initialize_uninitialized_regs): Likewise.
14648 * ipa-inline.c (resolve_noninline_speculation): Likewise.
14649 (inline_small_functions): Likewise.
14650 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
14651 * ira.c (combine_and_move_insns): Likewise.
14652 (build_insn_chain): Likewise.
14653 * loop-invariant.c (find_invariants): Likewise.
14654 * lower-subreg.c (propagate_pseudo_copies): Likewise.
14655 * predict.c (tree_predict_by_opcode): Likewise.
14656 (predict_paths_leading_to): Likewise.
14657 (predict_paths_leading_to_edge): Likewise.
14658 (estimate_loops_at_level): Likewise.
14659 (estimate_loops): Likewise.
14660 * shrink-wrap.c (try_shrink_wrapping): Likewise.
14661 (spread_components): Likewise.
14662 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
14663 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
14664 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
14665 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
14666 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
14667 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
14668 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
14669 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
14670 (mark_threaded_blocks): Likewise.
14671 (thread_through_all_blocks): Likewise.
14672 * tree-ssa.c (verify_ssa): Likewise.
14673 (execute_update_addresses_taken): Likewise.
14674 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
14676 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14678 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
14680 (post_order_compute): Likewise.
14681 (inverted_post_order_compute): Likewise.
14682 (pre_and_rev_post_order_compute_fn): Likewise.
14684 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14686 * genrecog.c (int_set::int_set): Explicitly construct our
14687 auto_vec base class.
14688 * vec.h (auto_vec::auto_vec): New constructor.
14690 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14692 * bitmap.h (class auto_bitmap): New constructor taking
14693 bitmap_obstack * argument.
14695 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14697 * bitmap.h (class auto_bitmap): Change type of m_bits to
14698 bitmap_head, and adjust ctor / dtor and member operators.
14700 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
14702 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
14703 when returned register mode doesn't match original mode.
14705 2017-05-12 Jeff Law <law@redhat.com>
14706 Jakub Jelinek <jakub@redhat.com>
14708 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
14709 we look for cc setter after the compare-elim changes.
14710 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
14711 within the vector to match what compare-elim now expects.
14712 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
14713 (xorsi3_flags, one_cmplsi2_flags): Likewise.
14715 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
14716 after the compare-elim changes.
14717 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
14718 the vector to match what compare-elim now expects.
14719 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
14720 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
14721 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
14722 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
14723 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
14725 * config/visium/visium.c (single_set_and_flags): Fix where
14726 we look for cc setter after the compare-elim changes.
14727 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
14728 with the vector to match what compare-elim now expects.
14729 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
14730 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
14731 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
14732 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
14733 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
14734 (neg<mode>2_insn_set_overflow): Likewise.
14736 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
14738 PR middle-end/79794
14739 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
14740 maybe_expand_insn call, set ops[0].target. If still set after call,
14741 set alt_rtl. Add extra arg to recursive calls.
14742 (extract_bit_field): Add alt_rtl argument. Pass to
14744 * expmed.h (extract_bit_field): Fix prototype.
14745 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
14746 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
14747 to extract_bit_field_calls.
14748 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
14749 Pass alt_rtl to extract_bit_field calls.
14750 * calls.c (store_unaligned_arguments_into_psuedos)
14751 load_register_parameters): Pass extra NULL to extract_bit_field calls.
14752 * optabs.c (maybe_legitimize_operand): Clear op->target when call
14754 * optabs.h (struct expand_operand): Add target bitfield.
14756 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
14758 * compare-elim.c (try_eliminate_compare): Canonicalize
14759 operation with embedded compare to
14760 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
14761 (set (reg) (operation)].
14763 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
14765 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
14768 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
14769 cost of adding a carry flag for ADC instruction.
14770 [case MINUS]: Ignore the cost of subtracting a carry flag
14771 for SBB instruction.
14773 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
14775 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
14777 * config/rs6000/bmiintrin.h: New file.
14778 * config/rs6000/bmi2intrin.h: New file.
14779 * config/rs6000/x86intrin.h: New file.
14781 2017-05-12 Jeff Law <law@redhat.com>
14783 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
14786 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
14788 PR middle-end/80707
14789 * tree-cfg.c: Remove cfg edges of unreachable case statements.
14791 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14793 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
14794 early expansion of vector divide builtins.
14795 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
14796 builtins identified as having unsigned arguments.
14798 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14800 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
14801 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
14802 expansion of vector logical operations (and, andc, or, xor,
14805 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14807 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
14808 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
14810 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
14812 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
14813 early GIMPLE expansion of vector multiplies.
14815 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14817 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
14818 TARGET_HAVE_MOVT conditional.
14819 (movt splitter): Likewise.
14821 2017-05-12 Richard Biener <rguenther@suse.de>
14823 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
14824 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
14825 Fold all stmts not inplace.
14827 2017-05-12 Richard Biener <rguenther@suse.de>
14829 PR tree-optimization/80713
14830 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
14831 inserted_exprs bit for not removed stmts.
14833 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
14835 PR middle-end/69921
14836 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
14837 parallelized" attribute for parallelized OpenACC kernels.
14838 * omp-offload.c (execute_oacc_device_lower): Use it.
14840 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
14841 Set "oacc kernels" attribute.
14842 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
14843 parameter. Adjust all users.
14844 (oacc_fn_attrib_kernels_p): Remove function.
14845 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
14846 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
14847 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
14848 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
14849 assert "oacc kernels" attribute is set.
14851 2017-05-11 Carl Love <cel@us.ibm.com>
14853 * config/rs6000/rs6000-c: Add support for built-in functions
14854 vector unsigned char vec_popcnt (vector signed char)
14855 vector unsigned char vec_popcnt (vector unsigned char)
14856 vector unsigned short vec_popcnt (vector signed short)
14857 vector unsigned short vec_popcnt (vector unsigned short)
14858 vector unsigned int vec_popcnt (vector signed int)
14859 vector unsigned int vec_popcnt (vector unsigned int)
14860 vector unsigned long long vec_popcnt (vector signed long long)
14861 vector unsigned long long vec_popcnt (vector unsigned long long)
14862 vector signed long long vec_slo (vector signed long long,
14863 vector signed char)
14864 vector signed long long vec_slo (vector signed long long,
14865 vector unsigned char)
14866 vector unsigned long long vec_slo (vector unsigned long long,
14867 vector signed char)
14868 vector unsigned long long vec_slo (vector unsigned long long,
14869 vector unsigned char)
14870 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
14871 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
14872 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
14873 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
14874 * doc/extend.texi: Update the built-in documentation file for the
14875 new built-in functions.
14877 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
14879 * attribs.h (sorted_attr_string): Move machine independent
14880 functions for target clone support from the i386 port to common
14881 code. Rename ix86_function_versions to common_function_versions.
14882 Rename make_name to make_unique_name.
14883 (common_function_versions): Likewise.
14884 (make_unique_name): Likewise.
14885 (make_dispatcher_decl): Likewise.
14886 (is_function_default_version): Likewise.
14887 * attribs.c (attr_strcmp): Likewise.
14888 (sorted_attr_string): Likewise.
14889 (common_function_versions): Likewise.
14890 (make_unique_name): Likewise.
14891 (make_dispatcher_decl): Likewise.
14892 (is_function_default_version): Likewise.
14893 * config/i386/i386.c (attr_strcmp): Likewise.
14894 (sorted_attr_string): Likewise.
14895 (ix86_function_versions): Likewise.
14896 (make_name): Likewise.
14897 (make_dispatcher_decl): Likewise.
14898 (is_function_default_version): Likewise.
14899 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
14901 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14904 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
14905 Account for direct move costs for vec_construct of integer
14908 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
14911 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
14912 (UNSPEC_STX_ATOMIC): Ditto.
14913 (loaddi_via_sse): New insn.
14914 (storedi_via_sse): Ditto.
14915 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
14916 Update corresponding peephole2 patterns.
14917 (atomic_storedi_fpu): Ditto.
14919 2017-05-11 Julia Koval <julia.koval@intel.com>
14921 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
14922 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
14924 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
14925 (__builtin_ia32_rsqrt14ss_mask): New builtins.
14926 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
14928 2017-05-11 Nathan Sidwell <nathan@acm.org>
14930 * graphite-poly.c: Include dumpfile.h.
14932 * dumpfle.h (dump_function): Declare here ...
14933 * tree-dump.h (dump_function): ... not here.
14934 * dumpfile.c: #include tree-cfg.h.
14935 (dump_function): Move here from ...
14936 * tree-dump.c (dump_function): ... here.
14937 * gimplify.c: #include splay-tree.h, not tree-dump.h.
14938 * graphite-poly.c: Don't include tree-dump.h.
14939 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
14940 * print-tree.c: Likewise.
14941 * stor-layout.c: Likewise.
14942 * tree-nested.c: Likewise.
14944 * dumpfile.c (dump_start): Use TDF_FLAGS.
14945 (dump_enable_all): Fix TDF_KIND check thinko.
14947 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
14949 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14950 array entries to represent two legal parameterizations of the
14951 overloaded __builtin_cmpb function, as represented by the
14952 P6_OV_BUILTIN_CMPB constant.
14953 (altivec_resolve_overloaded_builtin): Add special case handling
14954 for the __builtin_cmpb function, as represented by the
14955 P6_OV_BUILTIN_CMPB constant.
14956 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
14957 (BU_P6_64BIT_2): New macro.
14958 (BU_P6_OVERLOAD_2): New macro
14959 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
14960 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
14961 (CMPB): Add overload support to represent both 32-bit and 64-bit
14962 compare-bytes function.
14963 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
14964 support for TARGET_CMPB.
14965 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
14966 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
14967 documentation of the __builtin_cmpb overloaded built-in function.
14969 2017-05-11 Richard Biener <rguenther@suse.de>
14971 PR tree-optimization/80705
14972 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
14973 bases are not vectorizable.
14975 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14977 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
14978 when counting register pressure.
14980 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14982 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
14983 (iv_ca_more_deps): Renamed to ...
14984 (iv_ca_compare_deps): ... this.
14985 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
14987 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14989 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
14991 (determine_group_iv_costs): ... here.
14992 (find_inv_vars_cb): Record inv var if it's not recorded before.
14994 2017-05-11 Bin Cheng <bin.cheng@arm.com>
14996 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
14997 (get_shiftadd_cost): Ditto.
14999 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15001 * tree-ssa-address.c: Include header file.
15002 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
15004 (add_to_parts): Refactor.
15005 (addr_to_parts): New parameter. Update use of move_hint_to_base.
15006 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
15009 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15011 PR tree-optimization/53090
15012 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
15014 (extract_cond_operands): Detect condition with IV on both sides
15015 and return COMP_IV_EXPR_2.
15016 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
15017 (rewrite_use_compare): Simplify by removing call to function
15018 extract_cond_operands.
15020 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15022 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
15023 (extract_cond_operands): Detect condition comparing against non-
15024 invariant bound and return appropriate enum value.
15025 (find_interesting_uses_cond): Update use of extract_cond_operands.
15026 Handle its return value accordingly.
15027 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
15029 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15031 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
15032 nonlinear iv_use computation in loop invariant sensitive way.
15034 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15036 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
15037 (find_iv_candidates): Call relate_compare_use_with_all_cands.
15039 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15041 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
15042 (dump_cand): Support iv_cand.inv_exprs.
15043 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
15045 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
15048 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15050 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
15052 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
15053 as local function. Include necessary header files.
15054 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
15056 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15058 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
15060 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15062 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
15063 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
15064 RSHIFT_EXPR and BIT_NOT_EXPR.
15066 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15068 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
15069 (adjust_setup_cost): New parameter supporting round up adjustment.
15070 (struct address_cost_data): Delete.
15071 (force_expr_to_var_cost): Don't bound cost with spill_cost.
15072 (split_address_cost, ptr_difference_cost): Delete.
15073 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
15074 (struct ainc_cost_data): New struct.
15075 (get_address_cost_ainc): New function.
15076 (get_address_cost, get_computation_cost): Reimplement.
15077 (determine_group_iv_cost_address): Record inv_expr for all uses of
15079 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
15080 (iv_ca_has_deps): Reimplemented to ...
15081 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
15083 (iv_ca_extend): Call iv_ca_more_deps.
15085 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15087 * tree-ssa-address.c (struct mem_address): Move to header file.
15088 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
15089 * tree-ssa-address.h (struct mem_address): Move from C file.
15090 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
15092 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15094 * tree-affine.h (aff_combination_type): New interface.
15095 (aff_combination_zero_p): Remove static.
15096 (aff_combination_const_p): New interface.
15097 (aff_combination_singleton_var_p): New interfaces.
15099 2017-05-11 Richard Biener <rguenther@suse.de>
15101 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
15102 Skip unreachable blocks and destinations.
15103 (eliminate): Move stmt removal and fixup ...
15104 (fini_eliminate): ... here. Skip inserted exprs.
15105 (pass_pre::execute): Move fini_pre after fini_eliminate.
15106 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
15107 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
15108 PRE to get rid of dead code that has invalid SSA form and
15109 split critical edges again.
15111 2017-05-11 Bin Cheng <bin.cheng@arm.com>
15113 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
15115 2017-05-11 Richard Biener <rguenther@suse.de>
15117 * passes.c (execute_function_todo): Verify loops if they are
15118 said to be up-to-date.
15119 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
15120 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
15122 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
15125 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
15126 handle calling assemble_external ourself.
15129 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
15130 modes with zero size. Enhance comment.
15132 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15134 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
15135 built-ins for vec_xl and vec_xst with short and char pointer
15138 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15140 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
15141 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
15142 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
15143 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
15144 (_mm_maskz_min_round_ss): New intrinsics.
15145 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
15146 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
15147 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
15148 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
15149 (__builtin_ia32_minss_mask_round): New builtins.
15150 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15151 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15152 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
15154 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
15155 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
15157 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
15160 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
15162 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
15163 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
15164 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
15165 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
15166 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
15167 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
15168 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
15169 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15170 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
15171 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
15172 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
15173 (__builtin_ia32_mulss_mask_round): New builtins.
15174 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15175 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15176 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
15178 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
15179 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
15181 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
15184 2017-05-10 Julia Koval <julia.koval@intel.com>
15186 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
15187 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
15188 (_mm256_setr_m128i): New intrinsics.
15190 2017-05-10 Julia Koval <julia.koval@intel.com>
15192 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
15193 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
15194 (_mm_maskz_rcp14_ss): New intrinsics.
15195 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
15196 (__builtin_ia32_rcp14ss_mask): New builtins.
15197 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
15199 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
15201 PR tree-optimization/51513
15202 * tree-cfg.c (gimple_seq_unreachable_p): New function.
15203 (assert_unreachable_fallthru_edge_p): Use it.
15204 (group_case_labels_stmt): Likewise.
15205 * tree-cfg.h: Prototype it.
15206 * stmt.c: Include cfghooks.h and tree-cfg.h.
15207 (emit_case_dispatch_table) <gap_label>: New local variable.
15208 Use it to fill dispatch table gaps.
15209 Test for default_label before updating probabilities.
15210 (expand_case) <default_label>: Remove unneeded initialization.
15211 Test for unreachable default case statement and remove its edge.
15212 Set default_label accordingly.
15213 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
15215 2017-05-10 Carl Love <cel@us.ibm.com>
15217 * config/rs6000/rs6000-c: Add support for built-in functions
15218 vector signed char vec_neg (vector signed char)
15219 vector signed short int vec_neg (vector short int)
15220 vector signed int vec_neg (vector signed int)
15221 vector signed long long vec_neg (vector signed long long)
15222 vector float vec_neg (vector float)
15223 vector double vec_neg (vector double)
15224 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
15226 * config/rs6000/altivec.h: Add define for vec_neg
15227 * doc/extend.texi: Update the built-in documentation for the
15228 new built-in functions.
15230 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15232 PR tree-optimization/77644
15233 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
15235 2017-05-10 Nathan Sidwell <nathan@acm.org>
15237 * dumpfile.h (TDI_lang_all): New.
15238 (TDF_KIND): New. Renumber others
15239 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
15241 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
15243 (get_dump_file_name): Adjust suffix generation.
15244 (dump_enable_all): Use TDF_KIND.
15245 * doc/invoke.texi (-fdump-lang-all): Document.
15247 * dumpfile.h: Tabify.
15249 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
15252 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
15253 Move member access before delete.
15255 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
15257 * tree-inline.c (expand_call_inline): Split block at stmt
15260 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
15263 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
15264 are now unused after splitting mov{sf,sd}_hardfloat.
15265 (f32_lr2): Likewise.
15266 (f32_lm): Likewise.
15267 (f32_lm2): Likewise.
15268 (f32_li): Likewise.
15269 (f32_li2): Likewise.
15270 (f32_lv): Likewise.
15271 (f32_sr): Likewise.
15272 (f32_sr2): Likewise.
15273 (f32_sm): Likewise.
15274 (f32_sm2): Likewise.
15275 (f32_si): Likewise.
15276 (f32_si2): Likewise.
15277 (f32_sv): Likewise.
15278 (f32_dm): Likewise.
15279 (f32_vsx): Likewise.
15280 (f32_av): Likewise.
15281 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
15282 For movsf, order stores so the VSX stores occur before the GPR
15283 store which encourages the register allocator to use a traditional
15284 FPR instead of a GPR. For movsd, order the stores so that the GPR
15285 store comes before the VSX stores to allow the power6 to work.
15286 This is due to the power6 not having a 32-bit integer store
15287 instruction from a FPR.
15288 (movsf_hardfloat): Likewise.
15289 (movsd_hardfloat): Likewise.
15291 2017-05-09 Martin Sebor <msebor@redhat.com>
15293 PR translation/80280
15294 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
15297 PR translation/80280
15298 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
15299 data member added in r247778.
15300 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
15302 2017-05-09 Nathan Sidwell <nathan@acm.org>
15304 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
15306 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
15309 2017-05-09 Marek Polacek <polacek@redhat.com>
15311 * doc/invoke.texi: Fix typo.
15313 2017-05-09 Richard Biener <rguenther@suse.de>
15315 * tree-vrp.c (vrp_val_is_max): Adjust comment.
15316 (vrp_val_is_min): Likewise.
15317 (set_value_range_to_value): Likewise.
15318 (set_value_range_to_nonnegative): Likewise.
15319 (gimple_assign_nonzero_p): Likewise.
15320 (gimple_stmt_nonzero_p): Likewise.
15321 (vrp_int_const_binop): Likewise. Remove unreachable case.
15322 (adjust_range_with_scev): Adjust comments.
15323 (compare_range_with_value): Likewise.
15324 (extract_range_from_phi_node): Likewise.
15325 (test_for_singularity): Likewise.
15327 2017-05-09 Richard Biener <rguenther@suse.de>
15329 * tree-vrp.c (get_single_symbol): Add assert that we don't
15330 get overflowed constants as invariant part.
15331 (compare_values_warnv): Add comment before the TREE_NO_WARNING
15332 checks. Use wi::cmp instead of recursing for integer constants.
15333 (compare_values): Just ignore whether we assumed undefined
15334 overflow instead of failing the compare.
15335 (extract_range_for_var_from_comparison_expr): Add comment before the
15336 TREE_NO_WARNING sets.
15337 (test_for_singularity): Likewise.
15338 (extract_range_from_comparison): Do not disable optimization
15339 when we assumed undefined overflow.
15340 (extract_range_basic): Remove init of unused var.
15342 2017-05-09 Richard Biener <rguenther@suse.de>
15344 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
15345 (extract_range_from_multiplicative_op_1): Adjust.
15346 (extract_range_from_binary_expr_1): Use int_const_binop.
15348 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
15351 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
15352 rs6000_store_data_bypass_p in seven define_bypass directives and
15353 in several comments.
15354 * config/rs6000/rs6000-protos.h: Add prototype for
15355 rs6000_store_data_bypass_p function.
15356 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
15357 function implements slightly different (rs6000-specific) semantics
15358 than store_data_bypass_p, returning false rather than aborting
15359 with assertion error when arguments do not satisfy the
15360 requirements of store data bypass.
15361 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
15362 rs6000_store_data_bypass_p.
15364 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
15366 * config/xtensa/xtensa-protos.h
15367 (xtensa_initial_elimination_offset): New declaration.
15368 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
15369 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
15370 macro definition, add case for FRAME_POINTER_REGNUM when
15371 FRAME_GROWS_DOWNWARD.
15372 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
15373 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
15374 xtensa_initial_elimination_offset.
15376 2017-05-08 Nathan Sidwell <nathan@acm.org>
15378 * doc/invoke.texi: Alphabetize -fdump options.
15380 2017-05-08 Martin Sebor <msebor@redhat.com>
15382 PR translation/80280
15383 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
15385 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
15387 * target.def (compute_frame_layout): New optional target hook.
15388 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
15389 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
15390 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
15392 * reload1.c (verify_initial_elim_offsets): Likewise.
15393 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
15394 (use_simple_return_p): Call arm_compute_frame_layout if needed.
15395 (arm_get_frame_offsets): Split up into this ...
15396 (arm_compute_frame_layout): ... and this function.
15398 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
15400 * config/aarch64/constraints.md (Usa): New constraint.
15401 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
15403 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
15405 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
15406 with_multilib_list after it has been checked.
15408 2017-05-08 Richard Biener <rguenther@suse.de>
15410 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
15411 (bitmap_set_subtract_values): Likewise.
15413 2017-05-08 Richard Biener <rguenther@suse.de>
15415 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
15416 (gimple_assign_nonzero): ... this and remove strict_overflow_p
15418 (gimple_stmt_nonzero_warnv_p): Rename to ...
15419 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
15421 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
15422 (extract_range_basic): Adjust, do not disable propagation on
15423 strict overflow sensitive simplification.
15424 (vrp_visit_cond_stmt): Likewise.
15426 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
15428 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
15429 body size unconditionally.
15431 2017-05-07 Jeff Law <law@redhat.com>
15434 2017-05-06 Jeff Law <law@redhat.com>
15435 PR tree-optimization/78496
15436 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15439 PR tree-optimization/78496
15440 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15441 (simplify_stmt_using_ranges): Call it.
15442 (vrp_dom_walker::before_dom_children): Extract equivalences
15443 from an ASSERT_EXPR with an equality comparison against a
15446 2017-05-06 Jeff Law <law@redhat.com>
15448 PR tree-optimization/78496
15449 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
15452 PR tree-optimization/78496
15453 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
15454 (simplify_stmt_using_ranges): Call it.
15455 (vrp_dom_walker::before_dom_children): Extract equivalences
15456 from an ASSERT_EXPR with an equality comparison against a
15459 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15461 * lra-constraints.c (lra_copy_reg_equiv): New function.
15462 (split_reg): Use it to copy equivalence information from the
15463 original register to the spill register.
15465 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
15467 PR rtl-optimization/75964
15468 * simplify-rtx.c (simplify_const_relational_operation): Remove
15469 invalid handling of comparisons of integer ABS.
15471 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
15473 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
15474 initialize to zero.
15475 (init_regs): Remove declaration.
15476 (function_arg_advance_32): Initialize error_p as boolean variable.
15478 2017-05-05 Nathan Sidwell <nathan@acm.org>
15480 * store-motion.c (remove_reachable_equiv_notes): Reformat long
15481 lines. Use for (;;).
15483 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15485 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
15486 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
15487 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
15488 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
15489 VF=2 that require versioning.
15491 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15493 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
15496 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15498 * diagnostic.h (diagnostic_override_option_index): Convert from
15499 macro to inline function.
15501 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15503 * diagnostic.c (last_module_changed_p): New function.
15504 (set_last_module): New function.
15505 (diagnostic_report_current_module): Convert macro usage to
15506 the above functions.
15507 * diagnostic.h (diagnostic_context::last_module): Strengthen
15508 from const line_map * to const line_map_ordinary *.
15509 (diagnostic_last_module_changed): Delete macro.
15510 (diagnostic_set_last_module): Delete macro.
15512 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15514 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
15515 with diagnostic_report_diagnostic.
15516 (diagnostic_n_impl_richloc): Likewise.
15517 * diagnostic.h (report_diagnostic): Delete macro.
15518 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
15519 with diagnostic_report_diagnostic.
15520 * substring-locations.c (format_warning_va): Likewise.
15522 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15524 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
15525 save/restor of format_spec. Move option-printing code to...
15526 (print_option_information): ...this new function, and
15527 reimplement by simply printing to the pretty_printer,
15528 rather than appending to the format string.
15530 2017-05-05 David Malcolm <dmalcolm@redhat.com>
15532 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
15533 handling logic into...
15534 (update_effective_level_from_pragmas): ...this new function.
15536 2017-05-04 Andrew Waterman <andrew@sifive.com>
15538 * config/riscv/riscv.opt (mstrict-align): New option.
15539 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
15540 (SLOW_UNALIGNED_ACCESS): Define.
15541 (riscv_slow_unaligned_access): Declare.
15542 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
15544 (riscv_slow_unaligned_access): New variable.
15545 (rocket_tune_info): Set slow_unaligned_access to true.
15546 (optimize_size_tune_info): Set slow_unaligned_access to false.
15547 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
15548 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
15549 (riscv_option_override): Set riscv_slow_unaligned_access.
15550 * doc/invoke.texi: Add -mstrict-align to RISC-V.
15552 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
15554 * config/riscv/riscv.md: Unify indentation.
15556 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
15561 * config/rs6000/rs6000.md (u code attribute): Add FIX and
15563 (extendsi<mode>2): Add support for doing sign extension via
15564 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
15565 don't have ISA 3.0 instructions.
15566 (extendsi<mode>2 splitter): Likewise.
15567 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
15568 generate the normal insns since SImode can now go in vector
15569 registers. Disallow the special UNSPECs needed for previous
15570 machines to hide SImode being used. Add new insns
15571 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
15572 (fix_trunc<mode>si2_stfiwx): Likewise.
15573 (fix_trunc<mode>si2_internal): Likewise.
15574 (fixuns_trunc<mode>si2): Likewise.
15575 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15576 (fctiw<u>z_<mode>_smallint): Likewise.
15577 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
15578 of floating point to 32-bit integer from doing a direct move to
15579 the GPR registers to do a store.
15580 (fctiwz_<mode>): Break long line.
15582 2017-05-05 Bin Cheng <bin.cheng@arm.com>
15584 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
15585 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
15586 (addr_list, addr_offset_valid_p): New.
15587 (split_address_groups): Check offset validity with above function.
15588 (gt-tree-ssa-loop-ivopts.h): Include header file.
15590 2017-05-05 Nathan Sidwell <nathan@acm.org>
15592 * config.gcc (arm*-*-*): Add missing 'fi'.
15594 2017-05-05 Steve Ellcey <sellcey@cavium.com>
15596 * doc/invoke.texi (-fopt-info): Explicitly say order of options
15597 included in -fopt-info does not matter.
15598 * doc/optinfo.texi (-fopt-info): Fix description of default
15599 behavour. Explicitly say order of options included in -fopt-info
15602 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
15604 * config.gcc: Allow combinations of aprofile and rmprofile values for
15605 --with-multilib-list.
15606 * config/arm/t-multilib: New file.
15607 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
15608 variables. Remove setting of ISA and floating-point ABI in
15609 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
15610 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
15611 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
15612 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
15614 * config/arm/t-rmprofile: Likewise except for the matches changes.
15615 * doc/install.texi (--with-multilib-list): Document the combination of
15616 aprofile and rmprofile values and warn about pitfalls in doing that.
15618 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
15620 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
15621 (movdi_aarch64): Likewise.
15623 2017-05-05 Jakub Jelinek <jakub@redhat.com>
15625 PR tree-optimization/80632
15626 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
15628 (build_arrays): Initialize it for virtual phis.
15629 (fix_phi_nodes): Use it for virtual phis.
15631 PR tree-optimization/80558
15632 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
15633 [x, y] op z into [x op, y op z] for op & or | if conditions
15636 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
15637 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15640 * config/arm/arm.md (use_literal_pool): Remove.
15641 (64-bit immediate split): No longer takes cost into consideration
15642 if arm_disable_literal_pool is enabled.
15643 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
15644 used when arm_disable_literal_pool is enabled.
15645 (arm_max_const_double_inline_cost): Remove use of
15646 arm_disable_literal_pool.
15647 (push_minipool_fix): Add assert.
15648 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
15649 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
15650 (no_literal_pool_sf_immediate): New.
15652 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15654 PR tree-optimization/80613
15655 * tree-ssa-dce.c (propagate_necessity): Remove cases for
15656 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
15658 2017-05-05 Richard Biener <rguenther@suse.de>
15660 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
15662 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
15664 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
15665 of this flag from insn conditions due to removal from r247495.
15667 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
15669 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
15671 (arm_early_store_addr_dep_ptr): Likewise.
15672 * config/arm/aarch-common-protos.h
15673 (arm_early_load_addr_dep_ptr): Add prototype.
15674 (arm_early_store_addr_dep_ptr): Likewise.
15675 * config/arm/cortex-a53.md: Add new bypasses.
15677 2017-05-05 Jakub Jelinek <jakub@redhat.com>
15679 * tree.c (next_type_uid): Change type to unsigned.
15680 (type_hash_canon): Decrement back next_type_uid if
15681 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
15682 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
15685 2017-05-04 Martin Sebor <msebor@redhat.com>
15687 * builtins.c: Fix a trivial typo in a comment.
15689 PR middle-end/79234
15690 * builtins.c (check_sizes): Adjust to handle reading past the end.
15691 Avoid printing excessive upper bound of ranges. Use %E to print
15692 tree nodes instead of converting them to %wu.
15693 (expand_builtin_memchr): New function.
15694 (compute_dest_size): Rename...
15695 (compute_objsize): ...to this.
15696 (expand_builtin_memcpy): Adjust.
15697 (expand_builtin_mempcpy): Adjust.
15698 (expand_builtin_strcat): Adjust.
15699 (expand_builtin_strcpy): Adjust.
15700 (check_strncat_sizes): Adjust.
15701 (expand_builtin_strncat): Adjust.
15702 (expand_builtin_strncpy): Adjust and simplify.
15703 (expand_builtin_memset): Adjust.
15704 (expand_builtin_bzero): Adjust.
15705 (expand_builtin_memcmp): Adjust.
15706 (expand_builtin): Handle memcmp.
15707 (maybe_emit_chk_warning): Check strncat just once.
15709 2017-05-04 Martin Sebor <msebor@redhat.com>
15711 PR preprocessor/79214
15712 PR middle-end/79222
15713 PR middle-end/79223
15714 * builtins.c (check_sizes): Add inlining context and issue
15715 warnings even when -Wno-system-headers is set.
15716 (check_strncat_sizes): Same.
15717 (expand_builtin_strncat): Same.
15718 (expand_builtin_memmove): New function.
15719 (expand_builtin_stpncpy): Same.
15720 (expand_builtin): Handle memmove and stpncpy.
15722 2017-05-04 Bin Cheng <bin.cheng@arm.com>
15724 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
15725 which is not used any more.
15727 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15729 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
15731 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15733 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
15734 (cortexa53_tunings): Likewise.
15735 (cortexa57_tunings): Likewise.
15736 (cortexa72_tunings): Likewise.
15737 (cortexa73_tunings): Likewise.
15739 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15741 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
15742 Set loop alignment to 8.
15744 2017-05-04 Martin Sebor <msebor@redhat.com>
15746 PR translation/80280
15747 * builtins.c (expand_builtin_object_size): Add missing quoting to
15748 %D and like directives.
15749 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
15750 (hsa_type_for_tree_type): Same.
15751 (verify_function_arguments): Same.
15752 * symtab.c (symbol_table::change_decl_assembler_name): Same.
15753 * varasm.c (get_section): Same.
15756 2017-05-04 Martin Sebor <msebor@redhat.com>
15758 PR translation/80280
15759 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
15761 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
15763 * config/aarch64/aarch64.c (generic_addrcost_table):
15764 Change HI/TI mode setting.
15766 2017-05-04 Martin Jambor <mjambor@suse.cz>
15768 PR tree-optimization/80622
15769 * tree-sra.c (comes_initialized_p): New function.
15770 (build_accesses_from_assign): Only set write lazily when
15771 comes_initialized_p is false.
15772 (analyze_access_subtree): Use comes_initialized_p.
15773 (propagate_subaccesses_across_link): Assert !comes_initialized_p
15774 instead of testing for PARM_DECL.
15776 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15778 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
15779 constraint on operand 0 to allow more general addressing modes.
15780 Adjust output template.
15781 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
15783 * config/aarch64/aarch64-protos.h
15784 (aarch64_address_valid_for_prefetch_p): Declare prototype.
15785 * config/aarch64/constraints.md (Dp): New address constraint.
15786 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
15789 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
15791 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
15792 update use of estimate_ipcp_clone_size_and_time.
15793 (estimate_local_effects): Update use of
15794 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
15795 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
15796 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
15797 Return nonspecialized time.
15799 2017-05-04 Richard Biener <rguenther@suse.de>
15801 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
15802 for the last VUSE which def dominates the PHI. Directly call
15804 (get_continuation_for_phi_1): Remove.
15806 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
15808 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
15809 to explain the use of truncating division. Cap the number of
15810 iterations to the maximum given by nb_iterations_upper_bound,
15813 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15815 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
15816 * configure: Regenerate.
15817 * config.in: Regenerate.
15818 * config/i386/driver-mingw32.c: new file.
15819 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
15820 * config.host: Link driver-mingw32.o on MinGW host.
15821 * doc/install.texi: Document new --enable-mingw-wildcard configure
15824 2017-05-04 Marek Polacek <polacek@redhat.com>
15826 PR tree-optimization/80612
15827 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
15829 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15830 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
15832 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
15833 (movt splitter): Likewise.
15834 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
15835 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
15836 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
15837 block for Thumb-1 with MOVT.
15838 (thumb2_legitimate_address_p): Move code block ...
15839 (can_avoid_literal_pool_for_label_p): ... into this new function.
15840 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
15842 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
15843 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
15844 "M-profile targets with the MOVT instruction".
15846 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15848 * config/arm/arm-builtins.c (arm_init_builtins): Rename
15849 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
15850 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
15852 2017-05-04 Martin Liska <mliska@suse.cz>
15854 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
15855 variable cond_code.
15857 2017-05-04 Richard Biener <rguenther@suse.de>
15859 * tree.c (array_at_struct_end_p): Handle arrays at struct
15860 end with flexarrays more conservatively. Refactor and treat
15861 arrays of arrays or aggregates more strict. Fix
15862 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
15863 * tree.c (array_at_struct_end_p): Adjust prototype.
15864 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
15865 * gimple-fold.c (get_range_strlen): Likewise.
15866 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
15868 2017-05-04 Richard Biener <rguenther@suse.de>
15870 PR tree-optimization/31130
15871 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
15873 (supports_overflow_infinity): Likewise.
15874 (is_negative_overflow_infinity): Likewise.
15875 (is_positive_overflow_infinity): Likewise.
15876 (is_overflow_infinity): Likewise.
15877 (stmt_overflow_infinity): Likewise.
15878 (overflow_infinity_range_p): Likewise.
15879 (usable_range_p): Remove as always returning true.
15880 (make_overflow_infinity): Remove.
15881 (negative_overflow_infinity): Likewise.
15882 (positive_overflow_infinity): Likewise.
15883 (avoid_overflow_infinity): Likewise.
15884 (set_value_range): Adjust accordingly.
15885 (set_value_range_to_nonnegative): Likewise, remove now unused
15886 overflow_infinity arg.
15887 (vrp_operand_equal_p): Adjust.
15888 (update_value_range): Likewise.
15889 (range_int_cst_singleton_p): Likewise.
15890 (operand_less_p): Likewise.
15891 (compare_values_warnv): Likewise.
15892 (extract_range_for_var_from_comparison_expr): Likewise.
15893 (vrp_int_const_binop): Likewise.
15894 (zero_nonzero_bits_from_vr): Likewise.
15895 (extract_range_from_multiplicative_op_1): Likewise.
15896 (extract_range_from_binary_expr_1): Likewise.
15897 (extract_range_from_unary_expr): Likewise.
15898 (extract_range_from_comparison): Likewise.
15899 (extract_range_basic): Likewise.
15900 (adjust_range_with_scev): Likewise.
15901 (compare_ranges): Likewise.
15902 (compare_range_with_value): Likewise.
15903 (dump_value_range): Likewise.
15904 (test_for_singularity): Likewise, remove strict_overflow_p parameter
15906 (simplify_cond_using_ranges): Adjust.
15908 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
15910 * brig-builtins.def: Added a builtin for class_f64.
15911 * builtin-types.def: Added a builtin type needed by class_f64.
15913 2017-05-03 Jason Merrill <jason@redhat.com>
15915 * timevar.def: Add TV_CONSTEXPR.
15917 2017-05-03 David Malcolm <dmalcolm@redhat.com>
15919 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
15921 2017-05-03 Martin Jambor <mjambor@suse.cz>
15923 * ipa-prop.c (ipa_update_after_lto_read): Removed.
15924 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
15925 * ipa-cp.c (ipcp_propagate_stage): Do not call
15926 ipa_update_after_lto_read.
15927 * ipa-inline.c (ipa_inline): Likewise.
15929 2017-05-03 Martin Jambor <mjambor@suse.cz>
15931 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
15932 tag. Added a default constructor and a destructor.
15933 (ipa_edge_args_sum_t): New class;
15934 (ipa_edge_args_sum): Declare.
15935 (ipa_edge_args_vector): Remove declaration.
15936 (IPA_EDGE_REF): Use ipa_edge_args_sum.
15937 (ipa_free_edge_args_substructures): Remove declaration.
15938 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
15939 (ipa_edge_args_info_available_for_edge_p): Likewise.
15940 * ipa-prop.c (ipa_edge_args_vector): Removed.
15941 (edge_removal_hook_holder): Likewise.
15942 (edge_duplication_hook_holder): Likewise.
15943 (ipa_edge_args_sum): New variable.
15944 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
15945 ipa_edge_args_vector.
15946 (ipa_free_edge_args_substructures): Likewise.
15947 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
15948 ipa_edge_args_vector.
15949 (ipa_edge_removal_hook): Turned into method
15950 ipa_edge_args_sum_t::remove.
15951 (ipa_edge_duplication_hook): Turned into method
15952 ipa_edge_args_sum_t::duplicate.
15953 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
15954 registering edge hooks.
15955 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
15956 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
15957 ipa_edge_args_sum instead of ipa_edge_args_vector.
15958 * ipa-profile.c (ipa_profile): Likewise.
15960 2017-05-03 Martin Jambor <mjambor@suse.cz>
15962 * symbol-summary.h (function_summary): New method exists.
15963 (function_summary::symtab_removal): Deallocate through release.
15964 (call_summary): New class.
15965 (gt_ggc_mx): New overload.
15966 (gt_pch_nx): Likewise.
15967 (gt_pch_nx): Likewise.
15969 2017-05-03 Jeff Law <law@redhat.com>
15971 PR tree-optimization/78496
15972 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
15973 from simplify_cond_using_ranges. Split off code to walk
15974 backwards through casts into ...
15975 (simplify_cond_using_ranges_2): New function.
15976 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
15977 (execute_vrp): After identifying jump threads, call
15978 simplify_cond_using_ranges_2.
15980 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
15983 * ipa-inline.h (inline_summary): Add ctor.
15984 (create_ggc): Do not use ggc_cleared_alloc.
15986 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
15987 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15989 * gcc.c (handle_braces): Support escaping in switch matching
15991 * doc/invoke.texi (Spec Files): Document it.
15992 Remove superfluous @code markup in items.
15994 2017-05-03 David Malcolm <dmalcolm@redhat.com>
15996 * diagnostic-show-locus.c (struct column_range): New struct.
15997 (get_affected_columns): New function.
15998 (get_printed_columns): New function.
15999 (struct correction): New struct.
16000 (correction::ensure_capacity): New function.
16001 (correction::ensure_terminated): New function.
16002 (struct line_corrections): New struct.
16003 (line_corrections::~line_corrections): New dtor.
16004 (line_corrections::add_hint): New function.
16005 (layout::print_trailing_fixits): Reimplement in terms of the new
16007 (selftest::test_overlapped_fixit_printing): New function.
16008 (selftest::diagnostic_show_locus_c_tests): Call it.
16010 2017-05-03 Nathan Sidwell <nathan@acm.org>
16012 Canonicalize canonical type hashing
16013 * tree.h (type_hash_canon_hash): Declare.
16014 * tree.c (type_hash_list, attribute_hash_list): Move into
16015 type_hash_canon_hash.
16016 (build_type_attribute_qual_variant): Break out hash code calc into
16017 type_hash_canon_hash.
16018 (type_hash_canon_hash): New. Generic type hash computation.
16019 (build_range_type_1, build_array_type_1, build_function_type,
16020 build_method_type_directly, build_offset_type, build_complex_type,
16021 make_vector_type): Call it.
16023 2017-05-03 Richard Biener <rguenther@suse.de>
16025 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
16026 When all DRs have unknown misaligned do not always peel
16027 when there is a store but apply the same costing model as if
16028 there were only loads.
16030 2017-05-03 Richard Biener <rguenther@suse.de>
16033 PR tree-optimization/80492
16034 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16035 compare_base_decls returning dont-know properly.
16037 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
16039 * config/arm/iterators.md (CCSI): New mode iterator.
16040 (arch): New mode attribute.
16041 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
16042 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
16043 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
16044 code iterator for success result mode.
16045 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
16046 the corresponding new insn generators.
16048 2017-05-03 Bin Cheng <bin.cheng@arm.com>
16051 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16052 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16054 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
16056 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
16057 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
16058 (DDR_A): Wrap DDR argument in brackets.
16059 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
16060 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
16061 (DDR_REVERSED_P): Likewise.
16063 2017-05-03 Jakub Jelinek <jakub@redhat.com>
16065 PR tree-optimization/79472
16066 * tree-switch-conversion.c (struct switch_conv_info): Add
16067 contiguous_range and default_case_nonstandard fields.
16068 (collect_switch_conv_info): Compute contiguous_range and
16069 default_case_nonstandard fields, don't clear final_bb if
16070 contiguous_range and only the default case doesn't have the required
16072 (check_all_empty_except_final): Set default_case_nonstandard instead
16073 of failing if contiguous_range and the default case doesn't have empty
16075 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
16076 and only the default case doesn't have the required constants. Skip
16078 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
16079 if default_case_nonstandard.
16080 (build_constructors): Build constant 1 just once. Assert that default
16081 values aren't inserted in between cases if contiguous_range. Skip
16083 (build_arrays): Skip virtual phis.
16084 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
16085 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
16086 Handle virtual phis.
16087 (gen_inbound_check): Handle default_case_nonstandard case.
16088 (process_switch): Adjust check_final_bb caller. Call
16089 gather_default_values with the first non-default case instead of
16090 default case if default_case_nonstandard.
16092 2017-05-02 Nathan Sidwell <nathan@acm.org>
16094 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
16095 check. Fix formatting.
16097 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
16099 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
16100 errors when comparing specialized and unspecialized times.
16102 2017-05-02 David Malcolm <dmalcolm@redhat.com>
16104 * diagnostic-show-locus.c
16105 (layout::should_print_annotation_line_p): Make private.
16106 (layout::print_annotation_line): Make private.
16107 (layout::annotation_line_showed_range_p): Make private.
16108 (layout::show_ruler): Make private.
16109 (layout::print_source_line): Make private. Pass in line and
16110 line_width, rather than calling location_get_source_line. Drop
16112 (layout::print_leading_fixits): New method.
16113 (layout::print_any_fixits): Rename to...
16114 (layout::print_trailing_fixits): ...this, and make private.
16115 Don't print newline fixits.
16116 (diagnostic_show_locus): Move logic for printing one row into...
16117 (layout::print_line): ...this new function. Move the
16118 location_get_source_line call and error-handling from
16119 print_source_line to here. Call print_leading_fixits, and rename
16120 print_any_fixits to print_trailing_fixits.
16121 (selftest::test_fixit_insert_containing_newline): Update now that
16122 newlines are partially supported.
16123 (selftest::test_fixit_insert_containing_newline_2): New test.
16124 (selftest::test_fixit_replace_containing_newline): Update comments.
16125 (selftest::diagnostic_show_locus_c_tests): Call the new test.
16126 * edit-context.c (class added_line): New class.
16127 (class edited_line): Describe newline handling in comment.
16128 (edited_line::actually_edited_p): New method.
16129 (edited_line::print_content): Delete redundant decl.
16130 (edited_line::m_predecessors): New field.
16131 (edited_file::print_content): Call edited_line::print_content.
16132 (edited_file::print_diff): Update to support newlines.
16133 (edited_file::print_diff_hunk): Likewise.
16134 (edited_file::print_run_of_changed_lines): New function.
16135 (edited_file::print_diff_line): Convert to...
16136 (print_diff_line): ...this.
16137 (edited_file::get_effective_line_count): New function.
16138 (edited_line::edited_line): Initialize new field m_predecessors.
16139 (edited_line::~edited_line): Clean up m_predecessors.
16140 (edited_line::apply_fixit): Handle newlines.
16141 (edited_line::get_effective_line_count): New function.
16142 (edited_line::print_content): New function.
16143 (edited_line::print_diff_lines): New function.
16144 (selftest::test_applying_fixits_insert_containing_newline): New
16146 (selftest::test_applying_fixits_replace_containing_newline): New
16148 (selftest::insert_line): New function.
16149 (selftest::test_applying_fixits_multiple_lines): Add example of
16151 (selftest::edit_context_c_tests): Call the new tests.
16153 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16155 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
16156 parameter cand. Update dump information.
16157 (get_computation_cost): Update uses.
16159 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16161 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
16162 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
16163 (get_computation_at, rewrite_use_address): Update use of
16164 get_computation_aff.
16166 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16168 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
16169 (get_computation): Delete.
16170 (get_computation_cost): Implement like get_computation_cost_at.
16171 Use get_computation_at.
16172 (get_computation_cost_at): Delete.
16173 (rewrite_use_nonlinear_expr): Use get_computation_at.
16174 (rewrite_use_compare, remove_unused_ivs): Ditto.
16176 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16178 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
16180 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16182 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
16183 (ivopts_global_cost_for_size): Rename parameter and update uses.
16184 (iv_ca_recount_cost): Update uses.
16185 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
16186 candidates seperately in n_invs and n_cands.
16187 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
16189 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16191 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
16192 (find_inv_vars_cb): New.
16193 (find_depends): Renamed to ...
16194 (find_inv_vars): ... this.
16195 (add_candidate_1, force_var_cost): Call find_inv_vars.
16196 (split_address_cost, determine_group_iv_cost_cond): Ditto.
16198 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16200 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
16201 inv_vars. Add inv_exprs.
16202 (struct iv_cand): Rename depends_on to inv_vars.
16203 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
16204 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
16205 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
16206 (dump_cand): Dump inv_vars.
16207 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
16208 (record_invariant, find_depends, add_candidate_1): Ditto.
16209 (set_group_iv_cost, force_var_cost): Ditto.
16210 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
16211 (get_computation_cost_at, get_computation_cost): Ditto.
16212 (determine_group_iv_cost_generic): Ditto.
16213 (determine_group_iv_cost_address): Ditto.
16214 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
16215 (determine_group_iv_costs): Ditto.
16216 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
16217 (iv_ca_set_remove_invariants): Renamed to ...
16218 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
16219 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
16220 (iv_ca_set_add_invariants): Renamed to ...
16221 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
16222 (iv_ca_set_cp): Use iv_ca_set_add_invs.
16223 (iv_ca_has_deps): Support inv_vars and inv_exprs.
16224 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
16225 (create_new_ivs): Remove useless dump.
16227 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16229 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
16231 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
16232 (iv_ca_set_no_cp, create_new_iv): Ditto.
16234 2017-05-02 Bin Cheng <bin.cheng@arm.com>
16236 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
16238 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16240 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
16241 function tree_check2.
16243 2017-05-02 Martin Liska <mliska@suse.cz>
16245 * doc/gcov.texi: Add missing preposition.
16246 * gcov.c (function_info::function_info): Properly fill up
16247 all member variables.
16249 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16251 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
16253 2017-05-02 Tamar Christina <tamar.christina@arm.com>
16255 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
16257 2017-05-02 Martin Liska <mliska@suse.cz>
16260 * lto-streamer-in.c (lto_read_tree_1): Remove
16261 LTO_STREAMER_DEBUG.
16262 * lto-streamer.c (struct tree_hash_entry): Likewise.
16263 (struct tree_entry_hasher): Likewise.
16264 (tree_entry_hasher::hash): Likewise.
16265 (tree_entry_hasher::equal): Likewise.
16266 (lto_streamer_init): Likewise.
16267 (lto_orig_address_map): Likewise.
16268 (lto_orig_address_get): Likewise.
16269 (lto_orig_address_remove): Likewise.
16270 * lto-streamer.h: Likewise.
16271 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
16272 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
16274 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
16276 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
16277 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
16278 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
16279 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
16280 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
16281 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
16282 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
16283 (mm_maskz_sub_ss): New intrinsics.
16284 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16285 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
16286 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
16287 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
16288 (__builtin_ia32_subss_mask_round): New builtins.
16289 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
16290 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
16291 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
16293 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
16294 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
16296 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
16299 2017-05-02 Martin Jambor <mjambor@suse.cz>
16301 PR tree-optimization/78687
16302 * tree-sra.c (access): New field parent.
16303 (process_subtree_disqualification): New function.
16304 (disqualify_candidate): Call it.
16305 (build_accesses_from_assign): Reset write flag if creating an
16307 (build_access_subtree): Fill in parent field and also prpagate
16308 down grp_write flag.
16309 (create_artificial_child_access): New parameter set_grp_write, set
16310 grp_write to its value.
16311 (propagate_subaccesses_across_link): Also propagate grp_write flag
16313 (propagate_all_subaccesses): Push the closest parent back to work
16314 queue if add_access_to_work_queue returned true.
16316 2017-05-02 Richard Biener <rguenther@suse.de>
16318 * common.opt (fstrict-overflow): Alias negative to fwrapv.
16319 * doc/invoke.texi (fstrict-overflow): Remove all traces of
16320 -fstrict-overflow documentation.
16321 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
16322 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
16323 flag_strict_overflow.
16324 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
16325 * lto-opts.c (lto_write_options): Do not stream it.
16326 * lto-wrapper.c (merge_and_complain): Do not handle it.
16327 * opts.c (default_options_table): Do not set -fstrict-overflow.
16328 (finish_options): Likewise do not clear it when sanitizing.
16329 * simplify-rtx.c (simplify_const_relational_operation): Do not
16330 test flag_strict_overflow.
16332 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16334 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
16335 using enabled attribute.
16336 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
16337 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
16338 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
16339 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
16340 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
16341 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
16342 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
16343 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
16344 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
16345 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
16347 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
16349 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
16351 2017-05-02 Richard Biener <rguenther@suse.de>
16353 PR tree-optimization/80591
16355 2017-04-10 Richard Biener <rguenther@suse.de>
16357 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16360 2017-05-02 Richard Biener <rguenther@suse.de>
16362 PR tree-optimization/80549
16363 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
16364 (cleanup_tree_cfg_noloop): Create forwarders to known loop
16365 headers if they do not have a preheader.
16367 2017-05-02 Martin Liska <mliska@suse.cz>
16370 * common.opt: Fix typo.
16371 * doc/invoke.texi: Likewise.
16373 2017-05-01 Jan Beulich <jbeulich@suse.com>
16375 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
16376 swapping, add (x,x,m,x,n) alternative.
16378 2017-05-01 Nathan Sidwell <nathan@acm.org>
16380 * calls.c (combine_pending_stack_adjustment_and_call): Remove
16381 unnecessary unadjusted_alignment check.
16383 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
16386 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
16388 * gimplify.c (gimplify_cilk_detach): New function.
16389 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
16390 * tree-core.h: Document EXPR_CILK_SPAWN.
16391 * tree.h (EXPR_CILK_SPAWN): Define.
16393 2017-05-01 David Malcolm <dmalcolm@redhat.com>
16395 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
16396 to use new fixit_hint representation, using the "replace" logic.
16397 (get_line_span_for_fixit_hint): Likewise.
16398 (layout::print_any_fixits): Likewise.
16399 (selftest::test_one_liner_many_fixits): Rename to...
16400 (selftest::test_one_liner_many_fixits_1): ...this, and update
16401 comment and expected output to reflect that the multiple fix-it
16402 hints are now consolidated into one insertion.
16403 (selftest::test_one_liner_many_fixits_2): New test.
16404 (selftest::test_diagnostic_show_locus_one_liner): Update for
16406 (selftest::test_fixit_consolidation): Update for fix-it API
16408 * diagnostic.c (print_parseable_fixits): Likewise.
16409 * edit-context.c (edited_line::m_line_events): Convert from
16410 auto_vec <line_event *> to auto_vec <line_event>.
16411 (class line_event): Convert from abstract base class to a concrete
16412 class, taking over the role of replace_event.
16413 (class insert_event): Delete.
16414 (class replace_event): Rename to class line_event. Convert to
16416 (edit_context::add_fixits): Reimplement.
16417 (edit_context::apply_insert): Delete.
16418 (edit_context::apply_replace): Rename to...
16419 (edit_context::apply_fixit): ...this. Convert to half-open range.
16420 (edited_file::apply_insert): Delete.
16421 (edited_file::apply_replace): Rename to...
16422 (edited_file::apply_fixit): ...this.
16423 (edited_line::~edited_line): Drop deletion of events.
16424 (edited_line::apply_insert): Delete.
16425 (edited_line::apply_replace): Rename to...
16426 (edited_line::apply_fixit): ...this. Convert to half-open range.
16427 Update for change to type of m_line_events.
16428 * edit-context.h (edit_context::apply_insert): Delete.
16429 (edit_context::apply_replace): Rename to...
16430 (edit_context::apply_fixit): ...this.
16432 2017-05-01 Martin Sebor <msebor@redhat.com>
16434 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
16437 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
16440 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
16441 __get_cpuid_max returns 0.
16442 (__get_cpuid_count): Ditto.
16444 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
16446 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
16447 replacement expression is another instance of one of its arguments.
16449 2017-05-01 Jakub Jelinek <jakub@redhat.com>
16452 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
16453 check for stack push/pop autoinc.
16454 * config/i386/i386.c (ix86_agi_dependent): Return false
16455 if the only reason why modified_in_p returned true is that
16456 addr is SP based and set_insn is a push or pop.
16458 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16460 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
16463 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
16466 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
16467 (account_size_time): Use two predicates - exec_pred and
16469 (evaluate_conditions_for_known_args): Compute both clause and
16471 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
16472 (inline_summary_t::duplicate): Update.
16473 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
16475 (compute_inline_parameters): Likewise.
16476 (estimate_edge_size_and_time): Update caluclation of time.
16477 (estimate_node_size_and_time): Compute both time and nonspecialized
16479 (estimate_ipcp_clone_size_and_time): Update.
16480 (inline_merge_summary): Update.
16481 (do_estimate_edge_time): Update.
16482 (do_estimate_edge_size): Update.
16483 (do_estimate_edge_hints): Update.
16484 (inline_read_section, inline_write_summary): Stream both new predicates.
16485 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
16487 (compute_inlined_call_time): Cleanup.
16488 (big_speedup_p): Update.
16489 (edge_badness): Update.
16490 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
16491 (size_time_entry): Replace predicate by exec_predicate and
16492 nonconst_predicate.
16493 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
16494 (estimate_edge_time): Return also nonspec_time.
16495 (reset_edge_growth_cache): Update.
16497 2017-04-29 Jakub Jelinek <jakub@redhat.com>
16499 PR rtl-optimization/80491
16500 * ifcvt.c (noce_process_if_block): When looking for x setter
16501 with missing else_bb, don't check only the insn right before
16502 cond_earliest, but look for the last insn that x is modified in
16503 within the same bb.
16505 PR rtl-optimization/80491
16506 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
16508 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
16510 PR tree-optimization/80487
16511 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
16513 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
16515 PR tree-optimization/79697
16516 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
16517 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
16518 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
16520 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
16521 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
16523 2017-04-28 Martin Sebor <msebor@redhat.com>
16525 PR tree-optimization/80523
16526 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
16527 (init_target_to_host_charmap, target_to_host, target_strtol10): New
16529 (maybe_warn, format_directive, parse_directive): Use new functions.
16530 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
16532 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
16534 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
16536 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
16538 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
16539 target_header_dir): Set correctly.
16540 * configure: Regenerated.
16541 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
16542 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
16543 instead of SYSTEM_HEADER_DIR.
16545 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
16547 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
16548 (estimate_local_effects): Likewise.
16549 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
16550 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
16551 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
16552 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
16553 do_estimate_edge_time, estimate_edge_time): Likewise.
16554 * ipa-inline-analysis.c (estimate_node_size_and_time,
16555 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
16556 (estimate_time_after_inlining): Remove.
16558 2017-04-28 Martin Liska <mliska@suse.cz>
16560 * doc/gcov.texi: Enhance documentation of gcov.
16562 2017-04-28 Martin Liska <mliska@suse.cz>
16564 * doc/gcov.texi: Sort options in alphabetic order.
16565 * doc/gcov-dump.texi: Likewise.
16566 * doc/gcov-tool.texi: Likewise.
16567 * gcov.c (print_usage): Likewise.
16568 * gcov-dump.c (print_usage): Likewise.
16569 * gcov-tool.c (print_merge_usage_message): Likewise.
16570 (print_rewrite_usage_message): Likewise.
16571 (print_overlap_usage_message): Likewise.
16573 2017-04-28 Martin Liska <mliska@suse.cz>
16575 PR gcov-profile/53915
16576 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
16578 2017-04-28 Martin Liska <mliska@suse.cz>
16580 PR gcov-profile/79891
16581 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
16582 is marked by compiler as living on a line.
16583 (get_cycles_count): Remove usage of the union.
16584 (output_intermediate_file): Likewise.
16585 (find_source): Fix GNU coding style.
16586 (accumulate_line_counts): Remove old non-all block mode.
16587 (output_lines): Remove usage of the union.
16588 * profile.c (output_location): Include all BBs, even if
16589 belonging to a same line (and file) as a previous BB.
16591 2017-04-28 Martin Liska <mliska@suse.cz>
16593 * gcov.c (process_args): Handle new argument 'w'.
16594 (read_graph_file): Assign ID to BBs.
16595 (output_branch_count): Display BB # if verbose flag is set.
16596 (output_lines): Likewise for arcs.
16597 (print_usage): Add '--verbose' option help.
16598 * doc/gcov.texi: Document --verbose (-w) option.
16600 2017-04-28 Martin Liska <mliska@suse.cz>
16602 * gcov.c (struct block_location_info): New struct.
16603 (process_file): Fill up the new structure.
16604 (read_graph_file): Replace usage of encoding by the newly added
16606 (add_line_counts): Likewise.
16607 (accumulate_line_counts): Remove usage of the union.
16608 (function_info::function_info): New function.
16609 (function_info::~function_info): Likewise.
16610 (process_file): Call delete instead of release_function.
16611 (release_function): Release the function.
16612 (release_structures): Call delete instead of release_function.
16613 (solve_flow_graph): Replace usage of num_blocks.
16614 (find_exception_blocks): Likewise.
16615 (output_lines): Fix GNU coding style.
16617 2017-04-28 Martin Liska <mliska@suse.cz>
16620 * coverage.c (coverage_remove_note_file): New function.
16621 * coverage.h: Declare the function.
16622 * toplev.c (finalize): Clean if an error has been seen.
16624 2017-04-28 Martin Liska <mliska@suse.cz>
16626 PR gcov-profile/80031
16627 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
16628 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
16629 * gcov.c (read_graph_file): Read just number of blocks.
16630 * profile.c (branch_prob): Do not stream 0 flags per a basic
16633 2017-04-28 Martin Liska <mliska@suse.cz>
16635 * gcov-dump.c (tag_*): Add new argument to declarations.
16636 (dump_gcov_file): Likewise.
16637 (tag_blocks): Add and use new argument depth.
16638 (tag_arcs): Likewise.
16639 (tag_lines): Likewise.
16640 (tag_counters): Likewise.
16641 (tag_summary): Likewise.
16642 (dump_working_sets): Use depth to do a proper indentation.
16644 2017-04-28 Jakub Jelinek <jakub@redhat.com>
16647 * cgraph.h (symtab_node::debug_symtab): No longer inline.
16648 * symtab.c (symtab_node::debug_symtab): Move definition here.
16650 2017-04-28 Richard Biener <rguenther@suse.de>
16652 * lto-streamer.h (LTO_major_version): Bump to 7.
16654 2017-04-28 Richard Biener <rguenther@suse.de>
16656 * tree-vrp.c (assert_info): New struct.
16657 (add_assert_info): New helper.
16658 (register_edge_assert_for_2): Refactor to add asserts to a vector
16660 (register_edge_assert_for_1): Likewise.
16661 (register_edge_assert_for): Likewise.
16662 (finish_register_edge_assert_for): New helper actually registering
16663 asserts where live on edge.
16664 (find_conditional_asserts): Adjust.
16665 (find_switch_asserts): Likewise.
16666 (evrp_dom_walker::try_find_new_range): Generalize.
16667 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
16669 2017-04-27 Marek Polacek <polacek@redhat.com>
16672 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
16673 arg10 and arg11 to itype.
16675 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
16677 * doc/extend.texi (Object Size Checking): Improve grammar.
16679 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
16682 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
16683 that the logic for permitting reciprocal estimates matches that
16686 2017-04-27 Jakub Jelinek <jakub@redhat.com>
16689 * tree.c (type_cache_hasher::equal): Only compare
16690 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
16691 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
16692 non-aggregate element types.
16693 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
16694 about the flag on ARRAY_TYPEs in the comment, formatting fix.
16696 2017-04-27 Richard Biener <rguenther@suse.de>
16698 PR middle-end/80533
16699 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
16700 stripping ARRAY_REFs from MEM_EXPR make sure we're not
16701 keeping a reference to a trailing array.
16703 2017-04-27 Richard Biener <rguenther@suse.de>
16705 PR middle-end/80539
16706 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
16707 being in loop-closed SSA form conservatively.
16708 (chrec_fold_multiply_poly_poly): Likewise.
16710 2017-04-27 Tamar Christina <tamar.christina@arm.com>
16712 PR middle-end/79665
16713 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
16714 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
16716 2017-04-27 Jakub Jelinek <jakub@redhat.com>
16719 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
16720 (aarch64_function_arg_alignment): Return unsigned int again, but still
16721 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
16722 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
16723 Don't emit -Wpsabi note.
16724 (aarch64_function_arg_boundary): Likewise.
16725 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
16728 2017-04-26 Nathan Sidwell <nathan@acm.org>
16730 * tree.h (crc32_unsigned_n): Declare.
16731 (crc32_unsigned, crc32_unsigned): Make inline.
16732 * tree.c (crc32_unsigned_bits): Replace with ...
16733 (crc32_unsigned_n): ... this.
16734 (crc32_unsigned, crc32_byte): Remove.
16735 (crc32_string): Remove unnecessary braces.
16737 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
16739 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
16740 * ipa-inline-analysis.c (MAX_TIME): Remove.
16741 (account_size_time): Use sreal for time.
16742 (dump_inline_summary): Update.
16743 (estimate_function_body_sizes): Update.
16744 (estimate_edge_size_and_time): Update.
16745 (estimate_calls_size_and_time): Update.
16746 (estimate_node_size_and_time): Update.
16747 (inline_merge_summary): Update.
16748 (inline_update_overall_summary): Update.
16749 (estimate_time_after_inlining): Update.
16750 (inline_read_section): Update.
16751 (inline_write_summary): Update.
16752 * ipa-inline.c (compute_uninlined_call_time): Update.
16753 (compute_inlined_call_time): Update.
16754 (recursive_inlining): Update.
16755 (inline_small_functions): Update.
16756 (dump_overall_stats): Update.
16757 * ipa-inline.h: Include sreal.h.
16758 (size_time_entry): Turn time to sreal.
16759 (inline_summary): Turn self_time nad time to sreal.
16761 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
16763 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
16765 (sreal::stream_out, sreal::stream_in): New.
16766 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
16768 2017-04-25 Jakub Jelinek <jakub@redhat.com>
16770 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
16773 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
16776 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
16778 (dimode_scalar_chain::compute_convert_gain): Ditto.
16779 (dimode_scalar_chain::make_vector_copies): Ditto.
16780 (dimode_scalar_chain::convert_reg): Ditto.
16781 (dimode_scalar_chain::convert_insn): Ditto.
16782 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
16783 (VI248_AVX512BW_1): New mode iterator.
16784 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
16785 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
16788 2017-04-25 Martin Sebor <msebor@redhat.com>
16790 PR tree-optimization/80497
16791 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
16792 constants are representable in HOST_WIDE_INT.
16793 (parse_directive): Ditto.
16795 2017-04-25 Martin Sebor <msebor@redhat.com>
16798 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
16799 (new_zero_array): Adjust signature.
16800 (dom_info::dom_init): Used unsigned rather that size_t.
16801 (dom_info::dom_info): Same.
16803 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16804 Jakub Jelinek <jakub@redhat.com>
16807 * config/arm/arm.c: Include gimple.h.
16808 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
16809 returns negative, increment ncrn only if it returned positive.
16810 (arm_needs_doubleword_align): Return int instead of bool,
16811 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
16812 members, but if there is any such non-FIELD_DECL
16813 > PARM_BOUNDARY aligned decl, return -1 instead of false.
16814 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
16815 returns negative, increment nregs only if it returned positive.
16816 (arm_setup_incoming_varargs): Likewise.
16817 (arm_function_arg_boundary): Emit -Wpsabi note if
16818 arm_needs_doubleword_align returns negative, return
16819 DOUBLEWORD_ALIGNMENT only if it returned positive.
16821 2017-04-25 Marek Polacek <polacek@redhat.com>
16824 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
16825 first argument to type.
16827 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
16830 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
16831 type checks to test for compatibility instead of equality.
16833 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16834 Jakub Jelinek <jakub@redhat.com>
16837 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
16839 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
16840 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
16841 the alignment computation, but return their maximum in warn_alignment.
16842 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
16843 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
16845 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
16846 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
16849 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16851 * config/arc/simdext.md (dmpyh): Fix typo.
16853 2017-04-25 Richard Biener <rguenther@suse.de>
16855 PR tree-optimization/80492
16856 * alias.c (compare_base_decls): Handle registers with asm
16857 specification conservatively.
16858 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
16859 compare_base_decls returning dont-know properly.
16861 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16863 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
16864 (legitimate_offset_address_p): New function.
16865 (arc_legitimate_address_p): Use above function.
16867 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16869 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
16871 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16873 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
16874 ACCH registers whenever they are available.
16876 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16878 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
16879 double regs fix when not used.
16881 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16883 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
16885 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
16886 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
16888 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16890 * config/arc/arc.c (arc_output_addsi): Check for h-register class
16891 when emitting short ADD instructions.
16893 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
16895 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
16897 (cmpsi_cc_c_insn): Likewise.
16898 (cbranchsi4_scratch): Compute proper instruction length using
16899 compact_hreg_operand.
16900 * config/arc/predicates.md (compact_hreg_operand): New predicate.
16902 2017-04-25 Richard Biener <rguenther@suse.de>
16904 PR middle-end/80509
16905 * passes.c (pass_manager::pass_manager): Initialize
16906 m_name_to_pass_map.
16908 2017-04-25 Richard Biener <rguenther@suse.de>
16910 PR tree-optimization/79201
16911 * tree-ssa-sink.c (statement_sink_location): Handle calls.
16913 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16916 * config/s390/vector.md: Split MEM->GPR vector moves for
16917 non-s_operand addresses.
16919 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16922 * config/s390/predicates.md (reload_const_wide_int_operand): New
16924 * config/s390/s390.md ("movti"): Remove d/P alternative.
16925 ("movti_bigconst"): New pattern definition.
16927 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16930 * s390-protos.h (s390_expand_cs_hqi): Removed.
16931 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
16932 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
16933 modes as well as CCZ1mode and CCZmode.
16934 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
16935 signature of s390_emit_compare_and_swap.
16936 (s390_expand_cs_hqi): Likewise, make static.
16937 (s390_expand_cs_tdsi): Generate an explicit compare before trying
16938 compare-and-swap, in some cases.
16939 (s390_expand_cs): Wrapper function.
16940 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
16942 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
16943 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
16944 patterns for small and large integers. Forbid symref memory operands.
16945 Move expander to s390.c. Require cc register.
16946 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
16947 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
16948 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
16949 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
16950 symref memory operands. Remove CC mode and call s390_match_ccmode
16952 ("atomic_exchange<mode>"): Allow and implement all integer modes.
16954 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16956 * config/s390/s390.md (define_peephole2): New peephole to help
16957 combining the load-and-test pattern with volatile memory.
16959 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
16961 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
16962 with CCZmode for TARGET_Z196.
16964 2017-04-25 Jakub Jelinek <jakub@redhat.com>
16966 PR rtl-optimization/80501
16967 * combine.c (make_compound_operation_int): Set subreg_code to SET
16968 even for AND with mask of the sign bit of mode.
16970 PR rtl-optimization/80500
16971 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
16972 sum's initial value.
16974 2017-04-25 Julian Brown <julian@codesourcery.com>
16975 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16977 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
16979 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
16981 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
16983 2017-04-25 Julian Brown <julian@codesourcery.com>
16984 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16986 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
16987 (thunderx2t99_sha): New Reservation.
16989 2017-04-25 Julian Brown <julian@codesourcery.com>
16990 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16992 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
16993 type for 1-element load.
16995 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
16997 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
16999 2017-04-24 Martin Jambor <mjambor@suse.cz>
17001 PR tree-optimization/80293
17002 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
17003 char arrays not totally scalarizable if it is false.
17004 (analyze_all_variable_accesses): Pass correct value in the new
17005 parameter. Add a statistics counter.
17007 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
17009 PR middle-end/79931
17010 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
17012 2017-04-24 Richard Biener <rguenther@suse.de>
17014 PR tree-optimization/80494
17015 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
17016 out for complex types.
17018 2017-04-24 Richard Biener <rguenther@suse.de>
17020 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
17021 * tree-ssa-sccvn.c (print_scc): Print SCC size.
17022 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
17023 (DFS): Adjust and never fail.
17024 (sccvn_dom_walker::fail): Remove.
17025 (sccvn_dom_walker::before_dom_children): Adjust.
17026 (run_scc_vn): Likewise and never fail.
17027 * tree-ssa-pre.c (pass_pre::execute): Adjust.
17028 (pass_fre::execute): Likewise.
17030 2017-04-24 Richard Biener <rguenther@suse.de>
17032 PR tree-optimization/79725
17033 * tree-ssa-sink.c (statement_sink_location): Return whether
17034 failure reason was zero uses. Move that check later.
17035 (sink_code_in_bb): Deal with zero uses by removing the stmt
17038 2017-04-24 Richard Biener <rguenther@suse.de>
17041 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
17042 pointer-based references.
17044 2017-04-24 Richard Biener <rguenther@suse.de>
17047 * pass_manager.h (pass_manager::operator new): Remove.
17048 (pass_manager::operator delete): Likewise.
17049 * passes.c (pass_manager::operator new): Remove.
17050 (pass_manager::operator delete): Likewise.
17051 (pass_manager::pass_manager): Zero individual pass members.
17053 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
17056 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
17057 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
17058 Check "XEXP (src, 1)" operand here.
17059 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
17060 Check "XEXP (src, 1)" operand here.
17061 (dimode_scalar_chain::make_vector_copies): Detect count register
17062 of a shift instruction. Zero extend count register from QImode
17063 to DImode to satisfy vector shift pattern count operand predicate.
17064 Substitute vector shift count operand with a DImode copy.
17065 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
17068 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
17070 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
17071 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17072 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
17073 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
17074 (UNSPEC_NOREX_MEM): Remove definition.
17076 2017-04-21 Richard Biener <rguenther@suse.de>
17078 PR tree-optimization/79547
17079 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
17080 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
17081 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
17082 without any constraints.
17084 2017-04-21 Richard Biener <rguenther@suse.de>
17086 PR tree-optimization/78847
17087 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
17089 2017-04-21 Richard Biener <rguenther@suse.de>
17091 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
17092 (build_distinct_type_copy): Likewise.
17093 (build_variant_type_copy): Likewise.
17094 * tree.c (build_qualified_type): Pass down mem-stat info.
17095 (build_distinct_type_copy): Likewise.
17096 (build_variant_type_copy): Likewise.
17098 2017-04-21 Richard Biener <rguenther@suse.de>
17100 PR tree-optimization/80237
17101 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
17103 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
17104 for a simplified result.
17106 2016-04-21 Richard Biener <rguenther@suse.de>
17108 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
17109 sth as strict as a simple_iv but a chrec without symbols and an
17110 operand defined in the loop we are peeling (and not some subloop).
17111 (propagate_constants_for_unrolling): Propagate all constants.
17113 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17116 * config/i386/i386.c (print_reg): Remove assert for disalowed
17117 regno values, call output_operand_lossage instead.
17119 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
17122 * config/i386/constraints.md (Yc): New register constraint.
17123 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
17124 Use Yc constraint for alternative 2 of operand 0. Remove
17125 preferred_for_speed attribute.
17127 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17129 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
17130 lastprivate clauses in SIMT case.
17132 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
17134 * doc/invoke.texi (-Wextra-semi): Document new warning option.
17136 2017-04-20 Richard Biener <rguenther@suse.de>
17138 PR tree-optimization/57796
17139 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
17140 as N scalar stores.
17141 (vect_model_load_cost): Cost gathers as N scalar loads.
17143 2017-04-20 Richard Biener <rguenther@suse.de>
17145 * ggc-page.c (ggc_allocated_p): Rename to ...
17146 (safe_lookup_page_table_entry): ... this and return the lookup
17148 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
17150 2017-04-20 Richard Biener <rguenther@suse.de>
17152 PR tree-optimization/80453
17153 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
17154 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
17155 from the conditions.
17156 (vn_phi_eq): Pass them down.
17157 (vn_phi_lookup): Record them.
17158 (vn_phi_insert): Likewise.
17160 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
17162 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
17163 uninitialized variable warning to avoid buffer overrun.
17165 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
17168 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
17169 is suppressed for '{ 0 }' in C.
17171 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17173 * BASE-VER: Set to 8.0.0.
17175 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
17177 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
17178 priority .init_array and .fini_array section with SECTION_NOTYPE
17181 2017-04-20 Jakub Jelinek <jakub@redhat.com>
17183 PR middle-end/80423
17184 * tree.h (build_array_type): Add typeless_storage default argument.
17185 * tree.c (type_cache_hasher::equal): Also compare
17186 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
17187 (build_array_type): Add typeless_storage argument, set
17188 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
17190 (build_nonshared_array_type): Adjust build_array_type_1 caller.
17191 (build_array_type): Likewise. Add typeless_storage argument.
17193 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17194 Jakub Jelinek <jakub@redhat.com>
17196 PR tree-optimization/80426
17197 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
17198 operation on symbolic operands, also compute the overflow for the
17199 invariant part when the operation degenerates into a negation.
17201 2017-04-19 Jakub Jelinek <jakub@redhat.com>
17204 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
17205 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
17208 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
17210 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
17213 * config/avr/avr.c (tree.h): Include it.
17214 (cgraph.h): Include it.
17215 (avr_encode_section_info): Don't warn for uninitialized progmem
17216 variable if it's just an alias.
17218 2017-04-19 Richard Biener <rguenther@suse.de>
17221 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
17222 when needed by AutoPGO.
17224 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
17227 * doc/lto.texi: Remove an extra 'that'.
17229 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
17231 PR rtl-optimization/80429
17232 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
17233 are only used in debug insns.
17235 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
17236 Vladimir Makarov <vmakarov@redhat.com>
17238 * config/sparc/predicates.md (input_operand): Add comment. Return
17239 true for any memory operand when LRA is in progress.
17240 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
17242 2017-04-18 Jeff Law <law@redhat.com>
17245 * mips.md ({return,simple_return}_internal): Do not overwrite
17248 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17250 PR tree-optimization/80443
17251 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
17252 instead of adding 1, subtract -1 and similarly instead of subtracting
17255 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
17257 PR rtl-optimization/80357
17258 * haifa-sched.c (tmp_bitmap): New variable.
17259 (model_recompute): Handle duplicate use records.
17260 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
17261 (free_global_sched_pressure_data): Free it.
17263 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
17266 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
17267 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
17268 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
17269 instead of SYSTEM_HEADER_DIR.
17271 2017-04-18 Jeff Law <law@redhat.com>
17273 PR middle-end/80422
17274 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
17275 predecessors after walking up the insn chain.
17277 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17280 * dwarf2out.c (modified_type_die): Try harder not to emit internal
17281 sizetype type into debug info.
17283 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
17286 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
17287 unneeded test for TARGET_UPPER_REGS_SF.
17288 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
17290 2017-04-18 Jakub Jelinek <jakub@redhat.com>
17293 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
17294 instead of gsi_after_labels.
17296 2017-04-18 Jeff Law <law@redhat.com>
17298 * regcprop.c (maybe_mode_change): Avoid creating copies of the
17302 2017-04-13 Jeff Law <law@redhat.com>
17303 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17304 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17306 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
17309 * config/avr/avr.c (intl.h): Include it.
17310 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
17312 2017-04-18 Martin Liska <mliska@suse.cz>
17314 PR gcov-profile/78783
17315 * gcov-tool.c (gcov_output_files): Validate that destination
17316 file is either removed by the tool or by a user.
17318 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17319 Guy Benyei <guybe@mellanox.com>
17321 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
17322 block, and do not negate it, the stored id is already negative.
17324 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
17326 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
17328 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17331 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
17332 masks of options that should be turned off if the VSX vector
17333 options are turned off.
17334 (OTHER_P8_VECTOR_MASKS): Likewise.
17335 (OTHER_VSX_VECTOR_MASKS): Likewise.
17336 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
17337 rs6000_disable_incompatible_switches to validate no type switches
17339 (rs6000_incompatible_switch): New function to disallow turning on
17340 other vector options if -mno-vsx, -mno-power8-vector, or
17341 -mno-power9-vector are specified.
17343 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17345 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
17347 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17349 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
17350 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
17351 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
17352 (ARG_POINTER_CFA_OFFSET): Likewise.
17354 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
17356 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
17357 conditions to take advantage of various optimizations.
17359 2017-04-13 Jeff Law <law@redhat.com>
17361 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
17362 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
17363 (zero_extendsidi2_dext): Likewise.
17365 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17368 * fold-const.c (fold_ternary_loc): Revert
17369 use op0 instead of fold_convert_loc (loc, type, arg0) part of
17372 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
17374 PR rtl-optimization/80343
17375 * lra-remat.c (update_scratch_ops): Assign original hard reg to
17376 new scratch pseudo.
17378 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
17381 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
17382 to ubsan_encode_value.
17384 2017-04-13 Jeff Law <law@redhat.com>
17386 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
17387 appearing in DEBUG_INSNs.
17389 2017-04-13 Martin Liska <mliska@suse.cz>
17391 PR gcov-profile/80413
17392 * gcov-io.c (gcov_write_string): Copy to buffer just when
17393 allocated size is greater than zero.
17395 2017-04-13 Jakub Jelinek <jakub@redhat.com>
17398 * dwarf2out.c (decls_for_scope): Ignore declarations of
17399 current_function_decl in BLOCK_NONLOCALIZED_VARS.
17401 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
17404 * ipa-visibility.c (non_local_p): Fix typos.
17405 (localize_node): When localizing symbol in same comdat group,
17406 dissolve the group only when we know external symbols are going
17408 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
17410 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17412 PR tree-optimization/79390
17413 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
17414 order does not result in usable sequence, retry with reversed operand
17420 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
17421 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
17422 op0 instead of fold_convert_loc (loc, type, arg0).
17424 2017-04-12 Jeff Law <law@redhat.com>
17426 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
17427 has a delay slot in the generated code.
17429 * config/cris/cris.md (cris_preferred_reload_class): Return
17430 GENNONACR_REGS rather than GENERAL_REGS.
17432 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17435 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
17436 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
17437 signedness of the result type.
17439 2017-04-12 Richard Biener <rguenther@suse.de>
17440 Jeff Law <law@redhat.com>
17442 PR tree-optimization/80359
17443 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
17444 trim stores to TARGET_MEM_REFs.
17446 2017-04-12 Richard Biener <rguenther@suse.de>
17448 PR tree-optimization/79390
17449 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
17450 threading case even more.
17452 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
17455 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
17456 for quad_address_p for TImode, instead of just not indexed_address.
17458 2017-04-12 Richard Biener <rguenther@suse.de>
17459 Bernd Edlinger <bernd.edlinger@hotmail.de>
17461 PR middle-end/79671
17462 * alias.c (component_uses_parent_alias_set_from): Handle
17463 TYPE_TYPELESS_STORAGE.
17464 (get_alias_set): Likewise.
17465 * tree-core.h (tree_type_common): Add typeless_storage flag.
17466 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
17467 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
17468 for types containing members with TYPE_TYPELESS_STORAGE.
17469 (place_field): Likewise.
17470 (layout_type): Likewise for ARRAY_TYPE.
17471 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
17472 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
17473 TYPE_TYPELESS_STORAGE.
17474 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
17476 2017-04-12 Jakub Jelinek <jakub@redhat.com>
17479 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
17480 first argument to type.
17482 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17486 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
17487 CONST0_RTX (mode) rather than const0_rtx where appropriate.
17488 (rs6000_expand_binop_builtin): Likewise.
17489 (rs6000_expand_ternop_builtin): Likewise; also add missing
17490 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
17491 vshasigma built-ins.
17492 * doc/extend.texi: Document that vec_xxpermdi's third argument
17493 must be a constant.
17495 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
17497 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
17498 Use shift_const cost parameter when calculating gain of STV shifts.
17500 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
17502 PR rtl-optimization/70478
17503 * lra-constraints.c (process_alt_operands): Check memory for
17504 disfavoring memory insn operand.
17506 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17508 PR middle-end/80100
17509 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
17510 left shift in unsigned HOST_WIDE_INT type.
17512 PR rtl-optimization/80385
17513 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
17514 (not (neg X)) into (plus X -1) for complex or non-integral modes.
17517 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
17518 if they have any depend clauses.
17520 2017-04-11 Martin Liska <mliska@suse.cz>
17523 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
17524 * ipa-split.c (split_function): Create a local comdat symbol
17525 if caller is in a comdat group.
17527 2017-04-11 Martin Liska <mliska@suse.cz>
17530 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
17533 2017-04-11 Martin Sebor <msebor@redhat.com>
17535 PR middle-end/80364
17536 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
17537 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
17539 (directive::set_width, directive::set_precision, format_character):
17541 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
17544 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
17547 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
17548 conflict, set target->arch_name instead of target->cpu_name.
17550 2017-04-11 Richard Biener <rguenther@suse.de>
17552 PR tree-optimization/80374
17553 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
17554 build_zero_cst, remove fold_convertible_p check again.
17556 2017-04-11 Martin Liska <mliska@suse.cz>
17559 * ubsan.c (instrument_object_size): Do not instrument register
17562 2017-04-11 Jakub Jelinek <jakub@redhat.com>
17565 * config/i386/i386-builtin-types.def
17566 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
17567 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
17568 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
17569 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
17570 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
17571 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
17572 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
17573 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
17574 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
17575 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
17576 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
17577 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
17578 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
17579 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
17580 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
17581 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
17582 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
17583 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
17584 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
17585 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
17586 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
17587 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
17588 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
17589 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
17590 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
17591 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
17592 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
17593 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
17594 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
17595 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
17596 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
17597 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
17598 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
17599 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
17600 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
17601 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
17602 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
17603 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
17604 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
17605 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
17606 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
17607 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
17608 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
17609 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
17610 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
17611 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
17613 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
17614 flag to second_arg_count, handle 4 argument function type _COUNT
17615 aliases, handle second_arg_count on second argument rather than last.
17617 2017-04-10 Jeff Law <law@redhat.com>
17619 PR tree-optimization/80374
17620 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
17621 record anything if we can not convert integer_zero_node to the
17624 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
17627 * config/rs6000/rs6000.c (rs6000_option_override_internal):
17628 Enhance special handling given to the TARGET_P9_MINMAX option in
17629 relation to certain other options.
17631 2017-04-10 Bin Cheng <bin.cheng@arm.com>
17633 PR tree-optimization/80153
17634 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
17635 remove POINTER_PLUS_EXPR's base part directly, rather than through
17638 2017-04-10 Richard Biener <rguenther@suse.de>
17639 Bin Cheng <bin.cheng@arm.com>
17641 PR tree-optimization/80153
17642 * tree-affine.c (aff_combination_to_tree): Get base pointer from
17643 the first element of pointer type aff_tree. Build result expr in
17645 (add_elt_to_tree): Convert to type unconditionally. Remove other
17646 fold_convert calls.
17647 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
17648 (rewrite_use_nonlinear_expr): Check invariant using iv information.
17650 2017-04-10 Richard Biener <rguenther@suse.de>
17652 * tree-ssa-structalias.c (find_func_aliases): Properly handle
17655 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
17657 PR rtl-optimization/70478
17658 * lra-constraints.c (curr_small_class_check): New.
17659 (update_and_check_small_class_inputs): New.
17660 (process_alt_operands): Update curr_small_class_check. Disfavor
17661 alternative insn memory operands. Check available regs for small
17664 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
17667 * config/mips/mips.opt (-mvirt): Update description.
17668 * doc/invoke.texi (-mvirt): Likewise.
17670 2017-04-10 Richard Biener <rguenther@suse.de>
17672 PR middle-end/80362
17673 * fold-const.c (fold_binary_loc): Look at unstripped ops when
17674 looking for NEGATE_EXPR in -A / -B to A / B folding.
17676 2017-04-10 Martin Liska <mliska@suse.cz>
17678 PR gcov-profile/80224
17679 * gcov.c (print_usage): Fix usage string.
17680 (get_gcov_intermediate_filename): Remove.
17681 (output_gcov_file): Use both for normal and intermediate format.
17682 (generate_results): Do not initialize special file for
17683 intermediate format.
17685 2017-04-10 Richard Biener <rguenther@suse.de>
17687 PR tree-optimization/80304
17688 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
17691 2017-04-10 Nathan Sidwell <nathan@acm.org>
17694 * config/rs6000/rs6000.c (rs6000_vector_type): New.
17695 (rs6000_init_builtins): Use it.
17697 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17699 * config/arm/arm.md (<mrc>): Add mode to SET source.
17700 (<mrrc>): Likewise.
17702 2017-04-10 Richard Biener <rguenther@suse.de>
17704 PR middle-end/80344
17705 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
17707 2017-04-10 Jakub Jelinek <jakub@redhat.com>
17710 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
17711 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
17712 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
17713 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
17714 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
17715 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
17716 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
17717 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
17718 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
17719 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
17720 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
17721 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
17722 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
17723 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
17724 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
17725 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
17726 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
17727 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
17728 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
17729 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
17730 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
17731 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
17732 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
17734 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
17736 PR rtl-optimization/70478
17737 * lra-constraints.c: Reverse the last patch.
17739 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
17741 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
17742 Add comment for WCHAR_T.
17744 2017-04-08 Martin Liska <mliska@suse.cz>
17747 2017-04-07 Martin Liska <mliska@suse.cz>
17750 * ipa-split.c (split_function): Add function part to a same comdat
17753 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17756 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
17758 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
17760 * rs6000/rs6000.c (vec_load_pendulum): Rename...
17761 (vec_pairing): ...to this.
17762 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
17763 (rs6000_sched_init): Adjust for name change.
17764 (struct rs6000_sched_context): Likewise.
17765 (rs6000_init_sched_context): Likewise.
17766 (rs6000_set_sched_context): Likewise.
17768 2017-04-07 Jakub Jelinek <jakub@redhat.com>
17774 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
17776 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
17777 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
17778 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
17780 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
17782 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
17784 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
17786 PR rtl-optimization/70703
17787 * ira-color.c (update_conflict_hard_regno_costs): Use
17788 int64_t instead of HOST_WIDE_INT.
17790 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
17792 PR rtl-optimization/70478
17793 * lra-constraints.c (process_alt_operands): Disfavor alternative
17794 insn memory operands.
17796 2017-04-07 Jeff Law <law@redhat.com>
17798 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
17799 CALL and NOTE_INSN_CALL_ARG_LOCATION.
17801 2017-04-07 Martin Liska <mliska@suse.cz>
17804 * config/aarch64/aarch64.c (aarch64_process_target_attr):
17805 Show error message instead of an ICE.
17807 2017-04-07 Martin Liska <mliska@suse.cz>
17810 * ipa-split.c (split_function): Add function part to a same comdat
17813 2017-04-07 Richard Biener <rguenther@suse.de>
17815 PR middle-end/80341
17816 * tree.c (get_unwidened): Also handle ! for_type case for
17818 * convert.c (do_narrow): Split out from ...
17819 (convert_to_integer_1): ... here. Do not pass final truncation
17820 type to get_unwidened for TRUNC_DIV_EXPR.
17822 2017-04-07 Richard Biener <rguenther@suse.de>
17824 * tree-affine.c (wide_int_ext_for_comb): Take type rather
17826 (aff_combination_const): Adjust.
17827 (aff_combination_scale): Likewise.
17828 (aff_combination_add_elt): Likewise.
17829 (aff_combination_add_cst): Likewise.
17830 (aff_combination_convert): Likewise.
17831 (add_elt_to_tree): Likewise. Remove unused argument.
17832 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
17834 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
17836 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
17838 * config/arm/arm.c (arm_default_short_enums): Use
17839 ARM_DEFAULT_SHORT_ENUMS.
17840 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
17842 2017-04-06 Jakub Jelinek <jakub@redhat.com>
17845 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
17846 members with redundant out-of-class redeclaration.
17848 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
17851 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
17852 * config/i386/i386.md (*zero_extendsidi2):
17853 Add (?*x,*x) and (?*v,*v) alternatives.
17855 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
17858 * config/i386/i386.c (ix86_expand_builtin)
17859 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
17860 mode from insn data. Convert operands to insn operand mode.
17861 Copy operands that don't satisfy insn predicate to a register.
17863 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
17865 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
17868 2017-04-06 Richard Biener <rguenther@suse.de>
17870 PR tree-optimization/80334
17871 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
17872 preserve alignment of accesses.
17874 2017-04-06 Richard Biener <rguenther@suse.de>
17876 PR tree-optimization/80262
17877 * tree-sra.c (build_ref_for_offset): Preserve address-space
17879 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
17880 Drop useless address-space information on MEM_REF offsets.
17882 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
17884 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
17886 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
17888 PR rtl-optimization/70703
17889 * ira-color.c (update_conflict_hard_regno_costs): Use
17890 HOST_WIDE_INT instead of long.
17892 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
17895 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
17896 not defined for x86_64 target. Add -mmmx target option when __SSE2__
17898 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
17899 for x86_64 target. Handle -m3dnowa option.
17901 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
17903 PR rtl-optimization/70703
17904 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
17905 (update_conflict_hard_regno_costs): Use long instead of unsigned
17906 arithmetic for cost calculation.
17908 2017-04-05 Jakub Jelinek <jakub@redhat.com>
17909 Bernd Edlinger <bernd.edlinger@hotmail.de>
17912 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
17915 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
17918 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
17919 ptr_mode with Pmode throughout.
17920 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
17921 into probe_stack_range and use DImode.
17923 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
17926 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
17927 call_eh_return is true.
17929 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17931 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
17932 Initialize last_match_fntype_index.
17934 2017-04-05 Jakub Jelinek <jakub@redhat.com>
17937 * tree-nvr.c: Include internal-fn.h.
17938 (pass_return_slot::execute): Ignore internal calls without
17941 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17942 Richard Biener <rguenther@suse.de>
17945 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
17946 captures used multiple times, except for the last use.
17947 * generic-match-head.c: Include gimplify.h.
17949 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17951 PR tree-optimization/79390
17952 * target.h (struct noce_if_info): Declare.
17953 * targhooks.h (default_noce_conversion_profitable_p): Declare.
17954 * target.def (noce_conversion_profitable_p): New target hook.
17955 * ifcvt.h (struct noce_if_info): New type, moved from ...
17956 * ifcvt.c (struct noce_if_info): ... here.
17957 (noce_conversion_profitable_p): Renamed to ...
17958 (default_noce_conversion_profitable_p): ... this. No longer
17960 (noce_try_store_flag_constants, noce_try_addcc,
17961 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
17962 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
17963 instead of noce_conversion_profitable_p.
17964 * config/i386/i386.c: Include ifcvt.h.
17965 (ix86_option_override_internal): Don't override
17966 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
17967 (ix86_noce_conversion_profitable_p): New function.
17968 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
17969 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
17970 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
17971 * doc/tm.texi: Regenerated.
17973 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17975 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
17978 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17981 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
17982 instructions for small multiply cores.
17984 2017-04-04 Jeff Law <law@redhat.com>
17986 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
17988 (mips_expand_vec_perm_const): Initialize elements in orig_perm
17989 that are not set by the loop over the elements.
17991 2017-04-04 Jakub Jelinek <jakub@redhat.com>
17994 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
17995 int mode, convert_modes it to mode as unsigned, otherwise use
17996 lowpart_subreg to mode rather than SImode.
17997 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
17998 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
17999 Use DImode instead of SImode for the shift count operand.
18000 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
18003 2017-04-04 Richard Biener <rguenther@suse.de>
18005 PR middle-end/80281
18006 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
18007 arithmetic done for the negate or the plus. Simplify.
18008 (A - (-B) -> A + B): Likewise.
18009 * fold-const.c (split_tree): Make sure to not negate pointers.
18011 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
18013 PR rtl-optimization/60818
18014 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
18015 a compare of comparisons with the thing compared if this results
18016 in a different machine mode.
18018 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
18020 * alias.c (base_alias_check): Fix typo in comment.
18021 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
18022 * cgraphunit.c (symbol_table::compile): Likewise.
18023 * collect2.c (maybe_run_lto_and_relink): Likewise.
18024 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
18025 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
18026 * config/avr/avr.c (avr_map_op_t): Likewise.
18027 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
18028 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
18029 * config/epiphany/epiphany.md (movcc): Likewise.
18030 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
18031 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
18033 * config/mips/mips.c (mips_save_restore_reg): Likewise.
18034 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
18035 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
18036 * config/sh/sh.c (sh_rtx_costs): Likewise.
18037 * fold-const.c (fold_truth_andor): Likewise.
18038 * genautomata.c (collapse_flag): Likewise.
18039 * gengtype.h (struct type::u::s): Likewise.
18040 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
18041 * input.c (FORMAT_AMOUNT): Likewise.
18042 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
18043 (known_aggs_to_agg_replacement_list): Likewise.
18044 * ipa-inline-analysis.c: Likewise.
18045 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
18046 * ipa-polymorphic-call.c
18047 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
18048 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
18049 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
18051 * modulo-sched.c (apply_reg_moves): Likewise.
18052 * omp-expand.c (build_omp_regions_1): Likewise.
18053 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
18054 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
18055 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
18056 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
18057 * value-prof.c: Likewise.
18058 * var-tracking.c (val_reset): Likewise.
18060 2017-04-03 Richard Biener <rguenther@suse.de>
18062 PR tree-optimization/80275
18063 * fold-const.c (split_address_to_core_and_offset): Handle
18066 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
18068 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
18069 descriptors is at least equal to that of functions.
18071 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18073 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
18075 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
18078 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
18079 (mov<IMOD4:mode>): New expander.
18080 (*mov<IMOD4:mode>_internal): New insn and split pattern.
18082 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
18084 PR rtl-optimization/79405
18085 * fwprop.c (propagations_left): New variable.
18086 (forward_propagate_into): Decrement it.
18087 (fwprop_init): Initialize it.
18088 (fw_prop): If the variable has reached zero, stop propagating.
18089 (fwprop_addr): Ditto.
18091 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18094 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
18095 a FUNCTION_DECL, pass it as decl instead of origin to
18098 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
18100 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
18103 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
18106 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
18107 TARGET_VSX_SMALL_INTEGER.
18109 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18111 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
18112 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
18114 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
18116 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
18117 extraction from odd-numbered MSA register.
18119 2017-03-31 Jakub Jelinek <jakub@redhat.com>
18121 PR middle-end/80173
18122 * expmed.c (store_bit_field_1): Don't attempt to create
18123 a word subreg out of hard registers wider than word if they
18124 have HARD_REGNO_NREGS of 1 for their mode.
18126 PR middle-end/80163
18127 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
18128 conversions to integer types wider than word and pointer.
18131 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
18132 (rtx_equal_for_cselib_p): Pass 0 to it.
18133 * cselib.c (cselib_hasher::equal): Likewise.
18134 (rtx_equal_for_cselib_1): Add depth argument. If depth
18135 is 128, don't look up VALUE locs and punt. Increment
18136 depth in recursive calls when walking VALUE locs.
18138 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
18140 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
18141 (make_gcov_file_name): Use the canonical path name for generating
18143 (read_line): Fix handling of files with ascii null bytes.
18145 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
18147 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
18148 to initialise a vector register instead
18149 of using a const_int.
18151 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18153 PR translation/80189
18154 * gimplify.c (omp_default_clause): Use %qs instead of %s in
18155 diagnostic messages.
18157 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
18160 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
18161 (dfp_diex_<mode>): Update mode of operand 1.
18162 * doc/extend.texi (dxex, dxexq): Document change to return type.
18163 (diex, diexq): Document change to argument type.
18165 2017-03-30 Martin Jambor <mjambor@suse.cz>
18168 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
18169 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
18170 it reflects the signature changes performed at the callee side.
18171 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
18172 to cgraph_build_function_type_skip_args.
18173 (build_function_decl_skip_args): Adjust call to the above function.
18175 2017-03-30 Jakub Jelinek <jakub@redhat.com>
18178 * config/i386/sse.md
18179 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
18180 register as dest whenever it is a MEM not rtx_equal_p to the
18181 corresponding dup operand, and when forcing into reg move the
18182 reg into the memory afterwards.
18183 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
18184 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
18185 for the force_reg mode.
18186 (avx512vl_vextractf128<mode>): Use register as dest either
18187 always when a MEM, or when it is a MEM not rtx_equal_p to the
18188 corresponding dup operand, or even not when it is a CONST_VECTOR
18189 depending on the mode and lo vs. hi.
18190 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
18192 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
18193 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
18194 Likewise. Require that operands[2] is even.
18195 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
18196 Remove extraneous parens. Require that operands[2] is a multiple
18198 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
18199 operands[0] is a MEM if <mask_applied>, the predicates/constraints
18200 disallow memory then.
18202 2017-03-30 Richard Biener <rguenther@suse.de>
18204 PR tree-optimization/77498
18205 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
18206 to non-constants over backedges.
18208 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
18210 PR rtl-optimization/80233
18211 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
18212 as last_combined_insn. Do not test for BARRIER_P separately.
18214 2017-03-29 Andreas Schwab <schwab@suse.de>
18217 * calls.c (prepare_call_address): Convert funexp to Pmode before
18218 copying to temp reg.
18220 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18222 PR tree-optimization/80158
18223 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
18224 Handle possible future case of more than one alternate
18226 (replace_rhs_if_not_dup): Likewise.
18227 (replace_one_candidate): Likewise.
18229 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
18231 PR rtl-optimization/80193
18232 * ira.c (ira): Do not check allocation for LRA.
18234 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
18236 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
18237 (nvptx_output_simt_exit): Declare.
18238 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
18239 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
18240 (init_softstack_frame): Move initialization of crtl->is_leaf to...
18241 (nvptx_declare_function_name): ...here. Emit declaration of local
18242 memory space buffer for omp_simt_enter insn.
18243 (nvptx_output_unisimt_switch): New.
18244 (nvptx_output_softstack_switch): New.
18245 (nvptx_output_simt_enter): New.
18246 (nvptx_output_simt_exit): New.
18247 * config/nvptx/nvptx.h (struct machine_function): New fields
18248 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
18249 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
18250 (UNSPECV_SIMT_EXIT): Ditto.
18251 (omp_simt_enter_insn): New insn.
18252 (omp_simt_enter): New expansion.
18253 (omp_simt_exit): New insn.
18254 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
18256 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
18257 (expand_GOMP_SIMT_ENTER_ALLOC): New.
18258 (expand_GOMP_SIMT_EXIT): New.
18259 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
18260 (GOMP_SIMT_ENTER_ALLOC): Ditto.
18261 (GOMP_SIMT_EXIT): Ditto.
18262 * target-insns.def (omp_simt_enter): New insn.
18263 (omp_simt_exit): Ditto.
18264 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
18266 (lower_rec_simd_input_clauses): Implement SIMT privatization.
18267 (lower_rec_input_clauses): Likewise.
18268 (lower_lastprivate_clauses): Handle SIMT privatization.
18270 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
18271 (ompdevlow_adjust_simt_enter): New.
18272 (find_simtpriv_var_op): New.
18273 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
18274 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
18276 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
18277 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
18278 (copy_decl_for_dup_finish): Ditto.
18280 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
18282 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
18285 * config/i386/i386.c (ix86_option_override_internal): Always
18286 allow -mpreferred-stack-boundary=3 for 64-bit targets.
18288 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18290 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
18292 2017-03-28 Bin Cheng <bin.cheng@arm.com>
18294 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
18295 mark new edge's irreducible flag accordign to it.
18296 (vect_do_peeling): Check loop preheader edge's irreducible flag
18297 and pass it to function slpeel_add_loop_guard.
18299 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
18301 PR tree-optimization/80218
18302 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
18303 Update block frequencies and counts.
18305 2017-03-28 Richard Biener <rguenther@suse.de>
18307 PR tree-optimization/78644
18308 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
18309 of a simplification result we may not use it at all.
18311 2017-03-28 Richard Biener <rguenther@suse.de>
18314 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
18315 without arguments, generate default definition of a SSA name.
18317 2017-03-28 Richard Biener <rguenther@suse.de>
18319 PR middle-end/80222
18320 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
18321 TYPE_REF_CAN_ALIAS_ALL references.
18322 * fold-const.c (fold_indirect_ref_1): Likewise.
18324 2017-03-28 Martin Liska <mliska@suse.cz>
18327 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
18328 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
18330 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18331 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
18333 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
18334 (EXTRA_SPECS): Define.
18335 (SUBTARGET_EXTRA_SPECS): Likewise.
18336 (SUBTARGET_CPP_SPEC): Likewise.
18337 * config/arc/elf.h (EXTRA_SPECS): Renamed to
18338 SUBTARGET_EXTRA_SPECS.
18339 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
18341 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
18343 * config/arc/simdext.md (vst64_insn): Update pattern.
18344 (vld32wh_insn): Likewise.
18345 (vld32wl_insn): Likewise.
18346 (vld64_insn): Likewise.
18347 (vld32_insn): Likewise.
18349 2017-03-28 Marek Polacek <polacek@redhat.com>
18352 * fold-const.c (fold_comparison): Use protected_set_expr_location
18353 instead of SET_EXPR_LOCATION.
18355 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
18357 * tree.c (add_expr): Avoid name lookup warning.
18359 2017-03-27 Jeff Law <law@redhat.com>
18361 PR tree-optimization/80216
18362 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
18363 function name. Limit recursion depth.
18364 (record_temporary_equivalences): Corresponding changes.
18366 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
18368 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
18371 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18374 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
18376 * cfgcleanup.c (reg_note_cfa_p): New array.
18377 (insns_have_identical_cfa_notes): New function.
18378 (old_insns_match_p): Don't cross-jump in between /f
18379 and non-/f instructions. If both i1 and i2 are frame related,
18380 verify all CFA notes, their order and content.
18382 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
18385 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
18386 HImode and SImode with zero extend to DImode to one insn.
18387 (bswap<mode>2_extenddi): Likewise.
18388 (bswapsi2_extenddi): Likewise.
18389 (bswaphi2_extendsi): Likewise.
18390 (bswaphi2): Combine bswap HImode and SImode into one insn.
18391 Separate memory insns from swapping register.
18392 (bswapsi2): Likewise.
18393 (bswap<mode>2): Likewise.
18394 (bswaphi2_internal): Delete, no longer used.
18395 (bswapsi2_internal): Likewise.
18396 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
18397 store, and gpr<-gpr swap insns.
18398 (bswap<mode>2_store): Likewise.
18399 (bswaphi2_reg): Register only splitter, combine with the splitter.
18400 (bswaphi2 splitter): Likewise.
18401 (bswapsi2_reg): Likewise.
18402 (bswapsi2 splitter): Likewise.
18403 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
18404 the insns into load, store, and register/register insns.
18405 (bswapdi2_ldbrx): Likewise.
18406 (bswapdi2_load): Likewise.
18407 (bswapdi2_store): Likewise.
18408 (bswapdi2_reg): Likewise.
18410 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
18412 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
18413 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
18415 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
18418 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
18420 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18421 special handling for target option conflicts between dform
18422 options (-mpower9-dform, -mpower9-dform-vector,
18423 -mpower9-dform-scalar) and -mno-direct-move.
18425 2017-03-27 Richard Biener <rguenther@suse.de>
18427 PR tree-optimization/80181
18428 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
18430 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18432 * config/arc/predicates.md (move_double_src_operand): Replace the
18433 call to move_double_src_operand with a call to address_operand.
18435 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18437 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
18438 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
18439 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
18441 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18443 * config/arc/predicates.md (long_immediate_loadstore_operand):
18444 Consider scaled addresses cases.
18446 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
18448 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
18449 restored when in interrupt.
18450 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
18451 doesn't have delay slot.
18453 2017-03-27 Richard Biener <rguenther@suse.de>
18456 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
18457 inlined thunk clones.
18459 2017-03-27 Jakub Jelinek <jakub@redhat.com>
18462 * asan.c (instrument_derefs): Copy over last operand from
18463 original COMPONENT_REF to the new COMPONENT_REF with
18464 DECL_BIT_FIELD_REPRESENTATIVE.
18465 * ubsan.c (instrument_object_size): Likewise.
18467 2017-03-27 Richard Biener <rguenther@suse.de>
18469 PR tree-optimization/80170
18470 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
18471 sure DR/SCEV didnt fold in constants we do not see when looking
18472 at the reference base alignment.
18474 2017-03-27 Richard Biener <rguenther@suse.de>
18476 PR middle-end/80171
18477 * gimple-fold.c (fold_ctor_reference): Properly guard against
18478 NULL return value from canonicalize_constructor_val.
18480 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
18483 * config/i386/i386.c (ix86_expand_builtin)
18484 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
18485 flags reg setting and flags reg using instructions.
18486 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
18487 clobbering instructions to zero extend op2.
18489 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
18491 * doc/install.texi (Configuration) <--with-aix-soname>:
18492 Update link to AIX ld.
18494 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
18496 PR rtl-optimization/80160
18497 PR rtl-optimization/80159
18498 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
18499 reg_alternate_class into account.
18501 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
18504 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
18505 to consider in curr_insn_transform.
18507 2017-03-24 Jakub Jelinek <jakub@redhat.com>
18509 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
18510 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
18511 and emit_mode_inner.
18513 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18515 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
18516 argument to the overloaded builtin variants. Use the new flag to
18517 deprecate certain builtin variants.
18518 * config/s390/s390-builtin-types.def: Add new builtin types.
18519 * config/s390/s390-builtins.h: Support new flags field for
18520 overloaded builtins.
18521 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
18522 (s390_macro_to_expand): Enable vector float data type.
18523 (s390_cpu_cpp_builtins_internal): Indicate support of the new
18524 builtins by incrementing the __VEC__ version number.
18525 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
18527 (s390_resolve_overloaded_builtin): Emit error messages depending
18528 on the builtin flags.
18529 * config/s390/s390.c (s390_expand_builtin): Support additional
18530 flags argument. Change error message to match the messages
18531 emitted in s390-c.c.
18532 * config/s390/s390.md: New UNSPEC_* constants.
18533 (op_type): Add new instruction types.
18534 * config/s390/vecintrin.h: Add new builtins and test data class
18536 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
18537 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
18538 (VEC_INEXACT, VEC_NOINEXACT): New constants.
18539 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
18540 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
18541 ("vec_mergel<mode>"): V_HW -> VEC_HW.
18543 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
18544 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
18545 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
18546 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
18548 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
18549 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
18550 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
18551 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
18553 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
18554 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
18555 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
18556 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
18557 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
18558 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
18559 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
18561 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
18562 ("vec_scatter_element<V_HW_4:mode>_DI")
18563 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
18564 ("vec_fpint<mode>", "vflls")
18565 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
18566 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
18567 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
18568 ("*vec_cmphe<mode>_cc"): ... these.
18570 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
18571 mode constant instead of magic value.
18573 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18575 * config/s390/s390.c (s390_expand_vec_compare): Support other
18576 vector floating point modes than just V2DF.
18577 (s390_expand_vcond): Likewise.
18578 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
18579 (s390_cannot_change_mode_class): Prevent mode changes between TF
18580 and V1TF in vector registers.
18581 * config/s390/s390.md (DF, SF): New mode attributes.
18582 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
18583 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
18584 SFmode support for VRs.
18585 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
18587 (VFT, VF_HW): New mode iterators.
18588 (vw, sdx): New mode attributes.
18589 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
18590 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
18591 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
18592 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
18593 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
18594 also the new vector floating point modes. Renaming to ...
18596 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
18597 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
18598 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
18599 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
18600 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
18601 ("vec_unordered<mode>"): ... these.
18603 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
18604 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
18605 ("*vec_extendv2df"): New insn definitions.
18607 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18609 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
18610 ("mulditi3_2", "*muldi3_sign"): New patterns.
18611 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
18612 rename the pattern definition.
18614 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18616 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
18618 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
18620 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18622 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
18623 instruction if possible.
18624 * config/s390/vector.md (vec_halfnumelts): New mode
18626 ("*vec_vllezlf<mode>"): New pattern.
18628 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18630 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
18631 ("popcountv4si2", "popcountv2di2"): Rename to ...
18632 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
18633 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
18635 ("popcount<mode>2_vxe"): New pattern.
18637 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18639 * common/config/s390/s390-common.c (processor_flags_table): Add
18641 * config.gcc: Add arch12.
18642 * config/s390/driver-native.c (s390_host_detect_local_cpu):
18643 Default to arch12 for unknown CPU model numbers.
18644 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
18645 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
18646 PROCESSOR_max sanity check.
18647 * config/s390/s390-opts.h (enum processor_type): Add
18649 * config/s390/s390.c (processor_table): Add arch12.
18650 (s390_expand_builtin): Add check for B_VXE flag.
18651 (s390_issue_rate): Add PROCESSOR_ARCH12.
18652 (s390_get_sched_attrmask): Likewise.
18653 (s390_get_unit_mask): Likewise.
18654 (s390_sched_score): Enable z13 scheduling for arch12.
18655 (s390_sched_reorder): Likewise.
18656 (s390_sched_variable_issue): Likewise.
18657 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
18659 (s390_tune_attr): Use z13 scheduling also for arch12.
18660 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
18661 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
18662 (TARGET_VXE_P): New macros.
18663 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
18664 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
18665 * config/s390/s390.opt: Add arch12 as processor_type.
18667 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18669 * config/s390/s390.md
18670 ("fixuns_truncdddi2", "fixuns_trunctddi2")
18671 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
18672 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
18674 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
18675 Rename expanders to ...
18677 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
18678 ("fixuns_truncdddi2_emu"): ... these.
18680 ("fixuns_trunc<mode>si2_emu"): New expander.
18682 ("*fixuns_truncdfdi2_z13"): Rename to ...
18683 ("*fixuns_truncdfdi2_vx"): ... this.
18685 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18687 * config/s390/2964.md: Remove the single element vector compare
18688 instructions which are no longer used.
18689 * config/s390/s390.c (s390_select_ccmode): Remove handling of
18691 (s390_canonicalize_comparison): Remove handling of DFmode
18693 (s390_expand_vec_compare_scalar): Remove function.
18694 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
18695 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
18697 ("*cmp<mode>_ccs"): Add wfcdb instruction.
18699 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18701 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
18703 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
18704 will anyway by matched by mov<mode>_64dfp.
18706 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18708 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
18709 vlef/vstef. Add missing operand to vleif.
18711 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18713 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
18714 pair for all vector types with 64 bit elements.
18715 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
18716 * config/s390/vector.md (V_HW_64): ... here.
18717 (V_128_NOSINGLE): New mode iterator.
18718 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
18719 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
18720 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
18721 ("*vec_load_pairv2di"): Change to ...
18722 ("*vec_load_pair<mode>"): ... this one.
18724 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18726 * config/s390/constraints.md: Add comments.
18727 (jKK): Reject element sizes > 8 bytes.
18728 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
18730 * config/s390/s390.md: Add the s_operand checks formerly in
18731 s390_split_ok_p to various splitters where they are still
18733 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
18734 for 128 bit vectors. Plus two splitters.
18736 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18738 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
18741 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18744 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
18745 error if the boundary argument is not constant.
18747 2017-03-24 Jakub Jelinek <jakub@redhat.com>
18749 PR rtl-optimization/80112
18750 * loop-doloop.c (doloop_condition_get): Don't check condition
18751 if cmp isn't SET with IF_THEN_ELSE src.
18753 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18755 PR tree-optimization/80158
18756 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
18757 replacing a candidate statement, also replace it for the
18758 candidate's alternate interpretation.
18759 (replace_rhs_if_not_dup): Likewise.
18760 (replace_one_candidate): Likewise.
18762 2017-03-24 Richard Biener <rguenther@suse.de>
18764 PR tree-optimization/80167
18765 * graphite-isl-ast-to-gimple.c
18766 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
18768 (translate_isl_ast_to_gimple::get_rename): Likewise.
18770 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
18772 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
18773 handling of certain combinations of target options, including the
18774 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
18775 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
18777 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18780 * config/arm/arm.md (*load_multiple): Add reload_completed to
18781 matching condition.
18783 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18784 Richard Biener <rguenth@suse.de>
18786 PR tree-optimization/79908
18787 PR tree-optimization/80136
18788 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
18789 been cast away, gimplify_and_add suffices.
18791 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
18793 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
18795 2017-03-23 Richard Biener <rguenther@suse.de>
18797 PR tree-optimization/80032
18798 * gimplify.c (gimple_push_cleanup): Forced unconditional
18799 cleanups still have to go to the conditional_cleanups
18802 2017-03-22 Jakub Jelinek <jakub@redhat.com>
18804 PR tree-optimization/80072
18805 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
18807 (next_operand_entry_id): Change type to unsigned int.
18808 (sort_by_operand_rank): Make sure to return the right return value
18809 even if unsigned fields are bigger than INT_MAX.
18810 (struct oecount): Change cnt and id type to unsigned int.
18811 (oecount_hasher::equal): Formatting fix.
18812 (oecount_cmp): Make sure to return the right return value
18813 even if unsigned fields are bigger than INT_MAX.
18814 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
18817 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
18818 TREE_READONLY on result if writing it more than once.
18821 * doc/invoke.texi (-fsanitize=thread): Document that with
18822 -fnon-call-exceptions atomics are not able to throw
18826 * tsan.c: Include tree-eh.h.
18827 (instrument_builtin_call): Call maybe_clean_eh_stmt or
18828 maybe_clean_or_replace_eh_stmt where needed.
18829 (instrument_memory_accesses): Add cfg_changed argument.
18830 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
18831 if it returned true.
18832 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
18834 PR rtl-optimization/63191
18835 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
18836 wrapper function, moved the whole old content into ...
18837 (ix86_delegitimize_address_1): ... this. New inline function.
18838 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
18839 true as last argument instead of ix86_delegitimize_address.
18841 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
18843 * config/aarch64/aarch64.c (generic_branch_cost): Copy
18844 cortexa57_branch_cost.
18846 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
18848 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
18850 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18853 * doc/md.texi (Constraints): Document wA constraint.
18854 * config/rs6000/constraints.md (wA): New.
18855 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
18856 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
18857 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
18858 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
18860 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
18863 * gimplify.c (is_oacc_declared): New function.
18864 (oacc_default_clause): Use it to set default flags for acc declared
18865 variables inside parallel regions.
18866 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
18867 declared variables.
18868 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
18869 declare attribute to any decl as necessary.
18871 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
18874 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
18875 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
18876 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
18877 (arm_arch_lpae): This.
18878 * config/arm/arm.c (arm_arch7ve): Rename into ...
18879 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
18880 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
18883 2017-03-22 Martin Liska <mliska@suse.cz>
18886 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
18887 error message instead of an ICE.
18889 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18891 * doc/extend.texi (6.11 Additional Floating Types): Revise.
18893 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
18895 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
18897 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
18900 2017-03-21 Martin Sebor <msebor@redhat.com>
18902 * doc/extend.texi: Use "cannot" instead of "can't."
18903 * doc/hostconfig.texi: Same.
18904 * doc/install.texi: Same.
18905 * doc/invoke.texi: Same.
18906 * doc/loop.texi: Same.
18907 * doc/md.texi: Same.
18908 * doc/objc.texi: Same.
18909 * doc/rtl.texi: Same.
18910 * doc/tm.texi: Same.
18911 * doc/tm.texi.in: Same.
18912 * doc/trouble.texi: Same.
18914 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
18917 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
18918 (collect_checksum_attributes): Set it.
18919 (die_checksum_ordered): Use it.
18921 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18923 PR tree-optimization/79908
18924 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
18925 change: For a VA_ARG whose LHS has been cast away, use
18926 force_gimple_operand to construct the side effects.
18928 2017-03-21 David Malcolm <dmalcolm@redhat.com>
18930 PR translation/80001
18931 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
18932 more amenable to translation.
18933 (oacc_loop_auto_partitions): Likewise.
18935 2017-03-21 Marek Polacek <polacek@redhat.com>
18936 Martin Sebor <msebor@redhat.com>
18938 PR tree-optimization/80109
18939 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
18940 on INTEGRAL_TYPE_P.
18942 2017-03-21 Jakub Jelinek <jakub@redhat.com>
18943 Segher Boessenkool <segher@kernel.crashing.org>
18946 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
18947 check reg_used_between_p between insn and one of succ or succ2
18948 depending on if succ is artificial insn not inserted into insn
18951 2017-03-21 Martin Liska <mliska@suse.cz>
18953 PR gcov-profile/80081
18954 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
18955 * doc/gcc.texi: Include gcov-dump stuff.
18956 * doc/gcov-dump.texi: New file.
18958 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
18960 PR rtl-optimization/79150
18961 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
18962 conditional jump, if the jump is the last insn of the loop.
18964 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18965 Richard Biener <rguenth@suse.de>
18967 PR tree-optimization/79908
18968 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
18969 been cast away, use force_gimple_operand to construct the side
18972 2017-03-21 Martin Liska <mliska@suse.cz>
18974 PR libfortran/79956
18975 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
18978 2017-03-21 Brad Spengler <spender@grsecurity.net>
18981 * plugin.c (htab_hash_plugin): New function.
18982 (add_new_plugin): Use it and adjust.
18983 (parse_plugin_arg_opt): Adjust.
18984 (init_one_plugin): Likewise.
18986 2017-03-21 Richard Biener <rguenther@suse.de>
18988 PR tree-optimization/80032
18989 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
18990 if set force the cleanup to happen unconditionally.
18991 (gimplify_target_expr): Push inserted clobbers with force_uncond
18992 to avoid them being removed by control-dependent DCE.
18994 2017-03-21 Richard Biener <rguenther@suse.de>
18996 PR tree-optimization/80122
18997 * tree-inline.c (copy_bb): Do not expans va-arg packs or
18998 va_arg_pack_len when the inlined call stmt requires pack
19000 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
19002 2017-03-21 Jakub Jelinek <jakub@redhat.com>
19005 * tsan.c (instrument_builtin_call): If the memory model argument
19006 is not a constant, assume it is valid.
19009 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
19012 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19014 PR rtl-optimization/79910
19015 * combine.c (can_combine_p): Do not allow combining an I0 or I1
19016 if its dest is used by an insn before I2 (other than the combined
19017 insns themselves, which are properly handled already).
19019 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
19022 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19024 * combine.c (record_used_regs): New static function.
19025 (try_combine): Handle situations where there is an additional
19026 instruction between I2 and I3 which needs to have a LOG_LINK
19030 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19032 * combine.c (try_combine): Delete redundant i1 test. Call
19033 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19035 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19038 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
19039 alternatives 13/14.
19041 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19043 PR tree-optimization/80054
19044 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
19045 the optimization if a PHI or any of its arguments is not dominated
19046 by the candidate's basis. Use gphi* rather than gimple* as
19048 (replace_profitable_candidates): Clean up a gimple* variable that
19049 should be a gphi* variable.
19051 2017-03-20 Martin Sebor <msebor@redhat.com>
19054 * doc/extend.texi (attribute constructor): Document present limitation.
19056 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
19059 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
19060 __POWER9_VECTOR__ #ifdef control, change template definition to
19061 use Power9-specific built-in function.
19062 (vec_any_eq): Likewise.
19063 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
19064 to control outcomes from this test.
19065 (vector_ae_<mode>p): For VEC_F modes, likewise.
19067 2017-03-20 Ian Lance Taylor <iant@google.com>
19069 * config/i386/i386.c (ix86_function_regparm): Save an extra
19070 register for -fsplit-stack with DECL_STATIC_CHAIN.
19072 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19075 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
19076 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
19078 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
19080 * config/riscv/riscv.c (riscv_print_operand): Use "fence
19082 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
19085 2017-03-20 Marek Polacek <polacek@redhat.com>
19088 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
19090 2017-03-20 Richard Biener <rguenther@suse.de>
19092 PR tree-optimization/80113
19093 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
19094 allocate extra SSA name for PHI def.
19095 (add_close_phis_to_outer_loops): Likewise.
19096 (add_close_phis_to_merge_points): Likewise.
19097 (copy_loop_close_phi_args): Likewise.
19098 (copy_cond_phi_nodes): Likewise.
19100 2017-03-20 Martin Liska <mliska@suse.cz>
19102 PR middle-end/79753
19103 * tree-chkp.c (chkp_build_returned_bound): Do not build
19104 returned bounds for a LHS that's not a BOUNDED_P type.
19106 2017-03-20 Martin Liska <mliska@suse.cz>
19110 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
19111 COMPLEX_CST and VECTOR_CST.
19113 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19116 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
19117 target operand. A new splitter adds the clobber statement in case
19118 the target operand is dead anyway.
19120 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
19122 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
19123 to age-old versions of binutils and glibc.
19125 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
19127 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
19129 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19131 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
19133 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
19135 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
19136 requirement for binutils 2.13.
19138 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
19140 * combine.c (try_combine): Delete redundant i1 test. Call
19141 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
19143 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
19145 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
19146 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
19148 <riscv64-*-elf>: Re-arrange section
19149 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
19150 <riscv32-*-linux>: Likewise.
19151 <riscv64-*-elf>: Likewise
19152 <riscv64-*-linux>: Likewise.
19154 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
19157 * aarch64.opt(verbose-cost-dump): Fix typo.
19159 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
19162 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
19163 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
19165 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
19167 * reload.c (find_reloads): When reloading a nonoffsettable address,
19168 use RELOAD_OTHER for it and its address reloads.
19170 PR rtl-optimization/79910
19171 * combine.c (record_used_regs): New static function.
19172 (try_combine): Handle situations where there is an additional
19173 instruction between I2 and I3 which needs to have a LOG_LINK
19176 2017-03-17 Jeff Law <law@redhat.com>
19178 PR tree-optimization/71437
19179 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
19180 conditional in the hash table first.
19181 (vrp_dom_walker::before_dom_children): Extract condition from
19182 ASSERT_EXPR. Record condition, its inverion and any implied
19183 conditions as well.
19185 2017-03-17 Marek Polacek <polacek@redhat.com>
19186 Markus Trippelsdorf <markus@trippelsdorf.de>
19188 PR tree-optimization/80079
19189 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
19192 2017-03-17 Richard Biener <rguenther@suse.de>
19194 PR middle-end/80075
19195 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
19196 Properly verify the LHS before the RHS possibly claims to be
19198 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
19201 2017-03-17 Martin Jambor <mjambor@suse.cz>
19203 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
19204 (List of -O2 options): Likewise.
19205 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
19208 2017-03-17 Tom de Vries <tom@codesourcery.com>
19210 * gcov-dump.c (print_usage): Print bug_report_url.
19212 2017-03-17 Richard Biener <rguenther@suse.de>
19214 PR middle-end/80050
19215 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
19216 (parser::peek): Likewise.
19218 2017-03-17 Richard Biener <rguenther@suse.de>
19220 PR tree-optimization/80048
19221 * sese.c (free_sese_info): Properly release rename_map and
19222 copied_bb_map elements.
19224 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
19226 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
19227 Add linked-list forward and backlinks. Insert on
19228 construction, remove on destruction.
19229 (class pass_store_merging): Add m_stores_head field.
19230 (pass_store_merging::terminate_and_process_all_chains):
19231 Iterate over m_stores_head list.
19232 (pass_store_merging::terminate_all_aliasing_chains):
19234 (pass_store_merging::execute): Check for debug stmts first.
19235 Push new chains onto the m_stores_head stack.
19237 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
19240 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
19241 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
19242 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
19244 2017-03-16 Jeff Law <law@redhat.com>
19246 PR tree-optimization/71437
19247 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
19248 member function. Implementation moved into after_dom_children
19249 member function and into the threader's thread_outgoing_edges
19251 (dom_opt_dom_walker::after_dom_children): Simplify by moving
19252 some code into new thread_outgoing_edges.
19253 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
19254 definition. Simplify marker handling (do it here). Assume we always
19255 have the available expression and the const/copies tables.
19256 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
19258 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
19259 * tree-vrp.c (equiv_stack): No longer file scoped.
19260 (vrp_dom_walker): New class.
19261 (vrp_dom_walker::before_dom_children): New member function.
19262 (vrp_dom_walker::after_dom_children): Likewise.
19263 (identify_jump_threads): Setup domwalker. Use it rather than
19264 walking edges in a random order by hand. Simplify setup/finalization.
19265 (finalize_jump_threads): Remove.
19266 (vrp_finalize): Do not call identify_jump_threads here.
19267 (execute_vrp): Do it here instead and call thread_through_all_blocks
19270 PR tree-optimization/71437
19271 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
19273 (simplify_stmt_for_jump_threading): Add basic_block argument. All
19275 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
19276 (dom_opt_dom_walker::thread_across_edge): Remove
19277 handle_dominating_asserts argument. All callers changed.
19278 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
19279 changes. Remove calls to lhs_of_dominating_assert. Other
19280 uses of handle_dominating_asserts turn into unconditional code
19281 (simplify_control_stmt_condition_1): Likewise.
19282 (simplify_control_stmt_condition): Likewise.
19283 (thread_through_normal_block, thread_across_edge): Likewise.
19284 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
19285 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
19286 object if it is not an SSA_NAME.
19287 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
19288 before calling into the VRP specific simplifiers.
19289 (identify_jump_threads): Remove handle_dominating_asserts
19292 2017-03-16 Jakub Jelinek <jakub@redhat.com>
19295 * tree-diagnostic.c (default_tree_printer): No longer static.
19296 * tree-diagnostic.h (default_tree_printer): New prototype.
19298 2017-03-16 Tamar Christina <tamar.christina@arm.com>
19300 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
19301 Change ins into fmov.
19303 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19305 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
19306 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
19307 Use h_con constraint for operand 1.
19308 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
19309 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
19311 2017-03-15 Jeff Law <law@redhat.com>
19313 PR tree-optimization/71437
19314 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
19315 (record_temporary_equivalences): Use it.
19317 PR tree-optimization/71437
19318 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
19319 tree-ssa-scopedtables.
19320 (lookup_avail_expr, build_and_record_new_cond): Likewise.
19321 (record_conditions, record_cond, vuse_eq): Likewise.
19322 (record_edge_info): Adjust to API tweak of record_conditions.
19323 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
19324 (record_temporary_equivalences, optimize_stmt): Likewise.
19325 (eliminate_redundant_computations): Likewise.
19326 (record_equivalences_from_stmt): Likewise.
19327 * tree-ssa-scopedtables.c: Include options.h and params.h.
19328 (vuse_eq): New function, moved from tree-ssa-dom.c
19329 (build_and_record_new_cond): Likewise.
19330 (record_conditions): Likewise. Accept vector of conditions rather
19331 than edge_equivalence structure for first argument.
19332 for the first argument.
19333 (avail_exprs_stack::lookup_avail_expr): New member function, moved
19334 from tree-ssa-dom.c.
19335 (avail_exprs_stack::record_cond): Likewise.
19336 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
19337 from tree-ssa-dom.c.
19338 (avail_exprs_stack): Add new member functions lookup_avail_expr
19340 (record_conditions): Declare.
19342 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
19345 * lra-constraints.c (process_alt_operands): Increase reject for
19346 reloading an input/output operand.
19348 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
19351 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
19352 insns to convert from signed/unsigned char/short to IEEE 128-bit
19354 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
19356 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
19359 * config/i386/i386.c (ix86_vector_duplicate_value): Create
19360 subreg of inner mode for values already in registers.
19362 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
19364 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
19365 iteration reg is used after the loop.
19367 2017-03-14 Martin Sebor <msebor@redhat.com>
19369 PR tree-optimization/79800
19370 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
19371 precision in negative-positive range.
19372 (format_floating): Call non-const overload with adjusted precision.
19374 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
19377 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
19380 2017-03-14 Martin Sebor <msebor@redhat.com>
19382 PR middle-end/80020
19383 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
19384 * builtins.def (aligned_alloc): Use it.
19387 * Makefile.in (GTFILES): Add calls.c.
19388 * calls.c: Include "gt-calls.h".
19390 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
19392 PR rtl-optimization/79728
19393 * regs.h (struct target_regs): New field
19394 x_contains_allocatable_regs_of_mode.
19395 (contains_allocatable_regs_of_mode): New macro.
19396 * reginfo.c (init_reg_sets_1): Initialize it, and change
19397 contains_reg_of_mode so it includes global regs as well.
19398 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
19399 rather than contains_regs_of_mode.
19401 2017-03-14 Martin Liska <mliska@suse.cz>
19403 * doc/invoke.texi: Document options that can't be combined with
19404 -fcheck-pointer-bounds.
19406 2017-03-14 Martin Liska <mliska@suse.cz>
19408 PR middle-end/79831
19409 * doc/invoke.texi (-Wchkp): Document the option.
19411 2017-03-14 Martin Liska <mliska@suse.cz>
19413 * Makefile.in: Install gcov-dump.
19415 2017-03-14 Martin Liska <mliska@suse.cz>
19417 * multiple_target.c (expand_target_clones): Bail out for
19418 an invalid attribute.
19420 2017-03-14 Richard Biener <rguenther@suse.de>
19422 * alias.c (struct alias_set_entry): Pack properly.
19423 * cfgloop.h (struct loop): Likewise.
19424 * cse.c (struct set): Likewise.
19425 * ipa-utils.c (struct searchc_env): Likewise.
19426 * loop-invariant.c (struct invariant): Likewise.
19427 * lra-remat.c (struct cand): Likewise.
19428 * recog.c (struct change_t): Likewise.
19429 * rtl.h (struct address_info): Likewise.
19430 * symbol-summary.h (function_summary): Likewise.
19431 * tree-loop-distribution.c (struct partition): Likewise.
19432 * tree-object-size.c (struct object_size_info): Likewise.
19433 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
19434 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
19435 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
19436 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
19437 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
19438 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
19439 (struct _stmt_vec_info): Likewise.
19441 2017-03-14 Martin Liska <mliska@suse.cz>
19444 * multiple_target.c (create_dispatcher_calls): Check that
19445 a target can create a function dispatcher.
19447 2017-03-14 Martin Liska <mliska@suse.cz>
19450 * multiple_target.c (expand_target_clones): Drop local.local
19451 flag for default implementation.
19453 2017-03-14 Richard Biener <rguenther@suse.de>
19455 PR tree-optimization/80030
19456 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
19458 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
19460 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
19461 gcc_fallthrough() instead of __attribute__((fallthrough));
19463 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19465 * doc/gcc.texi: Remove "up" link to (DIR).
19466 * doc/gccint.texi: Ditto.
19468 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
19470 * doc/install.texi (Specific) <avr>: Remove reference to
19473 2017-03-13 Jeff Law <law@redhat.com>
19475 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
19476 attribute rather than comments.
19478 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
19479 match_scratch operand is highest.
19481 2017-03-13 Martin Liska <mliska@suse.cz>
19483 PR middle-end/78339
19484 * ipa-pure-const.c (warn_function_noreturn): If the declarations
19485 is a CHKP clone, use original declaration.
19487 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19489 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
19490 (arc_conditional_register_usage): Use a different allocation order
19491 when optimizing for size.
19492 * common/config/arc/arc-common.c (arc_option_optimization_table):
19493 Section anchors default on when optimizing for size.
19495 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19497 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
19499 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19501 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
19502 * config/arc/arc.md (cpu_facility): Add cd variant.
19503 (*movqi_insn): Add code density variant.
19504 (*movhi_insn): Likewise.
19505 (*movqi_insn): Likewise.
19506 (*addsi3_mixed): Likewise.
19507 (subsi3_insn): Likewise.
19509 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19511 * config/arc/arc.md (movsi_cond_exec): Update constraint.
19513 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
19515 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
19516 expressions with MINUS and UNARY ops.
19518 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19521 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
19523 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
19524 between vec_select and vector argument.
19525 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
19526 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
19527 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
19528 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
19529 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
19530 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
19532 2017-03-13 Richard Biener <rguenther@suse.de>
19535 * params.def (vect-max-peeling-for-alignment): Fix typo.
19537 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19539 * doc/install.texi (Specific) <mips-*-*>: Remove description of
19540 issue that only occurred with binutils below 2.18.
19542 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19544 * doc/install.texi (Specific) <cris-axis-elf>: No longer
19545 refer to binutils 2.11/2.12 minimum.
19547 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
19549 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
19550 ftp.kernel.org and simplify binutils requirement.
19552 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
19554 * doc/invoke.texi (Warning Options): Fix spelling of link-time
19556 (Optimize Options): Ditto. Also remove redundancy.
19558 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19560 PR translation/79848
19561 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
19563 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
19564 to G_ to avoid double translation.
19566 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19568 PR translation/79923
19569 * auto-profile.c (get_combined_location): Convert leading
19570 character of diagnostics to lower case and remove trailing period.
19571 (read_profile): Likewise for various diagnostics.
19572 * config/arm/arm.c (arm_option_override): Remove trailing period
19573 from various diagnostics.
19574 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
19575 (msp430_expand_delay_cycles): Likewise.
19577 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19580 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
19581 full command-line argument, rather than just "str".
19582 (aarch64_validate_march): Likewise.
19583 (aarch64_validate_mtune): Likewise.
19585 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
19587 PR rtl-optimization/78911
19588 * lra-assigns.c (must_not_spill_p): New function.
19589 (spill_for): Use it.
19591 2017-03-10 Jakub Jelinek <jakub@redhat.com>
19593 PR tree-optimization/79981
19594 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
19595 ATOMIC_COMPARE_EXCHANGE ifn result.
19596 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
19597 IFN_ATOMIC_COMPARE_EXCHANGE.
19599 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19602 * opts.c (parse_sanitizer_options): Add missing question mark to
19603 "did you mean" message.
19605 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19607 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
19609 (VMULEUH_UNS): Likewise.
19610 (VMULOUB_UNS): Likewise.
19611 (VMULOUH_UNS): Likewise.
19612 * config/rs6000/rs6000.c (builtin_function_type): Remove
19613 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
19615 2017-03-10 David Malcolm <dmalcolm@redhat.com>
19618 * read-rtl-function.c (function_reader::read_rtx_operand): Update
19619 x with result of extra_parsing_for_operand_code_0.
19620 (function_reader::extra_parsing_for_operand_code_0): Convert
19621 return type from void to rtx, returning x. When reading
19622 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
19623 larger size containing struct block_symbol.
19625 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
19627 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
19628 -mfloat128-hardware without -m64.
19630 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
19633 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
19634 entries to the case statement that marks unsigned arguments to
19635 overloaded functions.
19637 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
19639 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
19640 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
19642 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
19645 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
19646 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
19648 2017-03-10 Martin Liska <mliska@suse.cz>
19652 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
19653 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
19656 2017-03-10 Olivier Hainque <hainque@adacore.com>
19658 * tree-switch-conversion (array_value_type): Start by resetting
19659 candidate type to it's main variant.
19661 2017-03-10 Jakub Jelinek <jakub@redhat.com>
19663 PR rtl-optimization/79909
19664 * combine.c (try_combine): Use simplify_replace_rtx on individual
19665 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
19666 of the whole CALL_INSN_FUNCTION_USAGE.
19668 PR tree-optimization/79972
19669 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
19670 get_range_info on SSA_NAMEs. Formatting fixes.
19672 2017-03-10 Richard Biener <rguenther@suse.de>
19673 Jakub Jelinek <jakub@redhat.com>
19675 PR tree-optimization/77975
19676 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
19677 edge to be constant.
19678 (get_val_for): For constant x return it. Formatting fix.
19679 (loop_niter_by_eval): Avoid pointless looping if the next iteration
19680 would use the same bases as the current one.
19682 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19684 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
19685 instead of vec_select for V1TImode.
19686 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
19688 (VSX_LE_128): Add V1TI to this mode iterator.
19689 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
19690 (*vsx_le_perm_store_<mode>): Likewise.
19691 (pre-reload splitter for VSX stores): Likewise.
19692 (post-reload splitter for VSX stores): Likewise.
19693 (*vsx_xxpermdi2_le_<mode>): Likewise.
19694 (*vsx_lxvd2x2_le_<mode>): Likewise.
19695 (*vsx_stxvd2x2_le_<mode>): Likewise.
19697 2017-03-09 Michael Eager <eager@eagercon.com>
19699 Correct failures with --enable-checking=yes,rtl.
19701 * config/microblaze/microblaze.c (microblaze_expand_shift):
19702 Replace GET_CODE test with CONST_INT_P and INTVAL test with
19703 test for const0_rtx.
19704 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
19705 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
19707 2017-03-09 Richard Biener <rguenther@suse.de>
19709 PR tree-optimization/79977
19710 * graphite-scop-detection.c (scop_detection::merge_sese):
19711 Handle the case of extra exits to blocks dominating the entry.
19713 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
19715 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
19718 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
19720 PR rtl-optimization/79949
19721 * lra-constraints.c (process_alt_operands): Check memory when
19722 trying to predict a cycle. Print about the overall increase.
19724 2017-03-09 Richard Biener <rguenther@suse.de>
19726 PR middle-end/79971
19727 * gimple-expr.c (useless_type_conversion_p): Preserve
19728 TYPE_SATURATING for fixed-point types.
19730 2017-03-09 Richard Biener <rguenther@suse.de>
19733 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
19734 alignment of BLKmode params.
19736 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19739 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
19740 (VALL_NO_V2Q): Likewise.
19742 * config/aarch64/aarch64-simd.md
19743 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
19745 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
19746 VALL_NO_V2Q mode iterator.
19747 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
19749 2017-03-09 Martin Liska <mliska@suse.cz>
19751 PR tree-optimization/79631
19752 * tree-chkp-opt.c (chkp_is_constant_addr): Call
19753 tree_int_cst_sign_bit just for INTEGER constants.
19755 2017-03-09 Martin Liska <mliska@suse.cz>
19759 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
19762 2017-03-09 Marek Polacek <polacek@redhat.com>
19765 * tree.c (inchash::add_expr): Handle TREE_VEC.
19767 2017-03-09 Martin Liska <mliska@suse.cz>
19770 (chkp_narrow_size_and_offset): New function.
19771 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
19772 (void chkp_parse_bit_field_ref): New function.
19773 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
19774 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
19776 2017-03-09 Martin Liska <mliska@suse.cz>
19779 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
19780 (chkp_find_bounds_1): Remove gcc_unreachable.
19782 2017-03-09 Jakub Jelinek <jakub@redhat.com>
19785 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
19786 BUILT_IN_SYNC*, determine the access type from the size suffix and
19787 always build a MEM_REF with that type. Handle forgotten
19788 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
19791 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
19792 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
19793 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
19794 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
19795 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
19796 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
19797 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
19798 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
19799 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
19800 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
19801 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
19802 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
19803 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
19804 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
19805 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
19806 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
19807 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
19808 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
19809 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
19810 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
19811 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
19812 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
19813 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
19814 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
19815 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
19816 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
19817 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
19818 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
19819 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
19820 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
19821 definitions outside of __OPTIMIZE__ guarded section.
19824 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
19825 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
19826 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
19827 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
19830 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19832 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
19833 ("vfenez<mode>"): Add missing constraints.
19835 2017-03-08 Martin Sebor <msebor@redhat.com>
19838 * config/nds32/nds32.c (nds32_option_override):
19839 Fix misspelled diagnostic.
19841 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19844 * gimplify.c (gimplify_omp_for): Replace index var in outer
19845 taskloop statement with an artificial variable and add
19846 OMP_CLAUSE_PRIVATE clause for it.
19848 2017-03-08 Richard Biener <rguenther@suse.de>
19850 PR tree-optimization/79955
19851 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
19852 for accesses that are completely outside of the variable.
19854 2017-03-08 Andrew Haley <aph@redhat.com>
19856 PR tree-optimization/79943
19857 * tree-ssa-loop-split.c (compute_new_first_bound): When
19858 calculating the new upper bound, (END-BEG) should be added, not
19861 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19863 * config/avr/avr.md (setmemhi): Make sure match_dup
19864 operand number comes before match_scratch.
19866 2017-03-08 Richard Biener <rguenther@suse.de>
19868 PR tree-optimization/79920
19869 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
19870 with ncopies == 1 to ...
19871 (vect_transform_slp_perm_load): ... here. Properly compute
19872 all element loads by iterating VF times over the group. Do
19873 not handle ncopies (computed in a broken way) in
19874 vect_create_mask_and_perm.
19876 2017-03-08 Jakub Jelinek <jakub@redhat.com>
19879 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
19880 is a uniform vector, use uniform_vector_p return value instead of
19881 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
19883 2017-03-07 Marek Polacek <polacek@redhat.com>
19885 PR middle-end/79809
19886 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
19887 (alloca_call_type): Likewise.
19889 2017-03-07 Martin Liska <mliska@suse.cz>
19891 * gcov.c (process_args): Put comment to correct location.
19893 2017-03-07 Martin Liska <mliska@suse.cz>
19895 PR middle-end/68270
19896 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
19897 Use array_at_struct_end_p instead of DECL_CHAIN (field).
19898 (chkp_narrow_bounds_for_field): Likewise.
19899 (chkp_parse_array_and_component_ref): Pass one more argument to
19902 2017-03-07 Richard Biener <rguenther@suse.de>
19904 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
19907 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
19909 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
19910 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
19912 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19915 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
19916 to end of description.
19917 (PARAM_MAX_STORES_TO_MERGE): Likewise.
19919 2017-03-07 Jakub Jelinek <jakub@redhat.com>
19921 PR rtl-optimization/79901
19922 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
19924 (*avx512f_<code><mode>3<mask_name>): ... this.
19925 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
19926 iterator instead of VI8_AVX2_AVX512BW.
19928 PR rtl-optimization/79901
19929 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
19930 min/max expander, expand it using expand_vec_cond_expr.
19933 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
19936 2017-03-06 Jakub Jelinek <jakub@redhat.com>
19939 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
19940 to void * for PCH reasons.
19941 * dwarf2out.c (output_loc_operands, output_die): Cast
19942 v.val_vec.array to unsigned char *.
19944 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
19947 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
19950 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
19952 PR rtl-optimization/79571
19953 * lra-constraints.c (process_alt_operands): Calculate static
19954 reject and subtract it from overall when only addresses will be
19957 2017-03-06 Julia Koval <julia.koval@intel.com>
19960 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
19961 incoming stack boundary to 128 for 64-bit targets.
19963 2017-03-06 Richard Biener <rguenther@suse.de>
19965 PR tree-optimization/79894
19966 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
19967 to NULL after folding it.
19969 2017-03-06 Richard Biener <rguenther@suse.de>
19971 PR tree-optimization/79824
19972 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
19973 check disabling peeling for gaps.
19975 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
19977 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
19978 attributes): Document gettimeofday.
19980 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
19982 * config/s390/s390.c (s390_option_override_internal): Set
19983 PARAM_MIN_VECT_LOOP_BOUND
19985 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
19987 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
19988 * config/s390/s390.md: Likewise.
19990 2017-03-06 Jakub Jelinek <jakub@redhat.com>
19993 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
19994 (<avx2_avx512>_perm<mode>): Rename to ...
19995 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
19997 (<avx512>_perm<mode>_mask): Rename to ...
19998 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
20000 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
20001 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
20002 instead of VI8F_256_512.
20003 (avx512f_perm<mode>): New define_expand.
20004 (avx512f_perm<mode>_mask): Likewise.
20005 (avx512f_perm<mode>_1<mask_name>): New define_insn.
20006 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
20008 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20010 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
20011 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
20013 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
20015 2017-03-06 Martin Liska <mliska@suse.cz>
20018 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
20019 when having a SSA NAME w/o VAR_DECL assigned to it.
20021 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20023 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
20024 msa_dpsub_<su>_d): Fix MODE for vec_select.
20026 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
20028 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
20030 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
20032 2017-03-06 Richard Biener <rguenther@suse.de>
20034 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
20035 * plugin.c (register_plugin_info): Likewise.
20036 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
20038 2017-03-05 Jakub Jelinek <jakub@redhat.com>
20040 * config/i386/sse.md (sse_storehps, sse_storelps,
20041 avx_<castmode><avxsizesuffix>_<castmode>,
20042 avx512f_<castmode><avxsizesuffix>_<castmode>,
20043 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
20044 in condition that at least one operand is not a MEM.
20046 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20048 PR middle-end/79805
20049 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
20050 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
20052 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
20053 gimple_call_nothrow_p flag based on whether original builtin can throw.
20054 If it can, emit following stmts on the fallthrough edge.
20055 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
20056 don't create new bb if inserting just debug stmts on the edge, try to
20057 insert them on the fallthru bb or just reset debug stmts.
20059 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
20062 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
20063 restore recog_data (including the operand rtxes inside it) around
20064 the call to get_insn_template.
20066 2017-03-03 Martin Sebor <msebor@redhat.com>
20068 PR tree-optimization/79699
20069 * context.c (context::~context): Free MPFR caches to avoid
20070 a memory leak on program exit.
20072 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20074 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
20075 Use wide_int::ulow () instead of .elt (0).
20077 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20079 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
20080 (*pushxf): Limit oF constraint to 32bit targets and add oC
20081 constraint for 64bit targets.
20082 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
20083 (*pushdf): Change rmF constraint to rmC.
20085 2017-03-03 Martin Liska <mliska@suse.cz>
20087 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
20088 Remove unused variable.
20090 2017-03-03 Jakub Jelinek <jakub@redhat.com>
20093 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
20094 is a memory operand, increase num_memory.
20095 (ix86_expand_args_builtin): Likewise.
20097 2017-03-03 Jan Hubicka <jh@suse.cz>
20100 * ipa-devirt.c (maybe_record_node): Properly handle
20101 __cxa_pure_virtual visibility.
20103 2017-03-03 Martin Liska <mliska@suse.cz>
20105 PR tree-optimization/79803
20106 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
20108 (pass_loop_prefetch::execute): Disabled optimization if an
20109 assumption about L1 cache size is not met.
20111 2017-03-03 Martin Liska <mliska@suse.cz>
20113 PR rtl-optimization/79574
20114 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
20115 (hash_scan_set): Likewise.
20116 (dump_hash_table): Likewise.
20117 (hoist_code): Likewise.
20119 2017-03-03 Richard Biener <rguenther@suse.de>
20121 * fixed-value.c (fixed_from_string): Restore use of elt (1)
20122 in place of uhigh ().
20123 (fixed_convert_from_real): Likewise.
20125 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
20128 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
20130 2017-03-03 Richard Biener <rguenther@suse.de>
20132 PR middle-end/79818
20133 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
20134 TYPE_OVERFLOW_UNDEFINED check.
20136 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20138 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
20140 (vector_ae_<mode>_p): Likewise.
20141 (vector_nez_<mode>_p): Likewise.
20142 (vector_ne_v2di_p): Likewise.
20143 (vector_ae_v2di_p): Likewise.
20144 (vector_ne_<mode>_p): Likewise.
20145 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
20147 (vsx_tsqrt<mode>2_fe): Likewise.
20149 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
20152 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
20154 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20156 PR rtl-optimization/79780
20157 * cprop.c (one_cprop_pass): When second and further conditional trap
20158 in a single basic block is turned into an unconditional trap, turn it
20159 into a deleted note to avoid RTL verification failures.
20161 2017-03-02 Richard Biener <rguenther@suse.de>
20163 * fold-const.c (const_binop): Use ulow () instead of elt (0).
20165 2017-03-02 Richard Biener <rguenther@suse.de>
20167 PR tree-optimization/79345
20169 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
20170 param and abort the walk, returning -1 if it is hit.
20171 (walk_aliased_vdefs): Take a limit param and pass it on.
20172 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
20173 defaulting to 0 and return a signed int.
20174 * tree-ssa-uninit.c (struct check_defs_data): New struct.
20175 (check_defs): New helper.
20176 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
20177 about uninitialized memory.
20178 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
20179 bogus uninitialized warning.
20180 (fixed_convert_from_real): Likewise.
20182 2017-03-02 Bin Cheng <bin.cheng@arm.com>
20184 PR tree-optimization/66768
20185 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
20186 iv_use if base object can't be determined.
20188 2017-03-02 Jakub Jelinek <jakub@redhat.com>
20190 PR tree-optimization/79345
20191 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
20192 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
20193 (get_pattern_stats): Initialize it.
20194 * genemit.c (gen_expand): Verify match_scratch numbers come after
20195 match_operand/match_dup numbers.
20196 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
20197 match_scratch numbers.
20198 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
20200 * config/s390/s390.md (trunctdsd2): Likewise.
20202 2017-03-02 Richard Biener <rguenther@suse.de>
20204 * wide-int.h (wide_int_storage::operator=): Implement in terms
20207 2017-03-02 Richard Biener <rguenther@suse.de>
20209 PR tree-optimization/79777
20210 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
20211 the to insert expression to sth existing.
20213 2017-03-01 Martin Sebor <msebor@redhat.com>
20215 PR middle-end/79692
20216 * gimple-ssa-sprintf.c
20217 (directive::known_width_and_precision): New function.
20218 (format_integer): Use it.
20219 (get_mpfr_format_length): Consider the full range of precision
20220 when computing %g output with the # flag. Set the likely byte
20221 count to 3 rather than 1 when precision is indeterminate.
20222 (format_floating): Correct the lower bound of precision.
20224 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20226 * doc/invoke.texi: Document default code model for 64-bit Linux.
20228 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20231 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
20232 udiv rather than div since input pattern is unsigned.
20234 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
20236 * config/i386/i386.c (print_reg): Warn for values of
20237 unsupported size in integer register.
20239 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
20242 * config/rs6000/predicates.md (current_file_function_operand): Do
20243 not allow self calls to be local if the function is replaceable.
20245 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
20248 * config/rs6000/altivec.h (vec_ctz and others): Change the
20249 preprocessor macro that controls conditional compilation from
20250 _ARCH_PWR9 to __POWER9_VECTOR__.
20251 (vec_all_ne): Change parameterization of __altivec_scalar_pred
20252 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
20253 control (instead of _ARCH_PWR9 control) so that template
20254 definition uses power9-specific function.
20255 (vec_any_eq): Likewise.
20256 (vec_all_ne): Change macro definition to use a power9-specific
20257 expansion under #ifdef __POWER9_VECTOR__ control (instead of
20258 _ARCH_PWR9 control).
20259 (vec_any_eq) Likewise.
20260 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
20261 expansion for CMPNEF to remove support for xvcmpnesp instruction.
20262 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
20263 support for xvcmpnedp instruction.
20264 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
20265 macro expansion so that Power9 implementation of vec_all_ne does
20266 not use the AltiVec predicate framework.
20267 (VCMPNEH_P): Likewise.
20268 (VCMPNEW_P): Likewise.
20269 (VCMPNED_P): Likewise.
20270 (VCMPNEFP_P): Likewise.
20271 (VCMPNEDP_P): Likewise.
20272 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
20273 implementation of vec_any_eq to not use AltiVec predicate
20275 (VCMPAEH_P): Likewise.
20276 (VCMPAEW_P): Likewise.
20277 (VCMPAED_P): Likewise.
20278 (VCMPAEFP_P): Likewise.
20279 (VCMPAEDP_P): Likewise.
20280 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
20281 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
20282 not use the AltiVec predicate framework.
20283 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
20284 of vec_any_eq to not use AltiVec predicate framework.
20285 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
20286 support for predefined __POWER9_VECTOR__ macro to indicate that
20287 Power9 instruction selection is enabled.
20288 (altivec_overloaded_builtins): Remove extraneous
20289 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
20290 function argument types RS6000_BTI_bool_V16QI and
20291 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
20292 entry for overloaded function argument types RS6000_BTI_bool_V4SI
20293 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
20294 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
20295 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
20296 Power9 for implementations of vec_cmpne. Change the signature for
20297 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
20298 (representing vec_all_ne) to remove the previously described first
20299 argument of type RS6000_BTI_INTSI, as this was an artifact of
20300 reliance on the AltiVec predicate framework, which is no longer
20301 used in the implementation of these functions. Add
20302 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
20303 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
20304 since, unlike the AltiVec predicate framework implementation, we
20305 do not share function descriptors between vec_alle and vec_anyeq.
20306 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
20307 set of modes that receive special treatment even when
20308 TARGET_P9_VECTOR is true. The special treatment emits code that
20309 does not depend on Power9 instructions.
20310 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
20311 define_expand to not rely on AltiVec predicate framework.
20312 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20314 (vector_ne_v2di_p): Change this define_expand to not rely on
20315 AltiVec predicate framework.
20316 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
20318 (vector_ne_<mode>_p): Change this define_expand to not rely on
20319 AltiVec predicate framework.
20320 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
20322 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
20323 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
20324 define_insn pattern.
20325 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
20326 define_insn pattern because the xvcmpne<VSs>. instruction is not
20328 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
20329 instruction is not supported.
20331 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20333 * config/nvptx/nvptx.c: Include intl.h.
20335 2017-03-01 Martin Jambor <mjambor@suse.cz>
20338 * ipa-prop.h (ipa_bits): Removed field known.
20339 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
20340 to pointers. Adjusted their comments to warn about their sharing.
20341 (ipcp_transformation_summary): Change bits to a vector of pointers.
20342 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
20343 (ipa_get_ipa_bits_for_value): Declare.
20344 * tree-vrp.h (value_range): Mark as GTY((for_user)).
20345 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
20346 (ipa_bits_hash_table): Likewise.
20347 (ipa_vr_ggc_hash_traits): Likewise.
20348 (ipa_vr_hash_table): Likewise.
20349 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
20350 being pointers and vr_known being removed.
20351 (ipa_set_jf_unknown): Likewise.
20352 (ipa_get_ipa_bits_for_value): New function.
20353 (ipa_set_jfunc_bits): Likewise.
20354 (ipa_get_value_range): New overloaded functions.
20355 (ipa_set_jfunc_vr): Likewise.
20356 (ipa_compute_jump_functions_for_edge): Use the above functions to
20357 construct bits and vr parts of jump functions.
20358 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
20359 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20361 (ipcp_grow_transformations_if_necessary): Also allocate
20362 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
20364 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
20365 them. Fix too long lines.
20366 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
20367 vr_known being removed.
20368 (ipa_read_jump_function): Use new setter functions to construct bits
20369 and vr parts of jump functions or set them to NULL.
20370 (write_ipcp_transformation_info): Adjust for bits being pointers.
20371 (read_ipcp_transformation_info): Likewise.
20372 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
20374 Include gt-ipa-prop.h.
20375 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
20377 (ipcp_store_bits_results): Likewise.
20378 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
20379 Do not write to existing jump functions but use a temporary instead.
20381 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20384 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
20385 attempt to use its first operand as BIT_FIELD_REF base.
20387 2017-03-01 Richard Biener <rguenther@suse.de>
20389 PR middle-end/79721
20390 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
20391 interpolating formula in wrapping arithmetic.
20392 (chrec_apply): Convert chrec_evaluate return value to wanted type.
20394 2017-03-01 Jakub Jelinek <jakub@redhat.com>
20396 PR tree-optimization/79734
20397 * tree-vect-generic.c (expand_vector_condition): Optimize
20398 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
20399 Handle VEC_COND_EXPR where comparison has different inner width from
20400 type's inner width.
20402 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
20404 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
20405 markup, and similar issues. Remove @opindex entries for things
20406 that aren't options. Add missing -mmpy-option entries.
20408 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20410 PR tree-optimization/79737
20411 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
20412 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
20413 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
20414 instead of byte_size. Formatting fix.
20415 (shift_bytes_in_array_right): Formatting fix.
20417 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
20420 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
20421 condition on optimize for the leaf function test.
20423 2017-02-28 Martin Liska <mliska@suse.cz>
20426 * read-rtl-function.c (function_reader::handle_unknown_directive):
20427 Bail out when one uses -flto.
20429 2017-02-28 Martin Liska <mliska@suse.cz>
20431 * common.opt: Replace space with tabular for options of <number>
20433 * config/i386/i386.opt: Show <number> value for
20434 -mlarge-data-threshold.
20435 * opts.c (print_filtered_help): Do not display number in hexadecimal
20438 2017-02-28 Martin Liska <mliska@suse.cz>
20440 * common.opt: Fix --help=option -Q for options which are of
20443 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
20445 * config/i386/i386.c (print_reg): Error out for values
20446 of 8-bit size in invalid integer register.
20448 2017-02-28 Martin Sebor <msebor@redhat.com>
20450 PR tree-optimization/79691
20451 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
20453 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20456 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
20457 gcc_unreachable with output_operand_lossage.
20459 2017-02-28 Richard Biener <rguenther@suse.de>
20461 PR tree-optimization/79740
20462 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
20464 (visit_nary_op): Insert the nary into the hashtable if we
20465 pattern-matched sth.
20466 * tree-ssa-pre.c (eliminate_insert): Robustify.
20468 2017-02-28 Richard Biener <rguenther@suse.de>
20470 PR middle-end/79731
20471 * fold-const.c (decode_field_reference): Reject out-of-bound
20474 2017-02-28 Jakub Jelinek <jakub@redhat.com>
20476 * config/i386/i386.c: Include intl.h.
20477 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
20478 instead of just cond ? "..." : "...".
20479 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
20480 * coverage.c (read_counts_file): Likewise.
20481 * omp-offload.c: Include intl.h.
20482 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
20483 of just cond ? "..." : "...".
20484 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
20485 of just cond ? "..." : "...".
20487 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
20490 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
20492 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
20493 'tune for' CPU name.
20494 * config/arm/arm-cpu-data.h: Regenerated.
20496 2017-02-28 Richard Biener <rguenther@suse.de>
20498 PR tree-optimization/79732
20499 * tree-inline.c (expand_call_inline): Do not shadow var.
20501 2017-02-28 Richard Biener <rguenther@suse.de>
20503 PR tree-optimization/79723
20504 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
20505 address-space properly.
20507 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
20509 * doc/optinfo.texi (Optimization groups): Fix option used for
20511 * doc/invoke.texi (-fopt-info): Document "omp".
20512 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
20513 (OPTGROUP_ALL): Add OPTGROUP_OMP.
20514 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
20515 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
20516 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
20518 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
20520 * dumpfile.c (optgroup_options): Instead of "openmp", associate
20521 OPTGROUP_OMP with "omp".
20523 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
20526 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
20527 for arithmetic shift of unsigned V2DI.
20529 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
20531 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
20532 arc/linux.h headers.
20533 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
20534 (LINK_SPEC): Likewise.
20535 (ARC_TLS_EXTRA_START_SPEC): Likewise.
20536 (EXTRA_SPECS): Likewise.
20537 (STARTFILE_SPEC): Likewise.
20538 (ENDFILE_SPEC): Likewise.
20539 (LIB_SPEC): Likewise.
20540 (TARGET_SDATA_DEFAULT): Likewise.
20541 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
20542 (MULTILIB_DEFAULTS): Likewise.
20543 (DWARF2_UNWIND_INFO): Likewise.
20544 * config/arc/big.h: New file.
20545 * config/arc/elf.h: Likewise.
20546 * config/arc/linux.h: Likewise.
20547 * config/arc/t-uClibc: Remove.
20549 2017-02-27 Bin Cheng <bin.cheng@arm.com>
20551 PR tree-optimization/77536
20552 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
20553 (tree_transform_and_unroll_loop): Use above function to compute the
20554 estimated niter of unrolled loop and use it when scaling profile.
20555 Also use count info rather than frequency if it's non-zero.
20556 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
20557 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
20558 (vect_transform_loop): Call above function.
20560 2017-02-27 Richard Biener <rguenther@suse.de>
20562 PR tree-optimization/45397
20563 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
20564 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
20565 (visit_nary_op): Add pattern matching for CSEing sign-changed
20566 or truncated operations with wider ones.
20568 2017-02-27 Richard Biener <rguenther@suse.de>
20570 PR tree-optimization/79690
20571 * tree-vect-stmts.c (vectorizable_store): Use vector type
20572 built from the DR with address-space.
20574 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
20576 * doc/invoke.texi (Optimize Options): Refine the description
20577 of asan-use-after-return.
20579 2017-02-25 Alan Modra <amodra@gmail.com>
20581 PR rtl-optimization/79584
20582 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
20583 base, not ad->base_term, the reg within base. Remove assertion
20584 that ad->base == ad->base_term. Replace gen_int_mode using
20585 bogus mode with const0_rtx.
20587 2017-02-25 Jakub Jelinek <jakub@redhat.com>
20589 PR middle-end/79396
20590 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
20591 FMA_EXPR like tcc_binary or tcc_unary.
20593 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
20596 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
20598 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
20599 (output_loc_operands): Handle DW_OP_call_ref and
20600 DW_OP_GNU_variable_value.
20601 (struct variable_value_struct): New type.
20602 (struct variable_value_hasher): Likewise.
20603 (variable_value_hash): New variable.
20604 (string_types): Remove.
20605 (copy_loc_descr): New function.
20606 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
20607 (prepend_loc_descr_to_each): New function.
20608 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
20609 instead of add_loc_descr_to_each if the first argument is single
20610 location list and the second has multiple.
20611 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
20612 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
20613 when looking for variable value which doesn't have other location info.
20614 (loc_list_from_tree): Formatting fix.
20615 (gen_array_type_die): Simplify DW_AT_string_length handling.
20616 (adjust_string_types): Remove.
20617 (gen_subprogram_die): Don't call adjust_string_types nor test/set
20618 string_types. Call resolve_variable_values.
20619 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
20620 (resolve_addr_in_expr): Likewise. Add A argument.
20621 (copy_deref_exprloc): Remove deref argument. Adjust for the
20622 original expression being DW_OP_GNU_variable_value with optionally
20623 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
20624 optionally after it.
20625 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
20626 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
20627 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
20628 (variable_value_hasher::hash, variable_value_hasher::equal): New
20630 (resolve_variable_value_in_expr, resolve_variable_value,
20631 resolve_variable_values, note_variable_value_in_expr,
20632 note_variable_value): New functions.
20633 (dwarf2out_early_finish): Call note_variable_value on all toplevel
20636 2017-02-24 Jakub Jelinek <jakub@redhat.com>
20639 * opts.h (handle_generated_option): Add GENERATED_P argument.
20640 * opts-common.c (handle_option): Adjust function comment.
20641 (handle_generated_option): Add GENERATED_P argument, pass it to
20643 (control_warning_option): Pass false to handle_generated_option
20645 * opts.c (maybe_default_option): Pass true to handle_generated_option
20647 * optc-gen.awk: Likewise.
20649 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20651 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
20652 a REG, look at the REG it is a SUBREG of.
20653 (splitter for cmpeqsi_t): Ditto.
20655 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20657 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
20658 the special USEs with the pattern of the insn, not the insn itself.
20660 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
20663 * doc/invoke.texi: Document -mload-store-pairs.
20665 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20666 Sandra Loosemore <sandra@codesourcery.com>
20668 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
20669 argument isn't a CONST_INT.
20670 (nios2_alternate_compare_const): Assert op is a CONST_INT.
20671 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
20672 (nios2_validate_compare): Bypass alternate compare logic if *op2
20673 is not a CONST_INT.
20674 (ldstwm_operation_p): Return false if first_base is not a REG or
20675 if first_offset is not a CONST_INT.
20677 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20679 * config/cris/cris.md: Use correct operand in a define_peephole2.
20681 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20683 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
20685 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
20687 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
20688 this_insn if it is an INSN or JUMP_INSN.
20689 (force_offsettable): Look at base, not at addr.
20690 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
20691 on things that aren't necessarily CONST_INTs.
20693 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
20695 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
20696 -mfpmath=sse is the default also for x86-32 targets with SSE2
20697 instruction set when @option{-ffast-math} is enabled
20699 2017-02-24 Jeff Law <law@redhat.com>
20701 PR rtl-optimizatoin/79286
20702 * ira.c (update_equiv_regs): Drop may_trap_p exception to
20705 2017-02-24 Richard Biener <rguenther@suse.de>
20707 PR tree-optimization/79389
20708 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
20711 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
20713 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
20714 function comment to reflect reality.
20715 (loop_exits_before_overflow): Fix typo in function description.
20717 2017-02-24 Richard Biener <rguenther@suse.de>
20719 PR tree-optimization/79389
20720 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
20721 properly that a threading opportunity exists. Detect conditional
20722 copy/constant propagation opportunities.
20724 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
20726 * config/visium/visium.md (type): Add trap.
20727 (b): New mode attribute.
20728 (*btst): Rename into...
20729 (*btst<mode>): ...this and adjust.
20730 (*cbranchsi4_btst_insn): Rename into...
20731 (*cbranch<mode>4_btst_insn): ...this and adjust.
20732 (trap): New define_insn.
20734 2017-02-23 Jakub Jelinek <jakub@redhat.com>
20736 PR tree-optimization/79389
20737 * ifcvt.c (struct noce_if_info): Add rev_cond field.
20738 (noce_reversed_cond_code): New function.
20739 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
20740 reversed_comparison_code. Formatting fix.
20741 (noce_try_store_flag): Test rev_cond != NULL in addition to
20742 reversed_comparison_code.
20743 (noce_try_store_flag_constants): Likewise.
20744 (noce_try_store_flag_mask): Likewise.
20745 (noce_try_addcc): Use rev_cond if non-NULL instead of
20746 reversed_comparison_code.
20747 (noce_try_cmove_arith): Likewise. Formatting fixes.
20748 (noce_try_minmax, noce_try_abs): Clear rev_cond.
20749 (noce_find_if_block): Initialize rev_cond.
20750 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
20751 instead of false as last argument never attempt to reverse it
20754 2017-02-23 Bin Cheng <bin.cheng@arm.com>
20756 PR tree-optimization/79663
20757 * tree-predcom.c (combine_chains): Process refs in reverse order
20758 only for ZERO length chains, and add explaining comment.
20760 2017-02-23 Jeff Law <law@redhat.com>
20762 PR tree-optimization/79578
20763 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
20764 in call to operand_equal_p.
20766 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
20769 * config/i386/cpuid.h: Fix another undefined behavior.
20771 2017-02-23 Richard Biener <rguenther@suse.de>
20773 PR tree-optimization/79683
20774 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
20775 vector types for data-refs.
20777 2017-02-23 Martin Liska <mliska@suse.cz>
20779 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
20781 2017-02-23 Jakub Jelinek <jakub@redhat.com>
20783 PR middle-end/79665
20784 * internal-fn.c (get_range_pos_neg): Moved to ...
20785 * tree.c (get_range_pos_neg): ... here. No longer static.
20786 * tree.h (get_range_pos_neg): New prototype.
20787 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
20788 are known to be in between 0 and signed maximum inclusive, try to
20789 expand both unsigned and signed divmod and use the cheaper one from
20792 2017-02-22 Jeff Law <law@redhat.com>
20794 PR tree-optimization/79578
20795 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
20796 to compare base operands.
20798 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
20801 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
20802 gpc_reg_operand instead of fpr_reg_operand.
20804 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
20806 * config/mips/mips.c (mips_return_in_memory): Force FP
20807 vector types to be returned in memory for o32 ABI.
20809 2017-02-22 Jakub Jelinek <jakub@redhat.com>
20811 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
20812 instead of DW_TAG_member for static data member declarations and don't
20813 set no_linkage_name for static inline data members.
20814 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
20817 2017-02-22 Martin Liska <mliska@suse.cz>
20819 * doc/invoke.texi: Replace inequality signs with square brackets
20822 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20824 PR tree-optimization/68644
20825 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
20827 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
20830 * lra-constraints.c (simplify_operand_subreg): Handle
20831 WORD_REGISTER_OPERATIONS targets.
20833 2017-02-22 Jakub Jelinek <jakub@redhat.com>
20836 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
20837 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
20838 elimination by swapping fld*.
20840 2017-02-22 Richard Biener <rguenther@suse.de>
20842 PR tree-optimization/79673
20843 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
20844 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
20845 irrelevant address-space qualifiers and avoiding a
20846 ADDR_SPACE_CONVERT_EXPR from fold_convert.
20848 2017-02-22 Richard Biener <rguenther@suse.de>
20850 PR tree-optimization/79666
20851 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
20852 to not symbolically negate if that may introduce undefined
20855 2017-02-22 Martin Liska <mliska@suse.cz>
20858 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
20859 * data-streamer-out.c (streamer_write_gcov_count_stream):
20861 * value-prof.c (stream_out_histogram_value): Make assert more
20862 precise based on type of counter.
20864 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
20867 * config/i386/i386.md (standard_x87sse_constant_load splitter):
20868 Use nonimmediate_operand instead of memory_operand for operand 1.
20869 (float-extend standard_x87sse_constant_load splitter): Ditto.
20871 2017-02-21 Jeff Law <law@redhat.com>
20873 PR tree-optimization/79621
20874 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
20875 blocks with edges to themselves.
20877 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20880 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
20881 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
20882 Use gimple_call_builtin_p.
20885 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
20886 on temporarily removed DEBUG_INSNs.
20888 PR tree-optimization/79649
20889 * tree-loop-distribution.c (classify_partition): Give up on
20890 non-generic address space loads/stores.
20892 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
20894 * doc/loop.texi (Loop manipulation): Remove nonexistent
20895 tree_ssa_loop_version from the documentation.
20896 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
20898 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20901 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
20902 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
20903 * config/rs6000/rs6000.c: Include except.h.
20904 (rs6000_expand_split_stack_prologue): Call
20905 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
20907 2017-02-21 Martin Jambor <mjambor@suse.cz>
20910 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
20911 have been analyzed.
20913 2017-02-21 Martin Jambor <mjambor@suse.cz>
20915 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
20916 for backward compatibility only.
20917 * doc/invoke.texi (Option Summary): Remove all references to
20918 -fipa-cp-alignment.
20920 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
20924 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20926 * lra-constraints.c (curr_insn_transform): Handle
20927 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
20929 2017-02-21 Martin Liska <mliska@suse.cz>
20931 * config/i386/i386.opt: Replace -masm-dialect with -masm.
20933 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
20935 PR translation/79638
20936 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
20938 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
20941 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
20942 (arm_function_ok_for_sibcall): Return false for an indirect call by
20943 descriptor if all the argument registers are used.
20944 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
20945 alignment of the function.
20947 2017-02-21 Jakub Jelinek <jakub@redhat.com>
20949 PR tree-optimization/61441
20950 * simplify-rtx.c (simplify_const_unary_operation): For
20951 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
20952 the sNaN unmodified.
20954 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
20956 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
20957 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
20958 instead of SYSTEM_HEADER_DIR.
20960 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
20961 Martin Liška <mliska@suse.cz>
20963 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
20964 Fix typos and grammar, use active voice, and clarify.
20966 2017-02-20 Marek Polacek <polacek@redhat.com>
20968 PR middle-end/79537
20969 * gimplify.c (gimplify_expr): Handle unused *&&L;.
20972 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
20974 2017-02-20 Jakub Jelinek <jakub@redhat.com>
20977 * config/i386/i386.c (ix86_expand_builtin): Handle
20978 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
20979 ix86_builtins_isa[fcode].isa as a requirement of those
20980 flags and any other flag in the bitmask.
20981 (ix86_init_mmx_sse_builtins): Use 0 instead of
20982 ~OPTION_MASK_ISA_64BIT as mask.
20983 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
20984 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
20985 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
20986 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
20988 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20991 * lra-constraints.c (split_reg): Check requested split mode
20992 is supported by the register.
20994 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
20996 * lra-constraints.c (simplify_operand_subreg): Remove early
20999 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21002 * lra-constraints.c (curr_insn_transform): Tighten condition
21003 for converting SUBREG reloads from OP_OUT to OP_INOUT.
21005 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
21008 * lra-constraints.c (curr_insn_transform): Handle
21009 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
21011 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
21014 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
21016 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
21018 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
21021 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
21024 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21026 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
21029 2017-02-18 Jakub Jelinek <jakub@redhat.com>
21031 * final.c (last_columnnum, override_columnnum): New variables.
21032 (final_start_function): Set last_columnnum, pass it to begin_prologue
21033 hook and pass 0 to dwarf2out_begin_prologue.
21034 (final_scan_insn): Update override_columnnum. Pass last_columnnum
21035 to source_line debug hook.
21036 (notice_source_line): Compute last_columnnum and for debug_column_info
21037 return true on column changes.
21038 * debug.h (struct gcc_debug_hooks): Add column argument to
21039 source_line and begin_prologue hooks.
21040 (debug_nothing_int_charstar_int_bool): Remove prototype.
21041 (debug_nothing_int_int_charstar,
21042 debug_nothing_int_int_charstar_int_bool): New prototypes.
21043 (dwarf2out_begin_prologue): Add column argument.
21044 * debug.c (do_nothing_debug_hooks): Adjust source_line and
21045 begin_prologue hooks.
21046 (debug_nothing_int_charstar_int_bool): Remove.
21047 (debug_nothing_int_int_charstar,
21048 debug_nothing_int_int_charstar_int_bool): New functions.
21049 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
21050 through to dwarf2out_source_line.
21051 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
21052 (dwarf2out_source_line): Add column argument, emit it if requested.
21053 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
21055 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21056 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
21057 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
21058 through to dwarf2out_begin_prologue.
21059 (vmsdbgout_source_line): Add column argument, pass it through to
21060 dwarf2out_source_line.
21061 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
21062 dbxout_source_line caller.
21063 (dbxout_source_line): Add column argument.
21065 * common.opt (gno-column-info, gcolumn-info): New options.
21066 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
21067 (check_die): Also test for multiple DW_AT_decl_column attributes.
21068 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
21069 DW_AT_decl_column if requested.
21070 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
21072 (gen_variable_die): Likewise.
21073 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
21074 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
21077 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
21078 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
21079 (ix86_handle_option): Handle OPT_m3dnowa.
21080 * doc/invoke.texi (-m3dnowa): Document.
21081 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
21082 -m3dnowa instead of -m3dnow -march=athlon.
21085 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
21086 instead of gcc_assert for K, r and R code checks. Formatting fixes.
21088 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21091 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
21092 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
21093 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
21094 generator for vsx_xxpermdi_<mode>_be.
21095 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
21096 force big-endian semantics.
21097 (vsx_xxpermdi_<mode>_be): New define_expand with same
21098 implementation as previous version of vsx_xxpermdi_<mode>.
21100 2017-02-17 Jakub Jelinek <jakub@redhat.com>
21102 PR tree-optimization/79327
21103 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
21104 variable, its initialization and use.
21106 2017-02-17 Julia Koval <julia.koval@intel.com>
21108 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
21109 (OPTION_MASK_ISA_PKU_UNSET): New.
21110 (ix86_handle_option): Handle -mrdpid.
21111 * config/i386/cpuid.h (bit_RDPID): New.
21112 * config/i386/driver-i386.c (host_detect_local_cpu):
21113 Detect RDPID feature.
21114 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
21115 * config/i386/i386-c.c (ix86_target_macros_internal):
21117 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
21118 (ix86_valid_target_attribute_inner_p): Add "rdpid".
21119 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
21120 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
21121 * config/i386/i386.md (define_insn "rdpid"): New.
21122 * config/i386/i386.opt Add -mrdpid.
21123 * config/i386/immintrin.h (_rdpid_u32): New.
21125 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
21127 PR rtl-optimization/79541
21128 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
21129 instead of transforming it into USE.
21131 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
21133 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
21134 If HONOR_SNANS (SFmode) force the input to a register.
21135 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
21136 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
21137 an frsp or similar insn.
21139 2017-02-17 Martin Liska <mliska@suse.cz>
21141 PR rtl-optimization/79577
21142 * params.def (selsched-max-sched-times): Increase minimum to 1.
21144 2017-02-17 Martin Liska <mliska@suse.cz>
21146 PR rtl-optimization/79574
21147 * gcse.c (want_to_gcse_p): Prevent integer overflow.
21149 2017-02-17 Martin Liska <mliska@suse.cz>
21151 PR tree-optimization/79529
21152 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
21153 ssa_defined_default_def_p to handle cases which are implicitly
21155 * tree-ssa.c (ssa_defined_default_def_p): New function.
21156 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
21157 which are implicitly defined.
21158 * tree-ssa.h (ssa_defined_default_def_p): Declare.
21160 2017-02-17 Richard Biener <rguenther@suse.de>
21162 PR middle-end/79576
21163 * params.def (max-ssa-name-query-depth): Limit to 10.
21165 2017-02-17 Richard Biener <rguenther@suse.de>
21167 PR tree-optimization/79552
21168 * tree-ssa-structalias.c (visit_loadstore): Properly verify
21171 2017-02-17 Richard Biener <rguenther@suse.de>
21174 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
21176 2017-02-17 Marek Polacek <polacek@redhat.com>
21178 PR middle-end/79536
21179 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
21180 (fold_negate_expr): New wrapper.
21182 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
21184 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
21185 Correct terminology and de-emphasize pre-standard behavior.
21187 2017-02-16 Alan Modra <amodra@gmail.com>
21189 PR rtl-optimization/79286
21190 * ira.c (def_dominates_uses): New function.
21191 (update_equiv_regs): Don't create an equivalence for insns that
21192 may trap where the register def does not dominate the use.
21194 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
21196 PR rtl-optimization/78127
21197 * lra.c (lra): Call lra_eliminate before finish the loop after
21200 2017-02-16 Richard Biener <rguenther@suse.de>
21202 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
21204 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
21205 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
21206 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
21207 (isl_val_int_from_wi): New function.
21208 (extract_affine_gmp): Rename to ...
21209 (extract_affine_wi): ... this, take a widest_int.
21210 (extract_affine_int): Just wrap extract_affine_wi.
21211 (add_param_constraints): Use isl_val_int_from_wi.
21212 (add_loop_constraints): Likewise, and extract_affine_wi.
21214 2017-02-15 Jeff Law <law@redhat.com>
21216 PR middle-end/79521
21217 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
21218 ira_init_register_move_cost_if_necessary.
21220 2017-02-15 Martin Sebor <msebor@redhat.com>
21222 PR middle-end/32003
21223 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
21224 removed in a prior commit.
21226 2017-02-15 Bin Cheng <bin.cheng@arm.com>
21228 PR tree-optimization/79347
21229 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
21230 counters during peeling.
21232 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
21234 * Makefile.in (site.exp): Remove "set ISLVER".
21236 2017-02-15 Jakub Jelinek <jakub@redhat.com>
21239 * real.c (real_from_integer): Call real_convert even for decimal.
21241 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
21244 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
21246 2017-02-14 Andrew Pinski <apinski@cavium.com>
21248 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
21249 cores and change the partno/implementer to be correct.
21250 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
21251 the 'B" as the implementer.
21252 * config/aarch64/aarch64-tune.md: Regenerate.
21254 2017-02-14 Carl Love <cel@us.ibm.com>
21256 * config/rs6000/rs6000.c: Add case statement entry to make the
21257 xvcvuxdsp built-in argument unsigned.
21258 * config/rs6000/vsx.md: Fix the source and return operand types so they
21259 match the instruction definitions from the ISA document. Fix typo
21260 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
21263 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
21266 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
21267 member early_clobber_alts.
21268 * lra-lives.c (reg_early_clobber_p): New.
21269 (process_bb_lives): Use it.
21270 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
21271 (debug_operand_data): Initialize early_clobber_alts.
21272 (setup_operand_alternative): Set up early_clobber_alts.
21273 (collect_non_operand_hard_regs): Ditto. Pass early clobber
21274 alternatives to new_insn_reg.
21275 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
21277 (lra_update_insn_regno_info): Pass the new arg.
21279 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21281 PR middle-end/79505
21282 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
21283 (new_oacc_loop_raw): Don't clear already cleared fields.
21286 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
21287 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
21288 _mm512_prefetch_i64gather_ps): New inline functions and macros.
21290 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
21293 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
21295 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
21298 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
21299 the extra instruction to the right place to store 128-bit constant
21302 2017-02-14 Martin Sebor <msebor@redhat.com>
21304 PR middle-end/79448
21305 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
21306 warning for strings of unknown length.
21308 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
21310 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
21312 2017-02-14 Jeff Law <law@redhat.com>
21315 * ira-costs.c (scan_one_insn): Initialize register move costs
21316 for pseudos seen in USE/CLOBBER insns.
21318 PR tree-optimization/79095
21319 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
21320 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
21321 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
21322 if the operands are known to be not equal, then the resulting range
21324 (intersect_ranges): If the new range is ~[0,0] and the old range is
21325 wide, then prefer ~[0,0].
21326 * tree-vrp.c (overflow_comparison_p_1): New function.
21327 (overflow_comparison_p): New function.
21328 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
21329 if NAME is used in an overflow test.
21330 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
21331 overflow check that can be expressed as an equality test, then adjust
21332 ops to be that equality test.
21334 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21336 * config/s390/s390-builtin-types.def: Remove flags argument.
21337 * config/s390/s390.c (s390_init_builtins): Likewise.
21339 2017-02-14 Martin Liska <mliska@suse.cz>
21341 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
21342 vector. Fix trailing white spaces.
21344 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
21346 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
21349 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21351 PR rtl-optimization/68664
21352 * config/arm/arm.c (arm_sched_can_speculate_insn):
21353 New function. Declare prototype.
21354 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21356 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21358 PR rtl-optimization/68664
21359 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
21361 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
21363 2017-02-14 Amit Pawar <amit.pawar@amd.com>
21365 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
21366 max skip bytes for function, loop and jump.
21368 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21370 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
21371 ABS_EXPR for gimple dump.
21373 2017-02-14 Jakub Jelinek <jakub@redhat.com>
21376 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
21378 PR tree-optimization/79408
21379 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
21380 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
21381 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
21382 also if rhs1 is INTEGER_CST.
21384 2017-02-14 Richard Biener <rguenther@suse.de>
21386 PR middle-end/79432
21387 * tree-into-ssa.c (insert_phi_nodes): When the function can
21388 have abnormal edges rewrite SSA names with broken use-def
21389 dominance out of SSA and register them for PHI insertion.
21391 2017-02-13 Martin Sebor <msebor@redhat.com>
21393 PR middle-end/79496
21394 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
21395 clearing info.nowrite flag when snprintf size argument is a range.
21397 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21399 * cprop.c (cprop_jump): Add missing space in string literal.
21400 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
21401 (get_constraint_for_component_ref): Likewise.
21402 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
21403 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
21404 * lra-constraints.c (process_alt_operands): Likewise.
21405 * ipa-inline.c (inline_small_functions): Likewise.
21406 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
21407 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21408 * trans-mem.c (diagnose_tm_1_op): Likewise.
21409 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
21410 (grid_parallel_clauses_gridifiable): Likewise.
21412 * config/nvptx/mkoffload.c (process): Add space in between
21415 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
21416 "MOD4_SSE_REGS" and "ALL_REGS".
21418 * spellcheck.c (test_data): Add , in between "foo" and "food".
21420 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21423 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
21424 boundary crossing check and subsequent code generation agree.
21426 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21428 * config/aarch64/aarch64.c (has_memory_op): Delete.
21429 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
21432 2017-02-13 Jakub Jelinek <jakub@redhat.com>
21434 PR rtl-optimization/79388
21435 PR rtl-optimization/79450
21436 * combine.c (distribute_notes): When removing TEM_INSN for which
21437 corresponding dest has last value recorded, invalidate that last
21440 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21442 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
21443 of explicit '@'. Add missing assembly comment marker on branch costs
21446 2017-02-13 Nathan Sidwell <nathan@acm.org>
21448 * gengtype-lex.l (<in_struct>): Add '/'.
21450 2017-02-13 Martin Liska <mliska@suse.cz>
21453 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
21455 2017-02-13 Richard Biener <rguenther@suse.de>
21457 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
21459 * configure: Re-generate.
21460 * config.in: Likewise.
21461 * graphite-dependences.c: Simplify as if
21462 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
21463 * graphite-isl-ast-to-gimple.c: Likewise.
21464 * graphite-optimize-isl.c: Likewise.
21465 * graphite-poly.c: Likewise.
21466 * graphite-sese-to-poly.c: Likewise.
21467 * graphite.h: Likewise.
21468 * toplev.c: Include isl/version.h and use isl_version () for
21469 printing the ISL version.
21470 * doc/install.texi: Update ISL requirement.
21472 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21474 * doc/standards.texi (Standards): Update reference to
21477 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
21479 * doc/extend.texi (Named Address Spaces): sourceware.org now
21481 * doc/install.texi (Binaries): Ditto.
21484 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21486 * doc/cpp.texi: Replace "stringify"/"stringification" with C
21487 standard terminology "stringize"/"stringizing" throughout.
21488 * doc/cppinternals.texi: Likewise.
21490 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
21492 * doc/extend.texi: Fix some spelling mistakes and typos.
21493 * doc/invoke.texi: Likewise.
21495 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21498 * params.def (inline-min-speedup) Change from 10 to 8.
21500 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21502 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
21505 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
21508 * ipa-inline-analysis.c (get_minimal_bb): New function.
21509 (record_modified): Use it.
21510 (remap_edge_change_prob): Handle also ancestor functions.
21512 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
21514 * doc/contrib.texi (Contributors): Remove broken link into
21515 the Mauve CVS repository.
21517 2017-02-11 Jakub Jelinek <jakub@redhat.com>
21519 PR middle-end/79454
21520 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
21521 result computation whenever lhs doesn't have vector mode, not
21522 just when it has BLKmode.
21524 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21526 * doc/makefile.texi (profiledbootstrap): Refer to the
21527 installation instructions only in textual form.
21529 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21532 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
21534 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
21536 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
21537 (Specific): Update mingw-w64 reference.
21539 (Specific): Remove broken link to Renesas RX processor.
21541 2017-02-10 Richard Biener <rguenther@suse.de>
21543 * toplev.c (process_options): Do not mention obsolete graphite
21544 options when printing sorry message about missing graphite support.
21545 Mention -floop-nest-optimize.
21547 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
21549 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
21550 (vtst_p16): Likewise.
21551 (vtstq_p8): Likewise.
21552 (vtstq_p16): Likewise.
21554 (vtstq_p64): Likewise.
21555 * config/arm/arm_neon.h (vgetq_lane_p64): New.
21556 (vset_lane_p64): New.
21557 (vsetq_lane_p64): New.
21559 2017-02-10 Jakub Jelinek <jakub@redhat.com>
21561 PR tree-optimization/79411
21562 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
21563 stmt operands are SSA_NAMEs used in abnormal phis.
21564 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
21567 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21570 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
21573 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
21575 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
21577 2017-02-09 Jakub Jelinek <jakub@redhat.com>
21579 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
21583 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
21584 not arbitrary TREE_CONSTANT.
21587 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
21588 "omp declare target link" attribute unless is_global_var.
21589 * omp-offload.c (find_link_var_op): Likewise.
21591 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
21592 Chung-Lin Tang <cltang@codesourcery.com>
21594 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
21596 (gimplify_adjust_omp_clauses): Don't delete TILE.
21597 (gimplify_omp_for): Deal with TILE.
21598 * internal-fn.c (expand_GOACC_TILE): New function.
21599 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
21601 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
21602 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
21604 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
21605 avoid DIV for outermost collapse var.
21606 (expand_oacc_for): Insert tile element loop as needed. Adjust.
21607 Remove out of date comments, fix whitespace.
21608 * omp-general.c (omp_extract_for_data): Deal with tiling.
21609 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
21610 adjust OLF_DIM_BASE value.
21611 (struct omp_for_data): Add tiling field.
21612 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
21613 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
21614 for auto loops. Remove default auto determining, moved to
21615 oacc_loop_fixed_partitions.
21616 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
21617 stmts, add e_mask field.
21618 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
21619 (oacc_thread_numbers): Use oacc_dim_call.
21620 (oacc_xform_tile): New.
21621 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
21622 (finish_oacc_loop): Adjust for ifns vector.
21623 (oacc_loop_discover_walk): Append loop abstraction sites to list,
21624 add case for GOACC_TILE fns.
21625 (oacc_loop_xform_loop): Delete.
21626 (oacc_loop_process): Iterate over call list directly, and add
21627 handling for GOACC_TILE fns.
21628 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
21630 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
21631 vector partitioning to outer loops. Assign 2 partitions to loops
21632 when available. Add TILE handling.
21633 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
21634 (execite_oacc_device_lower): Process GOACC_TILE fns,
21635 ignore unknown specs.
21636 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
21637 * tree.c (omp_clause_num_ops): Adjust TILE ops.
21638 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
21640 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
21642 * configure.ac (ACX_BUGURL): Update.
21643 * configure: Regenerate.
21645 2017-02-09 Richard Biener <rguenther@suse.de>
21647 PR tree-optimization/69823
21648 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
21649 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
21651 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
21653 * config/arc/arc-c.def: Add __NPS400__ definition.
21654 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
21655 (TARGET_NPS400): Define.
21657 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
21659 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
21661 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
21662 pointer, arch_info.
21663 (arc_cpu_types): Fill the arch_info field with a pointer into the
21664 arc_arch_types table.
21665 (arc_selected_cpu): Declare.
21666 * config/arc/arc.c (arc_selected_cpu): Make global.
21667 (arc_selected_arch): Delete.
21668 (arc_base_cpu): Delete.
21669 (arc_override_options): Remove references to deleted variables,
21670 update access to arch information.
21671 (ARC_OPT): Update access to arch information.
21672 (ARC_OPTX): Likewise.
21673 * config/arc/arc.h (arc_base_cpu): Remove declaration.
21674 (TARGET_ARC600): Update access to arch information.
21675 (TARGET_ARC601): Likewise.
21676 (TARGET_ARC700): Likewise.
21677 (TARGET_EM): Likewise.
21678 (TARGET_HS): Likewise.
21679 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
21682 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
21685 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
21686 condition/operands for integer GE/LE/GEU/LEU operations.
21688 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
21690 PR translation/79397
21691 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
21694 2017-02-08 Martin Jambor <mjambor@suse.cz>
21697 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
21698 whether allocation happened.
21699 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
21700 nothing was allocated.
21702 2017-02-08 Jakub Jelinek <jakub@redhat.com>
21704 PR tree-optimization/79408
21705 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
21706 constant, but SSA_NAME with a known integer range, use the minimum
21707 of that range instead of op1 to determine if modulo can be replaced
21708 with its first operand.
21710 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21712 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
21714 2017-02-08 Richard Biener <rguenther@suse.de>
21716 PR tree-optimization/71824
21717 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
21718 Check all loops contained in the merged region.
21720 2017-02-07 Andrew Pinski <apinski@cavium.com>
21722 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
21724 2017-02-07 Andrew Pinski <apinski@cavium.com>
21726 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
21727 (thunderxt88): Likewise.
21728 (thunderxt81): Disable LSE and change v8.1 to v8.
21729 (thunderxt83): Likewise.
21731 2017-02-07 Jakub Jelinek <jakub@redhat.com>
21732 Richard Biener <rguenther@suse.de>
21734 PR middle-end/79399
21735 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
21736 type from int to size_t.
21737 * ira-costs.c (struct_costs_size): Change type from int to size_t.
21739 2017-02-07 Jakub Jelinek <jakub@redhat.com>
21741 PR rtl-optimization/79386
21742 * cprop.c (bypass_conditional_jumps): Initialize
21743 bypass_last_basic_block already before splitting bbs after
21744 unconditional traps...
21745 (bypass_conditional_jumps): ... rather than here.
21748 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
21749 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
21750 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
21751 fix -masm=intel patterns.
21753 2017-02-07 Richard Biener <rguenther@suse.de>
21755 PR tree-optimization/79256
21756 PR middle-end/79278
21757 * builtins.c (get_object_alignment_2): Use min_align_of_type
21758 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
21759 and ADJUST_FIELD_ALIGN.
21761 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
21763 * doc/tm.texi: Regenerate.
21764 * stor-layout.c (layout_decl): Adjust.
21765 (update_alignment_for_field): Likewise.
21766 (place_field): Likewise.
21767 (min_align_of_type): Likewise.
21768 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
21769 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
21770 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
21771 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
21772 * config/frv/frv.c (frv_adjust_field_align): Likewise.
21773 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
21774 * config/i386/i386.c (x86_field_alignment): Likewise.
21775 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
21776 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
21777 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
21778 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
21779 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
21780 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
21784 2017-01-30 Richard Biener <rguenther@suse.de>
21786 PR tree-optimization/79256
21787 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
21788 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
21791 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
21793 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
21794 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
21795 builtins to SImode and emit a zero-extend, if necessary.
21797 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21799 * docs/invoke.texi (RISC-V Options): Alphabetize.
21801 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21803 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
21806 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
21808 * config/riscv/riscv.c: New file.
21809 * common/config/riscv/riscv-common.c: Likewise.
21810 * config.gcc: Likewise.
21811 * config/riscv/constraints.md: Likewise.
21812 * config/riscv/elf.h: Likewise.
21813 * config/riscv/generic.md: Likewise.
21814 * config/riscv/linux.h: Likewise.
21815 * config/riscv/multilib-generator: Likewise.
21816 * config/riscv/peephole.md: Likewise.
21817 * config/riscv/pic.md: Likewise.
21818 * config/riscv/predicates.md: Likewise.
21819 * config/riscv/riscv-builtins.c: Likewise.
21820 * config/riscv/riscv-c.c: Likewise.
21821 * config/riscv/riscv-ftypes.def: Likewise.
21822 * config/riscv/riscv-modes.def: Likewise.
21823 * config/riscv/riscv-opts.h: Likewise.
21824 * config/riscv/riscv-protos.h: Likewise.
21825 * config/riscv/riscv.h: Likewise.
21826 * config/riscv/riscv.md: Likewise.
21827 * config/riscv/riscv.opt: Likewise.
21828 * config/riscv/sync.md: Likewise.
21829 * config/riscv/t-elf-multilib: Likewise.
21830 * config/riscv/t-linux: Likewise.
21831 * config/riscv/t-linux-multilib: Likewise.
21832 * config/riscv/t-riscv: Likewise.
21833 * configure.ac: Likewise.
21834 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
21835 Waterman as RISC-V maintainers.
21836 * doc/install.texi: Add RISC-V entries.
21837 * doc/invoke.texi: Add RISC-V options section.
21838 * doc/md.texi: Add RISC-V constraints section.
21839 * configure: Regenerated.
21841 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
21844 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
21845 false values to be constant vectors with all 0 or all 1 bits set.
21846 (vcondu<mode><mode>): Likewise.
21847 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
21849 (fpmask_comparison_operator): Update comment.
21850 (vecint_comparison_operator): New predicate.
21851 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
21852 vector conditionals when the true and false values are constant
21853 vectors with all 0 bits or all 1 bits set.
21855 2017-02-06 Martin Sebor <msebor@redhat.com>
21857 PR tree-optimization/79376
21858 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
21860 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
21862 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
21863 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
21864 to simplify split condition.
21866 2017-02-06 Jakub Jelinek <jakub@redhat.com>
21868 * omp-expand.c (oxpand_omp_atomic_fetch_op,
21869 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
21872 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
21874 PR rtl-optimization/68664
21875 * target.def (can_speculate_insn): New hook.
21876 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
21877 * doc/tm.texi: Regenerate.
21878 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
21879 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
21880 (rs6000_sched_can_speculate_insn): New function.
21882 2017-02-06 Jakub Jelinek <jakub@redhat.com>
21884 PR tree-optimization/79284
21885 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
21886 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
21887 vectorizable_mask_load_store, vectorizable_operation,
21888 vect_is_simple_cond, get_same_sized_vectype): Use it instead
21889 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
21890 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
21891 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
21892 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
21893 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
21894 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
21895 is_gimple_assign (stmt). Replace another such test with
21896 is_gimple_assign (stmt).
21898 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
21901 * config/avr/avr.c (rtl-iter.h): Include it.
21902 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
21903 (avr_legitimate_combined_insn): ...and implementation.
21905 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
21907 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
21908 * config/s390/s390.c (s390_const_operand_ok)
21909 (s390_canonicalize_comparison, s390_extract_part)
21910 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
21911 (s390_contiguous_bitmask_p, s390_rtx_costs)
21912 (legitimize_pic_address): Likewise.
21913 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
21914 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
21915 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
21916 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
21917 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
21919 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
21921 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
21922 REGNO($0) == REGNO($1).
21924 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21926 * config/s390/linux.h(SIZE_TYPE): Add comment.
21928 2017-02-06 Julian Brown <julian@codesourcery.com>
21929 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21930 Virendra Pathak <virendra.pathak@broadcom.com>
21932 * config/aarch64/aarch64-cores.def: Change the scheduler
21934 * config/aarch64/aarch64.md: Include thunderx2t99.md.
21935 * config/aarch64/thunderx2t99.md: New file.
21937 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
21939 * doc/standards.texi (Go Language): Update link to language
21942 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
21944 * tree-eh.c (lower_resx): Sanitize profile.
21945 (cleanup_empty_eh_move_lp): Likewise.
21947 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
21950 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
21952 * cfgloopmanip.h (loop_version): Update prototype.
21953 * modulo-sched.c (sms_schedule): Update call of loop_version.
21954 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
21955 * tree-parloops.c (gen_parallel_loop): Likewise.
21956 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
21957 * tree-ssa-loop-split.c (split_loop): Likewise.
21958 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
21959 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
21961 2017-02-05 Martin Liska <mliska@suse.cz>
21964 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
21966 (print_operand_address): Initialize a struct to zero.
21968 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
21970 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
21971 garbage collector only in textual form.
21973 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
21975 * doc/extend.texi (x86 specific memory model extensions for
21976 transactional memory): Simplify a phrase.
21978 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
21981 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
21982 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
21983 (atomic_storedi_1): Likewise.
21985 2017-02-04 Jakub Jelinek <jakub@redhat.com>
21987 PR tree-optimization/79338
21988 * tree-parloops.c (gather_scalar_reductions): Don't call
21989 vect_analyze_loop_form for loop->inner before destroying loop's
21992 2017-02-03 Martin Sebor <msebor@redhat.com>
21994 PR tree-optimization/79327
21995 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
21996 when precision has resulted in leading zeros.
21997 (format_integer): Adjust the likely counter to assume an unknown
21998 argument that may be zero is non-zero.
22000 2017-02-03 Jason Merrill <jason@redhat.com>
22003 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
22004 avoid copying non-taken branch.
22006 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22008 PR tree-optimization/79340
22009 * tree-vect-loop.c (vectorizable_reduction): Release
22010 vec_defs elements after safe_splicing them into other vectors.
22013 PR tree-optimization/79327
22014 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
22015 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
22017 (format_integer): Use wide_int_to_tree instead of build_int_cst
22018 + to_?hwi. If argmin is NULL, just set argmin and argmax to
22019 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
22020 of shortest and longest sequence.
22022 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
22024 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
22025 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
22027 2017-02-03 Walter Lee <walt@tilera.com>
22030 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
22031 after initial stackframe link reg save.
22032 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
22034 2017-02-03 Jakub Jelinek <jakub@redhat.com>
22037 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
22038 wu for stxssp alternative.
22040 2017-02-03 Martin Sebor <msebor@redhat.com>
22042 PR tree-optimization/79352
22043 * gimple-fold.c (get_range_strlen): Add argument.
22044 (get_range_strlen): Change return type to bool.
22045 (get_maxval_strlen): Pass in a dummy argument.
22046 * gimple-fold.h (get_range_strlen): Change return type to bool.
22047 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
22048 * tree.h (array_at_struct_end_p): Add argument.
22049 * tree.c (array_at_struct_end_p): Handle it.
22051 2017-02-03 Martin Liska <mliska@suse.cz>
22054 * multiple_target.c (create_dispatcher_calls): Redirect edge
22055 from a caller of a dispatcher.
22056 (expand_target_clones): Make the clones local.
22057 (ipa_target_clone): Do both target clones and resolvers.
22058 (ipa_dispatcher_calls): Remove the pass.
22059 (pass_dispatcher_calls::gate): Likewise.
22060 (make_pass_dispatcher_calls): Likewise.
22061 * passes.def (pass_target_clone): Put as very first IPA early
22064 2017-02-03 Martin Liska <mliska@suse.cz>
22066 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
22067 in case of a function with ifunc attribute.
22069 2017-02-03 Martin Liska <mliska@suse.cz>
22071 * cgraph.c (cgraph_node::dump): Dump function version info.
22072 * symtab.c (symtab_node::dump_base): Add missing new line.
22074 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22076 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
22077 (ifcombine_ifandif): Use it.
22079 2017-02-03 Martin Liska <mliska@suse.cz>
22081 * doc/invoke.texi: Document default value for
22082 use-after-scope-direct-emission-threshold.
22084 2017-02-03 Martin Liska <mliska@suse.cz>
22086 PR tree-optimization/79339
22087 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
22088 (format_floating): Likewise.
22090 2017-02-03 Martin Liska <mliska@suse.cz>
22093 * ipa-prop.c (ipa_node_params_t::insert): Remove current
22095 (ipa_node_params_t::remove): Likewise.
22096 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
22097 initialization from removed ipa_node_params_t::insert.
22098 (ipa_node_params::~ipa_node_params): Move from removed
22099 ipa_node_params_t::release.
22100 * symbol-summary.h (symbol_summary::m_released): New member.
22101 Do not release a summary twice. Do not allow to call finalizer
22102 for types of a summary that live in GGC memory.
22104 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
22106 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
22109 2017-02-02 Martin Sebor <msebor@redhat.com>
22111 PR middle-end/79275
22112 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
22113 (format_string): Tighten up the range of output for non-constant
22114 strings and correct the expected range for wide non-constant strings.
22116 2017-02-02 Martin Sebor <msebor@redhat.com>
22118 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
22120 PR middle-end/32003
22121 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
22123 (-fdump-tree-@var): Add to index and document how to come up
22124 with pass-specific option and dump file names.
22125 (-fdump-passes): Clarify where to look for output.
22127 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
22129 PR middle-end/77445
22130 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
22131 statistics of the analyzed path; allow threading for speed when
22132 any of BBs along the path are optimized for speed.
22134 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
22136 PR middle-end/78468
22137 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
22138 settings of the virtual registers.
22141 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
22143 * explow.c (get_dynamic_stack_size): Take known alignment of stack
22144 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
22147 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22149 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
22150 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
22152 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22154 * config/s390/s390.md: Add missing comments with the expanded
22156 * config/s390/vector.md: Likewise.
22157 * config/s390/vx-builtins.md: Likewise.
22159 2017-02-02 Jakub Jelinek <jakub@redhat.com>
22162 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
22163 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
22164 conditions on a single line.
22166 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22168 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22169 __S390_VX__ to __VX__.
22171 2017-02-01 Andrew Pinski <apinski@cavium.com>
22173 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
22174 stmt_info to record_stmt_cost.
22175 (vect_get_known_peeling_cost): Pass stmt_info if known to
22177 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
22178 cpu_vector_cost field into
22179 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
22180 field into vec_int_stmt_cost and vec_fp_stmt_cost.
22181 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
22182 splitting of scalar_stmt_cost and vec_stmt_cost.
22183 (thunderx_vector_cost): Likewise.
22184 (cortexa57_vector_cost): LIkewise.
22185 (exynosm1_vector_cost): Likewise.
22186 (xgene1_vector_cost): Likewise.
22187 (thunderx2t99_vector_cost): Improve after the splitting of the two
22189 (aarch64_builtin_vectorization_cost): Update for the splitting of
22190 scalar_stmt_cost and vec_stmt_cost.
22192 2017-02-01 Torvald Riegel <triegel@redhat.com>
22193 Richard Henderson <rth@redhat.com>
22195 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
22196 conditional on existance of a fast atomic load.
22197 * optabs-query.c (can_atomic_load_p): New function.
22198 * optabs-query.h (can_atomic_load_p): Declare it.
22199 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
22200 no fast atomic load is available for the particular size of access.
22201 (expand_atomic_compare_and_swap): Likewise.
22202 (expand_atomic_load): Likewise.
22203 (expand_atomic_store): Likewise.
22204 (expand_atomic_fetch_op): Likewise.
22205 * testsuite/lib/target-supports.exp
22206 (check_effective_target_sync_int_128): Remove x86 because it provides
22207 no fast atomic load.
22208 (check_effective_target_sync_int_128_runtime): Likewise.
22210 2017-02-01 Richard Biener <rguenther@suse.de>
22212 * graphite.c: Include tree-vectorizer.h for find_loop_location.
22213 (graphite_transform_loops): Provide opt-info for optimized nests.
22214 * tree-parloop.c (parallelize_loops): Provide opt-info for
22215 parallelized loops.
22217 2017-02-01 Richard Biener <rguenther@suse.de>
22219 PR middle-end/79315
22220 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
22221 was not set before.
22223 2017-02-01 Richard Biener <rguenther@suse.de>
22225 PR tree-optimization/71824
22226 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
22227 Verify the loops are valid in the merged SESE region.
22228 (scop_detection::can_represent_loop_1): Check analyzing the
22229 evolution of the number of iterations in the region succeeds.
22231 2017-01-31 Ian Lance Taylor <iant@golang.org>
22233 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
22234 REG_ARGS_SIZE note to 32-bit push insns and call insn.
22236 2017-01-31 David Malcolm <dmalcolm@redhat.com>
22238 PR preprocessor/79210
22239 * input.c (get_substring_ranges_for_loc): Replace line_width
22240 assertion with error-handling.
22242 2017-01-31 Richard Biener <rguenther@suse.de>
22244 PR tree-optimization/77318
22245 * graphite-sese-to-poly.c (extract_affine): Fix assert.
22246 (create_pw_aff_from_tree): Take loop parameter.
22247 (add_condition_to_pbb): Pass loop of the condition to
22248 create_pw_aff_from_tree.
22250 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22252 * config/s390/s390.c (s390_asan_shadow_offset): New function.
22253 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
22255 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
22259 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
22261 (convert_int_to_float128): Likewise.
22262 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
22263 (convert_int_to_float128): Likewise.
22264 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
22265 (UNSPEC_IEEE128_CONVERT): Likewise.
22266 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
22267 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
22268 Use local variables for IBM extended format.
22269 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
22270 (fix_trunc<mode>si2_fprs): Likewise.
22271 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
22272 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
22273 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
22274 to know that we can now have integers of all sizes in vector
22276 (fix<uns>_<mode>di2_hw): Likewise.
22277 (float<uns>_<mode>si2_hw): Likewise.
22278 (fix_<mode>si2_hw): Likewise.
22279 (fixuns_<mode>si2_hw): Likewise.
22280 (float<uns>_<mode>di2_hw): Likewise.
22281 (float_<mode>di2_hw): Likewise.
22282 (float_<mode>si2_hw): Likewise.
22283 (floatuns_<mode>di2_hw): Likewise.
22284 (floatuns_<mode>si2_hw): Likewise.
22285 (xscvqp<su>wz_<mode>): Delete, no longer used.
22286 (xscvqp<su>dz_<mode>): Likewise.
22287 (xscv<su>dqp_<mode>): Likewise.
22288 (ieee128_mfvsrd_64bit): Likewise.
22289 (ieee128_mfvsrd_32bit): Likewise.
22290 (ieee128_mfvsrwz): Likewise.
22291 (ieee128_mtvsrw): Likewise.
22292 (ieee128_mtvsrd_64bit): Likewise.
22293 (ieee128_mtvsrd_32bit): Likewise.
22295 2017-01-31 Martin Liska <mliska@suse.cz>
22298 * ipa-prop.c (ipa_free_all_node_params): Call release method
22299 instead of ~sumbol_summary to not to trigger double times
22302 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
22304 PR tree-optimization/71691
22305 * bitmap.h (class auto_bitmap): New.
22306 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
22307 is_maybe_undefined instead of ssa_undefined_value_p.
22309 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22311 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
22312 __S390_ARCH_LEVEL__ to __ARCH__.
22314 2017-01-31 Jakub Jelinek <jakub@redhat.com>
22316 PR tree-optimization/79267
22317 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
22318 if should_remove_lhs_p is true.
22320 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
22323 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
22324 (add_alignment_attribute): New.
22325 (base_type_die): Add alignment attribute.
22326 (subrange_type_die): Likewise.
22327 (modified_type_die): Likewise.
22328 (gen_array_type_die): Likewise.
22329 (gen_descr_array_type_die: Likewise.
22330 (gen_enumeration_type_die): Likewise.
22331 (gen_subprogram_die): Likewise.
22332 (gen_variable_die): Likewise.
22333 (gen_field_die): Likewise.
22334 (gen_ptr_to_mbr_type_die): Likewise.
22335 (gen_struct_or_union_type_die): Likewise.
22336 (gen_subroutine_type_die): Likewise.
22337 (gen_typedef_die): Likewise.
22338 (base_type_cmp): Compare alignment attribute.
22340 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22343 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
22344 (setb_unsigned) New pattern for setb with CCUNS.
22345 * config/rs6000/rs6000.c (expand_block_compare): Use a different
22346 subfc./subfe sequence to avoid overflow problems. Generate a
22347 shorter sequence with cmpld/setb for power9.
22348 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
22349 for generating subfc. instruction.
22350 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
22351 now uses this instruction.
22353 2017-01-30 Ian Lance Taylor <iant@google.com>
22356 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
22357 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
22359 2017-01-30 Martin Sebor <msebor@redhat.com>
22361 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
22362 Move constant to the right of a relational operator.
22363 (get_mpfr_format_length, format_character, format_string): Ditto.
22364 (should_warn_p, maybe_warn): Same.
22366 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
22368 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
22371 * asan.c (get_translation_unit_decl): Remove function.
22372 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
22374 2017-01-30 Martin Liska <mliska@suse.cz>
22376 PR gcov-profile/79259
22377 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
22378 -fprofile-generate.
22380 2017-01-30 Martin Liska <mliska@suse.cz>
22383 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
22384 Initialize variables with NULL value.
22386 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
22389 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
22391 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
22393 2017-01-30 Richard Biener <rguenther@suse.de>
22395 PR tree-optimization/79276
22396 * tree-vrp.c (process_assert_insertions): Properly adjust common
22397 when removing a duplicate.
22399 * gcc.dg/torture/pr79276.c: New testcase.
22401 2017-01-30 Richard Biener <rguenther@suse.de>
22403 PR tree-optimization/79256
22404 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
22405 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
22407 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
22409 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
22412 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
22413 ("*r<noxa>sbg_<mode>_sll_bitmask")
22414 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
22415 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
22416 Use contiguous_bitmask_nowrap_operand.
22418 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22421 * config/rs6000/altivec.h (vec_xl): Revise #define.
22422 (vec_xst): Likewise.
22424 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
22426 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
22428 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
22430 PR rtl-optimization/79194
22431 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
22432 traps before call to bypass_conditional_jumps.
22434 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22436 PR tree-optimization/71374
22437 * lra-constraints.c (check_conflict_input_operands): New.
22438 (match_reload): Use it.
22440 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
22443 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
22444 account to calculate conflict_set.
22446 2017-01-27 Bin Cheng <bin.cheng@arm.com>
22448 PR rtl-optimization/78559
22449 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
22450 other_insn in combine.
22452 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
22454 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
22455 uint16_type_node for BT_UINT16.
22457 2017-01-27 David Malcolm <dmalcolm@redhat.com>
22459 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
22460 "RTL Tests" to menu.
22461 (GIMPLE Tests): New node.
22462 (RTL Tests): New node.
22464 2017-01-27 Richard Biener <rguenther@suse.de>
22466 PR tree-optimization/79245
22467 * tree-loop-distribution.c (distribute_loop): Apply cost
22468 modeling also to detected patterns.
22470 2017-01-27 Richard Biener <rguenther@suse.de>
22472 PR tree-optimization/71433
22473 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
22474 (compare_assert_loc): New function.
22475 (process_assert_insertions): Sort and optimize assert locations
22476 to remove duplicates and push down identical assertions on
22477 edges to their destination block.
22479 2017-01-27 Richard Biener <rguenther@suse.de>
22481 PR tree-optimization/79244
22482 * tree-vrp.c (remove_range_assertions): Forcefully propagate
22483 out SSA names even if abnormal.
22485 2017-01-27 Jakub Jelinek <jakub@redhat.com>
22487 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
22488 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
22489 instead of MPFR_RNDN.
22491 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
22494 * arm.c (arm_option_override): Don't call build_target_option_node
22495 until after doing all option overrides.
22496 (arm_valid_target_attribute_tree): Likewise.
22498 2017-01-27 Martin Liska <mliska@suse.cz>
22500 * doc/invoke.texi (-fprofile-arcs): Document profiling support
22501 for {cd}tors and C++ {cd}tors.
22503 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
22505 * config/s390/s390.md ("*setmem_long_and")
22506 ("*setmem_long_and_31z"): Use zero_extend instead of and.
22508 2017-01-26 Martin Sebor <msebor@redhat.com>
22510 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
22513 2017-01-26 Martin Sebor <msebor@redhat.com>
22515 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
22516 HAVE_DFmode before using XFmode or DFmode.
22517 (parse_directive): Avoid using the z length modifier to avoid
22518 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
22520 PR middle-end/78703
22521 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
22522 to accept adjustment as an array.
22523 (get_int_range): New function.
22524 (struct directive): Make width and prec arrays.
22525 (directive::set_width, directive::set_precision): Call get_int_range.
22526 (format_integer, format_floating): Handle width and precision ranges.
22527 (format_string, parse_directive): Same.
22529 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22532 * dwarf2out.c (generate_skeleton_bottom_up): For children with
22533 comdat_type_p set, just clone them, but keep the children in the
22537 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
22538 which have direct callers with -fvar-tracking-assignments enabled
22540 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
22541 inside of type units.
22543 2017-01-26 Martin Sebor <msebor@redhat.com>
22545 PR middle-end/78703
22546 * gimple-ssa-sprintf.c (struct result_range): Add likely and
22548 (struct format_result): Replace number_chars, number_chars_min,
22549 and number_chars_max with a single member of struct result_range.
22551 (format_result::operator+=): Adjust.
22552 (struct fmtresult): Remove bounded. Handle likely and unlikely
22554 (fmtresult::adjust_for_width_or_precision): New function.
22555 (fmtresult:type_max_digits): New function.
22556 (bytes_remaining): Handle likely and unlikely counters.
22557 (min_bytes_remaining): Remove.
22558 (format_percent): Simplify.
22559 (format_integer, format_floating): Set likely and unlikely counters.
22560 (get_string_length, format_character, format_string): Same.
22561 (format_plain, should_warn_p): New function.
22562 (maybe_warn): Call should_warn_p. Update diagnostic messages
22563 and handle those for all directives, including plain strings.
22564 (format_directive): Handle likely and unlikely counters.
22565 Remove unnecessary quoting from diagnostics. Add an informational
22567 (add_bytes): Remove.
22568 (pass_sprintf_length::compute_format_length): Simplify.
22569 (try_substitute_return_value): Handle likely and unlikely counters.
22571 2017-01-26 Carl Love <cel@us.ibm.com>
22573 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
22574 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
22576 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
22579 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
22580 endianess for subregs into account.
22581 * lra-constraints.c (lra_constraints): Do risky transformations
22582 always on the first iteration.
22583 * lra-lives.c (check_pseudos_live_through_calls): Add arg
22584 last_call_used_reg_set.
22585 (process_bb_lives): Define and use last_call_used_reg_set.
22586 * lra.c (lra): Always continue after lra_constraints on the first
22589 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
22591 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
22593 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
22595 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22597 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
22598 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
22599 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
22600 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
22601 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
22602 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
22603 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
22604 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
22605 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
22607 2017-01-26 Marek Polacek <polacek@redhat.com>
22610 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
22611 for the third operand.
22613 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22615 PR middle-end/79236
22616 * omp-low.c (struct omp_context): Add simt_stmt field.
22617 (scan_omp_for): Return omp_context *.
22618 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
22619 context to the _simt_ SIMD stmt.
22620 (lower_omp_for): For combined SIMD with sibling _simt_
22621 SIMD, make sure to use the same decls in _looptemp_
22622 clauses as in the sibling.
22624 2017-01-26 David Sherwood <david.sherwood@arm.com>
22626 PR middle-end/79212
22627 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
22630 2017-01-26 Jakub Jelinek <jakub@redhat.com>
22633 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
22634 emit fld b; fld a; if possible.
22636 * brig-builtins.def: Update copyright years.
22637 * config/arm/arm_acle_builtins.def: Update copyright years.
22639 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
22642 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
22643 constraint instead of o for the stxsd instruction.
22645 2017-01-25 Carl Love <cel@us.ibm.com>
22647 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
22648 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
22650 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
22652 * doc/invoke.texi (C++ Dialect Options): Fix typo.
22654 2017-01-25 Richard Biener <rguenther@suse.de>
22656 PR tree-optimization/69264
22657 * target.def (vector_alignment_reachable): Improve documentation.
22658 * doc/tm.texi: Regenerate.
22659 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
22661 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
22662 earlier changes with respect to TYPE_USER_ALIGN.
22663 (vector_alignment_reachable_p): Likewise. Improve dumping.
22665 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22668 * config/arm/arm.md (xordi3): Force constant operand into a register
22671 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22673 * doc/invoke.texi (-fstore-merging): Correct default optimization
22674 levels at which it is enabled.
22675 (-O): Move -fstore-merging from list to...
22678 2017-01-25 Richard Biener <rguenther@suse.de>
22681 * omp-expand.c: Include debug.h.
22682 (expand_omp_taskreg): Make sure to generate early debug before
22683 outlining anything from a function.
22684 (expand_omp_target): Likewise.
22685 (grid_expand_target_grid_body): Likewise.
22687 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
22690 * asan.c (get_translation_unit_decl): New function.
22691 (asan_add_global): Extract modules file name from globals
22692 TRANSLATION_UNIT_DECL name.
22694 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
22697 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
22698 for long calls with APCS frame and VFP.
22700 2017-01-24 David Malcolm <dmalcolm@redhat.com>
22702 * cfg.c (original_copy_tables_initialized_p): New function.
22703 * cfg.h (original_copy_tables_initialized_p): New decl.
22704 * cfgrtl.c (relink_block_chain): Guard the call to
22705 free_original_copy_tables with a call to
22706 original_copy_tables_initialized_p.
22707 * cgraph.h (symtab_node::native_rtl_p): New decl.
22708 * cgraphunit.c (symtab_node::native_rtl_p): New function.
22709 (symtab_node::needed_p): Don't assert for early assembly output
22710 for __RTL functions.
22711 (cgraph_node::finalize_function): Set "force_output" for __RTL
22713 (cgraph_node::analyze): Bail out early for __RTL functions.
22714 (analyze_functions): Update assertion to support __RTL functions.
22715 (cgraph_node::expand): Bail out early for __RTL functions.
22716 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
22718 * function.h (struct function): Update comment for field
22720 * gimple-expr.c: Include "tree-pass.h".
22721 (gimple_has_body_p): Return false for __RTL functions.
22722 * Makefile.in (OBJS): Add run-rtl-passes.o.
22723 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
22725 (gcc::pass_manager::get_clean_slate): New accessor.
22726 * passes.c: Include "insn-addr.h".
22727 (should_skip_pass_p): Add logging. Update logic for running
22728 "expand" to be compatible with both __GIMPLE and __RTL. Guard
22729 property-provider override so it is only done for gimple passes.
22731 (skip_pass): New function.
22732 (execute_one_pass): Call skip_pass when skipping passes.
22733 * read-md.c (md_reader::read_char): Support filtering
22734 the input to a subset of line numbers.
22735 (md_reader::md_reader): Initialize fields
22736 m_first_line and m_last_line.
22737 (md_reader::read_file_fragment): New function.
22738 * read-md.h (md_reader::read_file_fragment): New decl.
22739 (md_reader::m_first_line): New field.
22740 (md_reader::m_last_line): New field.
22741 * read-rtl-function.c (function_reader::create_function): Only
22742 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
22743 curr_properties. Set DECL_INITIAL to a dummy block.
22744 (read_rtl_function_body_from_file_range): New function.
22745 * read-rtl-function.h (read_rtl_function_body_from_file_range):
22747 * run-rtl-passes.c: New file.
22748 * run-rtl-passes.h: New file.
22750 2017-01-24 Jeff Law <law@redhat.com>
22752 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
22755 2017-01-24 Bin Cheng <bin.cheng@arm.com>
22757 PR tree-optimization/79159
22758 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
22759 (record_nonwrapping_iv): Improve boundary using above function if no
22760 value range information.
22762 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
22763 Martin Jambor <mjambor@suse.cz>
22765 * brig-builtins.def: New file.
22766 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
22767 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
22768 (DEF_HSAIL_SAT_BUILTIN): Likewise.
22769 (DEF_HSAIL_INTR_BUILTIN): Likewise.
22770 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
22771 * builtin-types.def (BT_INT8): New.
22772 (BT_INT16): Likewise.
22773 (BT_UINT8): Likewise.
22774 (BT_UINT16): Likewise.
22775 (BT_FN_ULONG): Likewise.
22776 (BT_FN_UINT_INT): Likewise.
22777 (BT_FN_UINT_ULONG): Likewise.
22778 (BT_FN_UINT_LONG): Likewise.
22779 (BT_FN_UINT_PTR): Likewise.
22780 (BT_FN_ULONG_PTR): Likewise.
22781 (BT_FN_INT8_FLOAT): Likewise.
22782 (BT_FN_INT16_FLOAT): Likewise.
22783 (BT_FN_UINT32_FLOAT): Likewise.
22784 (BT_FN_UINT16_FLOAT): Likewise.
22785 (BT_FN_UINT8_FLOAT): Likewise.
22786 (BT_FN_UINT64_FLOAT): Likewise.
22787 (BT_FN_UINT16_UINT32): Likewise.
22788 (BT_FN_UINT32_UINT16): Likewise.
22789 (BT_FN_UINT16_UINT16_UINT16): Likewise.
22790 (BT_FN_INT_PTR_INT): Likewise.
22791 (BT_FN_UINT_PTR_UINT): Likewise.
22792 (BT_FN_LONG_PTR_LONG): Likewise.
22793 (BT_FN_ULONG_PTR_ULONG): Likewise.
22794 (BT_FN_VOID_UINT64_UINT64): Likewise.
22795 (BT_FN_UINT8_UINT8_UINT8): Likewise.
22796 (BT_FN_INT8_INT8_INT8): Likewise.
22797 (BT_FN_INT16_INT16_INT16): Likewise.
22798 (BT_FN_INT_INT_INT): Likewise.
22799 (BT_FN_UINT_FLOAT_UINT): Likewise.
22800 (BT_FN_FLOAT_UINT_UINT): Likewise.
22801 (BT_FN_ULONG_UINT_UINT): Likewise.
22802 (BT_FN_ULONG_UINT_PTR): Likewise.
22803 (BT_FN_ULONG_ULONG_ULONG): Likewise.
22804 (BT_FN_UINT_UINT_UINT): Likewise.
22805 (BT_FN_VOID_UINT_PTR): Likewise.
22806 (BT_FN_UINT_UINT_PTR: Likewise.
22807 (BT_FN_UINT32_UINT64_PTR): Likewise.
22808 (BT_FN_INT_INT_UINT_UINT): Likewise.
22809 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
22810 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
22811 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
22812 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
22813 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
22814 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
22815 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
22816 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
22817 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
22818 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
22819 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
22820 * doc/frontends.texi: List BRIG FE.
22821 * doc/install.texi (Testing): Add BRIG tesring requirements.
22822 * doc/invoke.texi (Overall Options): Mention BRIG.
22823 * doc/standards.texi (Standards): Doucment BRIG HSA version.
22825 2017-01-24 Richard Biener <rguenther@suse.de>
22827 PR translation/79208
22828 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
22830 2017-01-24 Martin Jambor <mjambor@suse.cz>
22833 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
22834 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
22835 and known_contexts.
22837 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
22839 PR middle-end/79123
22840 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
22841 casts from signed to unsigned really don't have a range.
22843 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
22845 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
22846 GMP_RNDx for compatiblity.
22848 2017-01-24 Martin Liska <mliska@suse.cz>
22851 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
22852 that would prevent us to call alloca with -1 as argument.
22854 2017-01-24 Jakub Jelinek <jakub@redhat.com>
22856 * dwarf2out.c (output_compilation_unit_header, output_file_names):
22857 Avoid -Wformat-security warning.
22859 2017-01-23 Andrew Pinski <apinski@cavium.com>
22861 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
22864 2017-01-23 Martin Sebor <msebor@redhat.com>
22866 PR middle-end/78703
22867 * gimple-ssa-sprintf.c (warn_level): New global.
22868 (format_integer): Use it here and throughout the rest of the file.
22869 Use the same switch to compute sign as base.
22870 (maybe_warn): New function.
22871 (format_directive): Factor out warnings into maybe_warn.
22872 Add debugging output. Use warn_level.
22873 (add_bytes): Use warn_level.
22874 (pass_sprintf_length::compute_format_length): Add debugging output.
22875 (try_substitute_return_value): Same.
22876 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
22878 PR middle-end/78703
22879 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
22880 (struct fmtresult, format_integer, format_floating): Adjust.
22881 (fmtresult::fmtresult): Set max correctly in two argument ctor.
22882 (get_string_length, format_string,format_directive): Same.
22883 (pass_sprintf_length::compute_format_length): Same.
22884 (try_substitute_return_value): Simplify slightly.
22886 PR middle-end/78703
22887 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
22888 (fmtresult::operator+=): Outlined.
22889 (struct fmtresult): Add ctors.
22890 (struct conversion_spec): Rename...
22891 (struct directive): ...to this. Add and remove data members.
22892 (directive::set_width, directive::set_precision): New functions.
22893 (format_percent): Use fmtresult ctor.
22894 (get_width_and_precision): Remove.
22895 (format_integer): Make naming changes. Avoid computing width and
22897 (format_floating): Same. Adjust indentation.
22898 (format_character, format_none): New functions.
22899 (format_string): Moved character handling to format_character.
22900 (format_directive): Remove arguments, change return type.
22901 (parse_directive): New function.
22902 (pass_sprintf_length::compute_format_length): Move directive
22903 parsing to parse_directive.
22905 2017-01-23 Jakub Jelinek <jakub@redhat.com>
22907 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
22908 (assign_assembler_name_if_needed): ... this.
22909 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
22910 (assign_assembler_name_if_needed): ... this.
22911 (free_lang_data_in_cgraph): Adjust callers.
22912 * cgraphunit.c (cgraph_node::analyze): Likewise.
22913 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
22916 2017-01-23 Richard Biener <rguenther@suse.de>
22918 PR tree-optimization/79088
22919 PR tree-optimization/79188
22920 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
22921 resetting loop bounds after last path deletion. Reset loop
22922 bounds of the target loop, make code match the comments.
22923 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
22924 Make sure loops need no fixups.
22926 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
22928 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
22929 exponent support with double type for first argument.
22930 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
22931 type returned by __builtin_vec_extract_sig,
22932 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
22933 functions from "vector int" to "vector unsigned int" or from
22934 "vector long long int" to "vector unsigned long long int".
22935 Changed type returned by __builtin_vec_extract_exp,
22936 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
22937 functions from "vector int" to "vector unsigned int" or from
22938 "vector long long int" to "vector unsigned long long int".
22939 Changed return type of __builtin_vec_test_data_class,
22940 __builtin_vec_test_data_class_sp, and
22941 __builtin_vec_test_data_class_dp from "vector int" to
22942 "vector bool int" or from "vector long long int" to "vector bool
22943 long long int" and changed second argument type from "unsigned
22944 int" to "int". Added new overloaded function forms "vector float
22945 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
22946 "vector float __builtin_vec_insert_exp_sp (vector float, vector
22947 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
22948 double, vector unsigned long long int)" and "vector double
22949 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
22950 long int)". Changed return type of
22951 __builtin_scalar_test_data_class and
22952 __builtin_scalar_test_data_class_sp and
22953 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
22954 int" and changed second argument from "unsigned int" to "int".
22955 Changed type returned by __builtin_scalar_test_neg,
22956 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
22957 from "int" to "bool int". Added new overloaded function form
22958 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
22959 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
22960 exponent double-precision with floating point first argument.
22961 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
22962 documentation of scalar_test_data_class, scalar_test_neg,
22963 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
22964 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
22965 vec_test_data_class built-in functions to reflect refinements in
22966 their type signatures.
22968 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
22970 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
22972 (aarch64_elf_asm_destructor): Likewise.
22974 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
22976 PR rtl-optimization/78634
22977 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
22978 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
22979 * ifcvt.c (noce_try_cmove): Add missing cost check.
22981 PR rtl-optimization/71724
22982 * combine.c (if_then_else_cond): Look for situations where it is
22983 beneficial to undo the work of one of the recursive calls.
22985 2017-01-23 Bin Cheng <bin.cheng@arm.com>
22987 PR tree-optimization/70754
22988 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
22989 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
22990 combined stmt before it if not NULL.
22991 (combine_chains): Process refs reversely and compute dominance point
22994 2017-01-23 Martin Liska <mliska@suse.cz>
22996 PR tree-optimization/79196
22997 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
22998 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
23000 (strlen_optimize_stmt): Call the renamed function.
23002 2017-01-23 Michael Matz <matz@suse.de>
23004 PR tree-optimization/78384
23005 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
23007 2017-01-23 Richard Biener <rguenther@suse.de>
23009 PR tree-optimization/79186
23010 * tree-vrp.c (register_new_assert_for): Make sure we've seen
23011 both incoming edges before moving an assert.
23013 2017-01-23 Martin Jambor <mjambor@suse.cz>
23015 * ipa-prop.c (load_from_param_1): Removed.
23016 (load_from_unmodified_param): Bits from load_from_param_1 put back
23018 (load_from_param): Removed.
23019 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
23020 with stmt. Reverted back to use of load_from_unmodified_param.
23022 2017-01-23 Martin Jambor <mjambor@suse.cz>
23025 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
23026 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
23027 field a pointer to garbage collected vector, mark lattices and
23028 ipcp_orig_node with GTY((skip)).
23029 (ipa_get_param_count): Adjust to descriptors being a pointer.
23030 (ipa_get_param): Likewise.
23031 (ipa_get_type): Likewise.
23032 (ipa_get_param_move_cost): Likewise.
23033 (ipa_set_param_used): Likewise.
23034 (ipa_get_controlled_uses): Likewise.
23035 (ipa_set_controlled_uses): Likewise.
23036 (ipa_is_param_used): Likewise.
23037 (ipa_node_params_t): Move into garbage collector. New methods insert
23039 (ipa_node_params_sum): Annotate wth GTY(()).
23040 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
23042 (ipa_load_from_parm_agg): Adjust declaration.
23043 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
23044 * ipa-profile.c (ipa_profile): Likewise.
23045 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
23046 (ipa_populate_param_decls): Make descriptors parameter garbage
23048 (ipa_dump_param): Adjust to descriptors being a pointer.
23049 (ipa_alloc_node_params): Likewise.
23050 (ipa_initialize_node_params): Likewise.
23051 (load_from_param_1): Make descriptors parameter garbage collected.
23052 (load_from_unmodified_param): Likewise.
23053 (load_from_param): Likewise.
23054 (ipa_load_from_parm_agg): Likewise.
23055 (ipa_node_params::~ipa_node_params): Removed.
23056 (ipa_free_all_node_params): Remove call to delete operator.
23057 (ipa_node_params_t::insert): New.
23058 (ipa_node_params_t::remove): Likewise.
23059 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
23060 copy known_csts and known_contexts vectors.
23061 (ipa_read_node_info): Adjust to descriptors being a pointer.
23062 (ipcp_modif_dom_walker): Make m_descriptors field garbage
23064 (ipcp_transform_function): Make descriptors variable garbage
23067 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
23069 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
23070 * config/i386/avx512dqintrin.h: Ditto.
23071 * config/i386/avx512fintrin.h: Ditto.
23072 * config/i386/i386.c: Handle new builtins.
23073 * config/i386/i386-builtin.def: Add new builtins.
23074 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
23075 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
23077 2017-01-23 Jakub Jelinek <jakub@redhat.com>
23078 Martin Liska <mliska@suse.cz>
23080 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
23081 * asan.c (asan_expand_poison_ifn): Support stores and use
23082 appropriate ASAN report function.
23083 * internal-fn.c (expand_ASAN_POISON_USE): New function.
23084 * internal-fn.def (ASAN_POISON_USE): Declare.
23085 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
23086 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
23087 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
23088 ASAN_POISON calls w/o LHS.
23089 * tree-ssa.c (execute_update_addresses_taken): Create clobber
23090 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
23091 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
23092 * gimplify.c (asan_poison_variables): Add attribute
23093 use_after_scope_memory to variables that really needs to live
23095 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
23096 having the attribute.
23098 2017-01-23 Martin Liska <mliska@suse.cz>
23100 * asan.c (create_asan_shadow_var): New function.
23101 (asan_expand_poison_ifn): Likewise.
23102 * asan.h (asan_expand_poison_ifn): New declaration.
23103 * internal-fn.c (expand_ASAN_POISON): Likewise.
23104 * internal-fn.def (ASAN_POISON): New builtin.
23105 * sanopt.c (pass_sanopt::execute): Expand
23106 asan_expand_poison_ifn.
23107 * tree-inline.c (copy_decl_for_dup_finish): Make function
23109 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
23110 * tree-ssa.c (is_asan_mark_p): New function.
23111 (execute_update_addresses_taken): Rewrite local variables
23112 (identified just by use-after-scope as addressable) into SSA.
23114 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23116 * doc/install.texi (Specific): opensource.apple.com uses https
23117 now. Remove trailing slash.
23119 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
23121 * README.Portability: Remove note on an Irix compatibility issue.
23123 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
23125 * gcov.c (INCLUDE_ALGORITHM): Define.
23126 (INCLUDE_VECTOR): Define.
23127 No longer include <vector> and <algorithm> directly.
23129 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
23131 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
23133 * doc/invoke.texi (Code Gen Options): Ditto.
23135 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
23138 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
23140 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
23142 rtl-optimization/79125
23143 * cprop.c (local_cprop_pass): Handle cases where we make an
23144 unconditional trap.
23146 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
23150 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
23151 read from, for big endian.
23153 2017-01-20 Jiong Wang <jiong.wang@arm.com>
23155 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
23156 register pauth builtins for LP64 only.
23158 2017-01-20 Marek Polacek <polacek@redhat.com>
23161 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
23164 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23166 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
23168 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
23169 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
23170 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
23172 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23175 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
23176 in big-endian mode when they are not a single duplicated value.
23178 2017-01-20 Richard Biener <rguenther@suse.de>
23180 * BASE-VER: Bump to 7.0.1.
23182 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
23184 * omp-low.c (omplow_simd_context): New struct. Use it...
23185 (lower_rec_simd_input_clauses): ...here and...
23186 (lower_rec_input_clauses): ...here to hold common data. Adjust all
23187 references to idx, lane, max_vf, is_simt.
23189 2017-01-20 Graham Markall <graham.markall@embecosm.com>
23191 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
23194 2017-01-20 Martin Jambor <mjambor@suse.cz>
23196 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
23197 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
23199 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
23200 (GTFILES): Rename hsa.c to hsa-common.c.
23201 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
23202 * hsa-dump.c: Likewise.
23203 * hsa-gen.c: Likewise.
23204 * hsa-regalloc.c: Likewise.
23205 * ipa-hsa.c: Likewise.
23206 * omp-expand.c: Likewise.
23207 * omp-low.c: Likewise.
23208 * toplev.c: Likewise.
23210 2017-01-20 Marek Polacek <polacek@redhat.com>
23213 * doc/invoke.texi: Document -Wduplicated-branches.
23214 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
23215 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
23216 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
23217 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
23218 return 0 only when not OEP_LEXICOGRAPHIC.
23219 (fold_build_cleanup_point_expr): Use the expression
23220 location when building CLEANUP_POINT_EXPR.
23221 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
23222 * tree.c (add_expr): Handle error_mark_node.
23224 2017-01-20 Martin Liska <mliska@suse.cz>
23227 * tree-profile.c (init_ic_make_global_vars): Do not call
23229 (gimple_init_gcov_profiler): Likewise.
23231 2017-01-20 Martin Liska <mliska@suse.cz>
23234 * cgraph.h (maybe_create_reference): Remove argument and
23236 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
23238 * ipa-cp.c (create_specialized_node): Likewise.
23239 * symtab.c (symtab_node::maybe_create_reference): Handle
23240 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
23242 2017-01-20 Martin Liska <mliska@suse.cz>
23244 * read-rtl-function.c (function_reader::create_function): Use
23245 build_decl instread of build_decl_stat.
23247 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
23249 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
23250 * config/i386/avx512dqintrin.h: Ditto.
23251 * config/i386/avx512fintrin.h: Ditto.
23252 * config/i386/i386-builtin-types.def: Add new types.
23253 * config/i386/i386.c: Handle new types.
23254 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
23255 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
23256 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
23257 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
23258 (__builtin_ia32_kshiftridi): New.
23259 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
23261 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
23265 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
23266 define to rs6000_init_stack_protect_guard.
23267 (rs6000_init_stack_protect_guard): New function.
23269 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23270 Yunqiang Su <yunqiang.su@imgtec.com>
23272 * config.gcc (supported_defaults): Add madd4.
23273 (with_madd4): Add validation.
23274 (all_defaults): Add madd4.
23275 * config/mips/mips.opt (mmadd4): New option.
23276 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23278 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23280 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
23281 (ISA_HAS_FUSED_MADD4): Likewise.
23282 * doc/invoke.texi (-mmadd4): Document the new option.
23283 * doc/install.texi (--with-madd4): Document the new option.
23285 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23287 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
23288 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
23289 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
23290 (aarch64_init_pauth_hint_builtins): New.
23291 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
23292 (aarch64_expand_builtin): Expand new builtins.
23294 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23296 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
23297 * combine-stack-adj.c (no_unhandled_cfa): Handle
23298 REG_CFA_TOGGLE_RA_MANGLE.
23299 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
23300 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
23301 info for return address signing.
23302 (aarch64_expand_epilogue): Likewise.
23304 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23306 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
23307 * config/aarch64/aarch64-protos.h
23308 (aarch64_return_address_signing_enabled): New declaration.
23309 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
23311 (aarch64_expand_prologue): Sign return address before it's pushed onto
23313 (aarch64_expand_epilogue): Authenticate return address fetched from
23315 (aarch64_override_options): Sanity check for ILP32 and ISA level.
23316 (aarch64_attributes): New function attributes for "sign-return-address".
23317 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
23318 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
23319 ("*do_return"): Generate combined instructions according to key index.
23320 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
23321 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
23323 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
23324 * config/aarch64/aarch64.opt (msign-return-address=): New.
23325 * doc/extend.texi (AArch64 Function Attributes): Documents
23326 "sign-return-address=".
23327 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
23329 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
23331 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
23332 overall option summary.
23334 2017-01-19 Jiong Wang <jiong.wang@arm.com>
23336 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
23337 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
23338 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
23339 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
23341 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
23343 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
23344 -mpower9-minmax by default for -mcpu=power9.
23345 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
23346 128-bit floating point.
23348 2017-01-20 Alan Modra <amodra@gmail.com>
23350 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
23351 optimizing for size.
23353 2017-01-20 Alan Modra <amodra@gmail.com>
23356 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
23357 for strcmp and strncmp from corresponding builtin decl.
23359 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23361 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
23362 instead of i386/rtems-64.h.
23363 * config/i386/rtems-64.h: Remove.
23365 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
23369 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
23371 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
23373 2017-01-19 Tamar Christina <tamar.christina@arm.com>
23375 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
23376 Change int to HOST_WIDE_INT.
23377 * config/aarch64/aarch64-protos.h
23378 (aarch64_simd_gen_const_vector_dup): Likewise.
23379 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
23381 2017-01-19 David Malcolm <dmalcolm@redhat.com>
23383 * langhooks-def.h (lhd_type_for_size): New decl.
23384 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
23385 * langhooks.c (lhd_type_for_size): New function, taken from
23388 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
23390 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
23391 define_bypass for CR latency.
23392 (power9-cracked-alu): Update bypass latency and remove power9-branch.
23393 (power9-alu2): Add define_bypass for CR latency.
23395 (power9-mul): Update insn latency.
23396 (power9-mul-compare): Update insn latency, bypass latency and remove
23399 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23401 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
23403 * config/aarch64/aarch64.md
23404 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
23405 aarch64_nopcrelative_literal_loads.
23406 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
23408 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
23410 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
23411 TARGET_LOONGSON_3A.
23412 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
23414 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
23417 * config.gcc (supported_defaults): Add lxc1-sxc1.
23418 (with_lxc1_sxc1): Add validation.
23419 (all_defaults): Add lxc1-sxc1.
23420 * config/mips/mips.opt (mlxc1-sxc1): New option.
23421 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
23423 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
23424 __mips_no_lxc1_sxc1.
23425 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
23426 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
23427 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
23429 2017-01-19 Richard Biener <rguenther@suse.de>
23431 PR tree-optimization/72488
23432 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
23433 sure to restore SSA info.
23434 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
23436 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
23438 PR rtl-optimization/79121
23439 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
23440 of the inner type when shifting an extended value.
23442 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
23445 * symtab.c (symtab_node::equal_address_to): Fix comparing of
23446 interposable aliases.
23448 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
23451 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
23452 Use the evmergelohi instruction.
23453 (mov_si<mode>_e500_subreg4_2_le): Likewise.
23454 (mov_sitf_e500_subreg8_2_be): Likewise.
23455 (mov_sitf_e500_subreg12_2_le): Likewise.
23456 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
23457 (mov_si<mode>_e500_subreg4_2_be): Likewise.
23458 (mov_sitf_e500_subreg8_2_le): Likewise.
23459 (mov_sitf_e500_subreg12_2_be): Likewise.
23461 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23463 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
23464 attribute from vecsimple to vecperm.
23465 (altivec_vbpermq2): Likewise.
23467 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23470 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
23472 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23473 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
23474 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
23475 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
23476 case where N arg is SIZE_MAX.
23477 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
23478 (cmpstrsi): Add pattern.
23480 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23482 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23483 __builtin_vec_revb builtins.
23484 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
23485 built-in functions to support generation of the ISA 3.0 XXBR<x>
23486 vector byte reverse instructions.
23487 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
23488 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
23489 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
23490 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
23491 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
23492 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
23493 (P9V_BUILTIN_VEC_REVB): Likewise.
23494 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
23495 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
23496 (p9_xxbrq_v16qi): Likewise.
23497 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
23498 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
23499 (p9_xxbrh_v8hi): Likewise.
23500 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
23501 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
23502 vec_revb built-in functions.
23504 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
23506 PR rtl-optimization/78952
23507 * config/i386/i386.md (any_extract): New code iterator.
23508 (*insvqi_2): Use any_extract for source operand.
23509 (*insvqi_3): Use any_shiftrt for source operand.
23511 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
23513 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
23515 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
23517 2017-01-18 Matthias Klose <doko@ubuntu.com>
23519 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
23521 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23523 * config/rs6000/altivec.h (vec_bperm): Change #define.
23524 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
23525 (altivec_vbpermq2): New define_insn.
23526 (altivec_vbpermd): Likewise.
23527 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
23528 function interface.
23529 (VBPERMD): Likewise.
23530 (VBPERM): New polymorphic function interface.
23531 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
23532 Add entries for P9V_BUILTIN_VEC_VBPERM.
23533 * doc/extend.texi: Add interfaces for vec_bperm.
23535 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23537 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
23538 first letter of error messages.
23539 (s390_resolve_overloaded_builtin): Likewise.
23540 * config/s390/s390.c (s390_expand_builtin): Likewise.
23541 (s390_invalid_arg_for_unprototyped_fn): Likewise.
23542 (s390_valid_target_attribute_inner_p): Likewise.
23543 * config/s390/s390.md ("tabort"): Likewise.
23545 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
23547 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
23548 (ISA_AVOID_DIV_HILO): New macro.
23549 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
23550 (ISA_HAS_DDIV): Likewise.
23552 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23554 * doc/invoke.texi (fabi-version): Correct number of occurrences.
23556 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23558 * doc/invoke.texi (fabi-version): Spelling fix.
23560 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23563 * doc/invoke.texi (fabi-version): Mention mangling fix for
23566 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
23569 * doc/invoke.texi (fabi-version): Document discriminator mangling.
23571 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
23574 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
23575 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
23577 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
23579 (stack_protect_test): Ditto.
23580 * config/rs6000/rs6000.opt (mstack-protector-guard=,
23581 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
23583 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
23584 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
23585 -mstack-protector-guard-offset=.
23586 (RS/6000 and PowerPC Options): Ditto.
23588 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
23590 * config/i386/i386.h (MASK_CLASS_P): New define.
23591 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
23592 there are no registers from different register sets also when
23593 mask registers are used. Update function comment.
23594 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
23595 to (*k/*r) and (*k/*km) alternatives.
23597 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
23599 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
23600 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
23601 (EH_RETURN_HANDLER_RTX): New define.
23602 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
23603 Force frame pointer in EH return functions.
23604 (aarch64_expand_epilogue): Add barrier for eh_return.
23605 (aarch64_final_eh_return_addr): Remove.
23606 (aarch64_eh_return_handler_rtx): New function.
23607 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
23609 (aarch64_eh_return_handler_rtx): New prototype.
23611 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23613 * config/rs6000/altivec.h (vec_rlmi): New #define.
23614 (vec_vrlnm): Likewise.
23615 (vec_rlnm): Likewise.
23616 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
23617 (UNSPEC_VRLNM): Likewise.
23618 (VIlong): New mode iterator.
23619 (altivec_vrl<VI_char>mi): New define_insn.
23620 (altivec_vrl<VI_char>nm): Likewise.
23621 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
23623 (VRLDNM): Likewise.
23624 (RLNM): New polymorphic function entry.
23625 (VRLWMI): New monomorphic function entry.
23626 (VRLDMI): Likewise.
23627 (RLMI): New polymorphic function entry.
23628 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
23629 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
23630 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
23633 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23636 * dwarf2out.c (field_byte_offset): Restore the
23637 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
23638 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
23639 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
23642 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
23645 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
23647 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23650 * dwarf2out.c (add_data_member_location_attribute): For constant
23651 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
23652 instead of DW_AT_data_member_location, DW_AT_bit_offset and
23653 DW_AT_byte_size attributes.
23655 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
23657 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
23658 after forcing to constant memory when the code model is medium.
23660 2017-01-17 Julia Koval <julia.koval@intel.com>
23663 * config/i386/avx512fintrin.h
23664 (_mm512_i32gather_ps): Change __addr type to void const*.
23665 (_mm512_mask_i32gather_ps): Ditto.
23666 (_mm512_i32gather_pd): Ditto.
23667 (_mm512_mask_i32gather_pd): Ditto.
23668 (_mm512_i64gather_ps): Ditto.
23669 (_mm512_mask_i64gather_ps): Ditto.
23670 (_mm512_i64gather_pd): Ditto.
23671 (_mm512_mask_i64gather_pd): Ditto.
23672 (_mm512_i32gather_epi32): Ditto.
23673 (_mm512_mask_i32gather_epi32): Ditto.
23674 (_mm512_i32gather_epi64): Ditto.
23675 (_mm512_mask_i32gather_epi64): Ditto.
23676 (_mm512_i64gather_epi32): Ditto.
23677 (_mm512_mask_i64gather_epi32): Ditto.
23678 (_mm512_i64gather_epi64): Ditto.
23679 (_mm512_mask_i64gather_epi64): Ditto.
23680 (_mm512_i32scatter_ps): Change __addr type to void*.
23681 (_mm512_mask_i32scatter_ps): Ditto.
23682 (_mm512_i32scatter_pd): Ditto.
23683 (_mm512_mask_i32scatter_pd): Ditto.
23684 (_mm512_i64scatter_ps): Ditto.
23685 (_mm512_mask_i64scatter_ps): Ditto.
23686 (_mm512_i64scatter_pd): Ditto.
23687 (_mm512_mask_i64scatter_pd): Ditto.
23688 (_mm512_i32scatter_epi32): Ditto.
23689 (_mm512_mask_i32scatter_epi32): Ditto.
23690 (_mm512_i32scatter_epi64): Ditto.
23691 (_mm512_mask_i32scatter_epi64): Ditto.
23692 (_mm512_i64scatter_epi32): Ditto.
23693 (_mm512_mask_i64scatter_epi32): Ditto.
23694 (_mm512_i64scatter_epi64): Ditto.
23695 (_mm512_mask_i64scatter_epi64): Ditto.
23696 * config/i386/avx512pfintrin.h
23697 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
23698 (_mm512_mask_prefetch_i32gather_ps): Ditto.
23699 (_mm512_mask_prefetch_i64gather_pd): Ditto.
23700 (_mm512_mask_prefetch_i64gather_ps): Ditto.
23701 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
23702 (_mm512_prefetch_i32scatter_ps): Ditto.
23703 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
23704 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
23705 (_mm512_prefetch_i64scatter_pd): Ditto.
23706 (_mm512_prefetch_i64scatter_ps): Ditto.
23707 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
23708 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
23709 * config/i386/avx512vlintrin.h
23710 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
23711 (_mm_mmask_i32gather_ps): Ditto.
23712 (_mm256_mmask_i32gather_pd): Ditto.
23713 (_mm_mmask_i32gather_pd): Ditto.
23714 (_mm256_mmask_i64gather_ps): Ditto.
23715 (_mm_mmask_i64gather_ps): Ditto.
23716 (_mm256_mmask_i64gather_pd): Ditto.
23717 (_mm_mmask_i64gather_pd): Ditto.
23718 (_mm256_mmask_i32gather_epi32): Ditto.
23719 (_mm_mmask_i32gather_epi32): Ditto.
23720 (_mm256_mmask_i32gather_epi64): Ditto.
23721 (_mm_mmask_i32gather_epi64): Ditto.
23722 (_mm256_mmask_i64gather_epi32): Ditto.
23723 (_mm_mmask_i64gather_epi32): Ditto.
23724 (_mm256_mmask_i64gather_epi64): Ditto.
23725 (_mm_mmask_i64gather_epi64): Ditto.
23726 (_mm256_i32scatter_ps): Change __addr type to void*.
23727 (_mm256_mask_i32scatter_ps): Ditto.
23728 (_mm_i32scatter_ps): Ditto.
23729 (_mm_mask_i32scatter_ps): Ditto.
23730 (_mm256_i32scatter_pd): Ditto.
23731 (_mm256_mask_i32scatter_pd): Ditto.
23732 (_mm_i32scatter_pd): Ditto.
23733 (_mm_mask_i32scatter_pd): Ditto.
23734 (_mm256_i64scatter_ps): Ditto.
23735 (_mm256_mask_i64scatter_ps): Ditto.
23736 (_mm_i64scatter_ps): Ditto.
23737 (_mm_mask_i64scatter_ps): Ditto.
23738 (_mm256_i64scatter_pd): Ditto.
23739 (_mm256_mask_i64scatter_pd): Ditto.
23740 (_mm_i64scatter_pd): Ditto.
23741 (_mm_mask_i64scatter_pd): Ditto.
23742 (_mm256_i32scatter_epi32): Ditto.
23743 (_mm256_mask_i32scatter_epi32): Ditto.
23744 (_mm_i32scatter_epi32): Ditto.
23745 (_mm_mask_i32scatter_epi32): Ditto.
23746 (_mm256_i32scatter_epi64): Ditto.
23747 (_mm256_mask_i32scatter_epi64): Ditto.
23748 (_mm_i32scatter_epi64): Ditto.
23749 (_mm_mask_i32scatter_epi64): Ditto.
23750 (_mm256_i64scatter_epi32): Ditto.
23751 (_mm256_mask_i64scatter_epi32): Ditto.
23752 (_mm_i64scatter_epi32): Ditto.
23753 (_mm_mask_i64scatter_epi32): Ditto.
23754 (_mm256_i64scatter_epi64): Ditto.
23755 (_mm256_mask_i64scatter_epi64): Ditto.
23756 (_mm_i64scatter_epi64): Ditto.
23757 (_mm_mask_i64scatter_epi64): Ditto.
23758 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
23759 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
23760 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
23761 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
23762 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
23763 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
23764 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
23765 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
23766 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
23767 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
23768 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
23769 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
23770 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
23771 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
23772 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
23773 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
23774 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
23775 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
23776 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
23777 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
23778 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
23779 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
23780 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
23781 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
23782 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
23783 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
23784 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
23785 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
23786 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
23787 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
23788 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
23789 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
23790 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
23791 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
23792 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
23793 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
23794 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
23795 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
23796 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
23797 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
23798 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
23799 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
23800 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
23801 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
23802 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
23803 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
23804 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
23805 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
23806 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
23807 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
23808 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
23809 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
23810 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
23811 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
23812 definitions accordingly.
23814 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
23815 Kuan-Lin Chen <kuanlinchentw@gmail.com>
23818 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
23821 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
23824 * ira-conflicts.c (ira_build_conflicts): Update total conflict
23825 hard regs for inner regno.
23827 2017-01-17 Martin Liska <mliska@suse.cz>
23830 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
23831 assumption and add comment.
23833 2017-01-17 Nathan Sidwell <nathan@acm.org>
23835 * ipa-visibility.c (localize_node): New function, broken out of ...
23836 (function_and_variable_visibility): ... here. Call it.
23838 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
23840 PR middle-end/77445
23841 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
23842 correctly set frequency of oudgoing edge.
23843 (duplicate_thread_path): Fix profile updating.
23845 2017-01-17 Jakub Jelinek <jakub@redhat.com>
23848 * configure.ac: Add GCC_BASE_VER.
23849 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
23850 version from BASE-VER file.
23851 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
23852 (gcc.o): Depend on $(BASEVER).
23853 * common.opt (dumpfullversion): New option.
23854 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
23855 * doc/invoke.texi: Document -dumpfullversion.
23856 * doc/install.texi: Document --with-gcc-major-version-only.
23857 * configure: Regenerated.
23859 2017-01-17 Richard Biener <rguenther@suse.de>
23861 PR tree-optimization/71433
23862 * tree-vrp.c (register_new_assert_for): Merge same asserts
23863 on all incoming edges.
23864 (process_assert_insertions_for): Handle insertions at the
23867 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
23869 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
23870 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
23872 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
23875 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
23878 2017-01-17 Alan Modra <amodra@gmail.com>
23881 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
23882 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
23883 symbolic stack limit when pic.
23885 2017-01-16 Martin Sebor <msebor@redhat.com>
23887 PR tree-optimization/78608
23888 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
23890 2017-01-16 Jeff Law <law@redhat.com>
23893 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
23894 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
23895 for several include directories that may be relative to sysroot.
23896 * config/i386/x-mingw32 (gplus_includedir): Define.
23897 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
23898 (native_system_includedir): Likewise.
23899 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
23900 override if TARGET_SYSTEM_ROOT is defined.
23901 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
23903 PR tree-optimization/79090
23904 PR tree-optimization/33562
23905 PR tree-optimization/61912
23906 PR tree-optimization/77485
23907 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
23908 and computed trims into the dump file.
23910 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
23912 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
23914 2017-01-16 Jakub Jelinek <jakub@redhat.com>
23917 * gimplify.c (gimplify_init_constructor): If want_value and
23918 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
23922 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
23923 sequence. Formatting fixes.
23924 (doloop_optimize): Formatting fixes.
23927 * gcc.c (debug_level_greater_than_spec_func): New function.
23928 (static_spec_functions): Add debug-level-gt spec function.
23929 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
23931 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
23932 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
23933 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
23934 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
23935 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
23936 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
23938 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
23940 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
23941 QImode fixups to general and mask registers only.
23943 2017-01-16 Carl Love <cel@us.ibm.com>
23945 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
23946 for built-in functions
23947 vector signed char vec_nabs (vector signed char)
23948 vector signed short vec_nabs (vector signed short)
23949 vector signed int vec_nabs (vector signed int)
23950 vector signed long long vec_nabs (vector signed long long)
23951 vector float vec_nabs (vector float)
23952 vector double vec_nabs (vector double)
23953 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
23955 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
23956 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
23957 * doc/extend.texi: Update the documentation file for the new built-in
23960 2017-01-16 Martin Sebor <msebor@redhat.com>
23962 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
23965 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23967 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
23968 UNSPEC_VSX__XXSPLTD to require special splat handling.
23970 2017-01-16 David Malcolm <dmalcolm@redhat.com>
23973 * system.h: Poison strndup.
23975 2017-01-16 Alan Modra <amodra@gmail.com>
23978 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
23981 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
23983 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
23985 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
23987 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
23988 call recog here. Assert that INSN_CODE (insn) is non-negative.
23990 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
23993 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
23995 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
23996 in the currently scheduled RTL fragment.
23998 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
24000 PR rtl-optimization/78751
24001 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
24004 2017-01-14 Jeff Law <law@redhat.com>
24006 PR tree-optimization/79090
24007 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
24008 variable length stores.
24009 (compute_trims): Delete dead assignment to *trim_tail.
24010 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
24013 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
24015 PR rtl-optimization/78626
24016 PR rtl-optimization/78727
24017 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
24018 of a block, and split such blocks after everything else is finished.
24020 2017-01-14 Alan Modra <amodra@gmail.com>
24023 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
24024 target legitimate_combined_insn.
24025 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
24026 (rs6000_legitimate_combined_insn): New function.
24027 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
24029 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
24030 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
24031 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
24033 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
24035 * doc/frontends.texi (G++ and GCC): Remove references to Java.
24037 2017-01-13 Jeff Law <law@redhat.com>
24039 PR tree-optimization/33562
24040 PR tree-optimization/61912
24041 PR tree-optimization/77485
24042 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
24044 (delete_dead_assignment): Likewise.
24045 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
24046 statement to delete_dead_call and delete_dead_assignment.
24048 2017-01-13 David Malcolm <dmalcolm@redhat.com>
24051 * substring-locations.c (format_warning_va): Strengthen case 1 so
24052 that both endpoints of the substring must be within the format
24053 range for just the substring to be printed.
24055 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
24057 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
24058 * config/i386/i386.c (ix86_target_string): Add missing options
24059 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
24060 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
24061 flags_other and ix86_target_other to flags2_other. Display unknown
24063 (ix86_valid_target_attribute_inner_p): Add missing options and
24064 reorder options by implied ISAs, as in ix86_target_string.
24066 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24068 * hash-table.h (hash_table::too_empty_p): New function.
24069 (hash_table::expand): Use it.
24070 (hash_table::traverse): Likewise.
24071 (hash_table::empty_slot): Use sizeof (value_type) instead of
24072 sizeof (PTR) to convert bytes to elements. Shrink the table
24073 if the current size is excessive for the current number of
24076 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
24078 * ira-costs.c (record_reg_classes): Break from the inner loop
24079 early once alt_fail is known to be true. Update outer loop
24080 handling accordingly.
24082 2017-01-13 Jeff Law <law@redhat.com>
24084 * tree-ssa-dse.c (decrement_count): New function.
24085 (increment_start_addr, maybe_trim_memstar_call): Likewise.
24086 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
24087 when we know the partially dead statement is a mem* function.
24089 PR tree-optimization/61912
24090 PR tree-optimization/77485
24091 * tree-ssa-dse.c: Include expr.h.
24092 (maybe_trim_constructor_store): New function.
24093 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
24095 PR tree-optimization/33562
24096 PR tree-optimization/61912
24097 PR tree-optimization/77485
24098 * doc/invoke.texi: Document new dse-max-object-size param.
24099 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
24100 * tree-ssa-dse.c: Include params.h.
24101 (dse_store_status): New enum.
24102 (initialize_ao_ref_for_dse): New, partially extracted from
24104 (valid_ao_ref_for_dse, normalize_ref): New.
24105 (setup_live_bytes_from_ref, compute_trims): Likewise.
24106 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
24107 (maybe_trim_partially_dead_store): Likewise.
24108 (maybe_trim_complex_store): Likewise.
24109 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
24110 Track what bytes live from the original store. Return tri-state
24111 for dead, partially dead or live.
24112 (dse_dom_walker): Add constructor, destructor and new private members.
24113 (delete_dead_call, delete_dead_assignment): New extracted from
24115 (dse_optimize_stmt): Make a member of dse_dom_walker.
24116 Use initialize_ao_ref_for_dse.
24118 PR tree-optimization/33562
24119 PR tree-optimization/61912
24120 PR tree-optimization/77485
24121 * sbitmap.h (bitmap_count_bits): Prototype.
24122 (bitmap_clear_range, bitmap_set_range): Likewise.
24123 * sbitmap.c (bitmap_clear_range): New function.
24124 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
24126 2017-01-13 Martin Liska <mliska@suse.cz>
24129 * function.c (set_cfun): Add new argument force.
24130 * function.h (set_cfun): Likewise.
24131 * ipa-inline-transform.c (inline_call): Use the function when
24132 strict alising from is dropped for function we inline to.
24134 2017-01-13 Richard Biener <rguenther@suse.de>
24136 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
24137 for dumping GIMPLE INTEGER_CSTs.
24139 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24141 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
24142 to 201112L since C++17.
24144 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
24147 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
24148 if -fsanitize=kernel-address is present.
24150 2017-01-13 Richard Biener <rguenther@suse.de>
24152 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
24153 as _Literal ( type ) number in case usual suffixes do not
24154 preserve all information.
24156 2017-01-13 Richard Biener <rguenther@suse.de>
24158 PR tree-optimization/77283
24159 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
24160 and ssa-iterators.h.
24161 (is_feasible_trace): Implement a cost model based on joiner
24164 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
24167 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
24168 char or short to __float128/_Float128 directly.
24170 2017-01-12 Martin Sebor <msebor@redhat.com>
24172 to -Wformat-overflow.
24173 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
24174 (min_bytes_remaining): Same.
24175 (get_string_length): Same.
24176 (format_string): Same.
24177 (format_directive): Same.
24179 (pass_sprintf_length::handle_gimple_call): Same.
24181 2017-01-12 Jakub Jelinek <jakub@redhat.com>
24183 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
24184 info.nowrite calls with no lhs that can't throw. Return bool
24185 whether gsi_remove has been called or not.
24186 (pass_sprintf_length::handle_gimple_call): Return bool whether
24187 try_substitute_return_value called gsi_remove. Formatting fix.
24188 (pass_sprintf_length::execute): Don't use gsi_remove if
24189 handle_gimple_call returned true.
24192 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
24193 be removed due to side-effects, don't remove following barrier nor
24194 turn the successor edge into fallthru edge.
24196 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24199 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
24200 element-reversing loads and stores as not swappable.
24202 2017-01-12 Nathan Sidwell <nathan@acm.org>
24203 Nicolai Stange <nicstange@gmail.com>
24205 * combine.c (try_combine): Don't ignore result of overlap checking
24206 loop. Combine overlap & asm check into single loop.
24208 2017-01-12 Richard Biener <rguenther@suse.de>
24210 * tree-pretty-print.c (dump_generic_node): Provide -gimple
24211 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
24213 2017-01-12 Richard Biener <rguenther@suse.de>
24215 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
24216 and TS_TARGET_OPTION directly derive from TS_BASE.
24217 * tree-core.h (tree_optimization_option): Derive from tree_base.
24218 (tree_target_option): Likewise.
24220 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
24222 * config/i386/i386.c (memory_address_length): Increase len
24223 only when rip_relative_addr_p returns false.
24225 2017-01-11 Julia Koval <julia.koval@intel.com>
24227 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
24228 (OPTION_MASK_ISA_SGX_SET): New.
24229 (ix86_handle_option): Handle OPT_msgx.
24230 * config.gcc: Added sgxintrin.h.
24231 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
24232 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
24233 * config/i386/i386.c (ix86_target_string): Add -msgx.
24235 (ix86_option_override_internal): Handle new options.
24236 (ix86_valid_target_attribute_inner_p): Add sgx.
24237 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
24238 * config/i386/i386.opt: Add msgx.
24239 * config/i386/sgxintrin.h: New file.
24240 * config/i386/x86intrin.h: Add sgxintrin.h.
24242 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24245 * fold-const.c (maybe_nonzero_address): Return 1 for function
24247 (tree_single_nonzero_warnv_p): Don't handle function local objects
24251 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
24254 2017-01-11 David Malcolm <dmalcolm@redhat.com>
24257 * opts.c: Include "spellcheck.h"
24258 (struct string_fragment): New struct.
24259 (struct edit_distance_traits<const string_fragment &>): New
24261 (get_closest_sanitizer_option): New function.
24262 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
24264 2017-01-11 Jakub Jelinek <jakub@redhat.com>
24266 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
24268 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
24269 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
24270 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
24271 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
24272 for initial die_offset if dwarf_split_debug_info.
24273 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
24274 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
24276 (output_skeleton_debug_sections): Formatting fix. Use
24277 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
24278 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
24280 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
24282 * config/arm/cortex-a53.md: Add bypasses for
24283 cortex_a53_r2f_cvt.
24284 (cortex_a53_r2f): Only use for transfers.
24285 (cortex_a53_f2r): Likewise.
24286 (cortex_a53_r2f_cvt): Add reservation for conversions.
24287 (cortex_a53_f2r_cvt): Likewise.
24289 2017-01-11 Tamar Christina <tamar.christina@arm.com>
24291 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
24292 to all inlined functions, change static to extern.
24294 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
24297 * config/arm/arm.c (legitimize_pic_address): Handle reference to
24299 (arm_assemble_integer): Likewise.
24301 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
24303 * config.gcc: Use new awk script to check CPU, FPU and architecture
24304 parameters for --with-... options.
24305 * config/arm/parsecpu.awk: New file
24306 * config/arm/arm-cpus.in: New file.
24307 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
24309 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
24311 * config/arm/t-arm: Update dependency rules.
24312 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
24313 of processing .def files.
24314 * config/arm/genopt.sh: Deleted.
24315 * config/arm/gentune.sh: Deleted.
24316 * config/arm/arm-cores.def: Deleted.
24317 * config/arm/arm-arches.def: Deleted.
24318 * config/arm/arm-fpus.def: Deleted.
24319 * config/arm/arm-tune.md: Regenerated.
24320 * config/arm/arm-tables.opt: Regenerated.
24321 * config/arm/arm-cpu.h: New generated file.
24322 * config/arm/arm-cpu-data.h: New generated file.
24323 * config/arm/arm-cpu-cdata.h: New generated file.
24325 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
24328 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
24330 (input_varpool_node): Unpack dynamically_initialized bit.
24332 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
24334 PR rtl-optimization/79032
24335 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
24336 the alignment of the adjusted memory reference against that of MODE,
24337 instead of the alignment of the original memory reference.
24339 2017-01-11 Martin Jambor <mjambor@suse.cz>
24341 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
24343 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
24344 decorated functions.
24346 2017-01-11 Richard Biener <rguenther@suse.de>
24348 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
24349 set range/nonnull info for PHI results. Do not set it on
24350 stmts marked for removal.
24352 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
24354 * expr.c (store_field): In the bitfield case, fetch the return value
24355 from the registers before applying a single big-endian adjustment.
24356 Always do a final load for a BLKmode value not larger than a word.
24358 2017-01-10 David Malcolm <dmalcolm@redhat.com>
24361 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
24362 that we correctly handle column numbers greater than
24363 LINE_MAP_MAX_COLUMN_NUMBER.
24365 2017-01-10 Martin Sebor <msebor@redhat.com>
24367 PR middle-end/78245
24368 * gimple-ssa-sprintf.c (get_destination_size): Call
24369 {init,fini}object_sizes.
24370 * tree-object-size.c (addr_object_size): Adjust.
24371 (pass_through_call): Adjust.
24372 (pass_object_sizes::execute): Adjust.
24373 * tree-object-size.h (fini_object_sizes): Declare.
24375 2017-01-10 Martin Sebor <msebor@redhat.com>
24377 PR tree-optimization/78775
24378 * builtins.c (get_size_range): Move...
24379 * calls.c: ...to here.
24380 (alloc_max_size): Accept zero argument.
24381 (operand_signed_p): Remove.
24382 (maybe_warn_alloc_args_overflow): Call get_size_range.
24383 * calls.h (get_size_range): Declare.
24385 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
24387 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
24388 from TI's devices.csv file as of September 2016.
24389 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
24391 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
24393 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
24394 * doc/invoke.texi: Likewise.
24395 * doc/md.texi: Likewise.
24396 * doc/objc.texi: Likewise.
24398 2017-01-10 Joshua Conner <joshconner@google.com>
24400 * config/arm/fuchsia-elf.h: New file.
24401 * config/fuchsia.h: New file.
24402 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
24403 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
24405 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
24407 2016-01-10 Richard Biener <rguenther@suse.de>
24409 PR tree-optimization/79034
24410 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
24411 Propagate out degenerate PHIs in the joiner.
24413 2017-01-10 Martin Liska <mliska@suse.cz>
24415 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
24416 (sort_congruence_classes_by_decl_uid): Likewise.
24417 (sort_congruence_class_groups_by_decl_uid): Likewise.
24418 (sem_item_optimizer::merge_classes): Sort class, groups in these
24419 classes and members in the groups by DECL_UID of declarations.
24420 This would make merge operations stable.
24422 2017-01-10 Martin Liska <mliska@suse.cz>
24424 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
24425 usage of m_classes_vec.
24426 (sem_item_optimizer::~sem_item_optimizer): Likewise.
24427 (sem_item_optimizer::get_group_by_hash): Likewise.
24428 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
24429 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
24430 (sem_item_optimizer::verify_classes): Likewise.
24431 (sem_item_optimizer::process_cong_reduction): Likewise.
24432 (sem_item_optimizer::dump_cong_classes): Likewise.
24433 (sem_item_optimizer::merge_classes): Likewise.
24434 * ipa-icf.h (congruence_class_hash): Rename from
24435 congruence_class_group_hash. Remove declaration of m_classes_vec.
24437 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
24439 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
24440 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
24441 * config.gcc: Add avx512vpopcntdqintrin.h.
24442 * config/i386/avx512vpopcntdqintrin.h: New.
24443 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
24444 * config/i386/i386-builtin-types.def: Add new types.
24445 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
24446 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
24447 __builtin_ia32_vpopcountq_v8di_mask): New.
24448 * config/i386/i386-c.c (ix86_target_macros_internal): Define
24449 __AVX512VPOPCNTDQ__.
24450 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
24451 (PTA_AVX512VPOPCNTDQ): Define.
24452 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
24453 TARGET_AVX512VPOPCNTDQ_P): Define.
24454 * config/i386/i386.opt: Add mavx512vpopcntdq.
24455 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
24456 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
24458 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24460 PR middle-end/77484
24461 * predict.def (PRED_CALL): Set to 67.
24463 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
24465 * expr.c (store_field): In the bitfield case, if the value comes from
24466 a function call and is of an aggregate type returned in registers, do
24467 not modify the field mode; extract the value in all cases if the mode
24468 is BLKmode and the size is not larger than a word.
24470 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
24473 * config/i386/cpuid.h: Fix undefined behavior.
24475 2017-01-04 Jeff Law <law@redhat.com>
24477 PR tree-optimization/79007
24478 PR tree-optimization/67955
24479 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
24480 conservative for pt.null when flag_non_call_exceptions is on.
24482 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24484 PR translation/79019
24485 PR translation/79020
24486 * params.def (PARAM_INLINE_MIN_SPEEDUP,
24487 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
24488 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
24490 * config/avr/avr.opt (maccumulate-args): Likewise.
24491 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
24492 * common.opt (freport-bug): Likewise.
24493 * cif-code.def (CIF_FINAL_ERROR): Likewise.
24494 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
24495 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
24496 translatable string.
24497 * config/i386/i386.c (function_value_32): Likewise.
24498 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
24499 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
24501 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
24502 * common/config/msp430/msp430-common.c (msp430_handle_option):
24504 * symtab.c (symtab_node::verify_base): Likewise.
24505 * opts.c (set_debug_level): Likewise.
24506 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
24507 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
24508 missing whitespace to translatable strings.
24509 * config/avr/avr.md (bswapsi2): Fix typo in comment.
24510 * config/sh/superh.h: Likewise.
24511 * config/i386/xopintrin.h: Likewise.
24512 * config/i386/znver1.md: Likewise.
24513 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
24514 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
24515 * double-int.h (struct double_int): Likewise.
24516 * double-int.c (div_and_round_double): Likewise.
24517 * wide-int.cc: Likewise.
24518 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
24519 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
24520 * cfgcleanup.c (crossjumps_occured): Renamed to ...
24521 (crossjumps_occurred): ... this.
24522 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
24525 PR tree-optimization/78899
24526 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
24527 returning bool return struct loop *, NULL for failure and the new
24529 (versionable_outer_loop_p): Don't version outer loop if it has
24530 dont_vectorized bit set.
24531 (tree_if_conversion): When versioning outer loop, ensure
24532 tree_if_conversion is performed also on the inner loop of the
24533 non-vectorizable outer loop copy.
24534 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
24535 LOOP_VECTORIZED in inner loop of the scalar outer loop and
24536 prevent vectorization of it.
24537 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
24538 the outer loop vectorization of the non-scalar version is attempted
24539 before vectorization of the inner loop in scalar version. If
24540 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
24541 vectorization of its inner loop.
24542 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
24543 has 2 inner loops, rename also on edges from bb whose single pred
24544 is outer_loop->header. Fix typo in function comment.
24546 2017-01-09 Martin Sebor <msebor@redhat.com>
24549 * asan.c (asan_emit_stack_protection): Increase local buffer size
24550 to avoid snprintf truncation warning.
24552 2017-01-09 Andrew Pinski <apinski@cavium.com>
24554 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
24555 to reference thunderx2t99 for the tuning structure
24556 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
24558 (thunderx2t99_extra_costs): This.
24559 * config/aarch64/aarch64-tune.md: Regenerate.
24560 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
24561 (vulcan_addrcost_table): This.
24562 (vulcan_regmove_cost): Rename to ...
24563 (thunderx2t99_regmove_cost): This.
24564 (vulcan_vector_cost): Rename to ...
24565 (thunderx2t99_vector_cost): this.
24566 (vulcan_branch_cost): Rename to ...
24567 (thunderx2t99_branch_cost): This.
24568 (vulcan_tunings): Rename to ...
24569 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
24570 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
24572 2017-01-09 Martin Jambor <mjambor@suse.cz>
24576 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
24577 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
24578 (propagate_vr_accross_jump_function): Use the above function for all
24579 value range computations for pass-through jump functions and type
24580 converasion from explicit value range values.
24581 (ipcp_propagate_stage): Do not attempt to deduce types of formal
24582 parameters from TYPE_ARG_TYPES.
24583 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
24584 (ipa_write_node_info): Stream type of the actual argument.
24585 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
24587 2017-01-09 Martin Liska <mliska@suse.cz>
24590 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
24591 (lookup_compiler): Do not show error message with have_E.
24593 2017-01-09 Jakub Jelinek <jakub@redhat.com>
24595 PR tree-optimization/78938
24596 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
24597 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
24598 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
24599 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
24602 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24604 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
24607 2017-01-09 Richard Biener <rguenther@suse.de>
24609 PR tree-optimization/78997
24610 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
24611 name condition properly.
24613 2017-01-09 Richard Biener <rguenther@suse.de>
24616 * dwarf2out.c (is_cxx): New overload with context.
24617 (is_naming_typedef_decl): Use it.
24619 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24621 * invoke.texi (Option Summary): Correct spacing in option lists
24622 and add line breaks to fix over-long lines.
24624 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24626 PR middle-end/17660
24628 * extend.texi (Common Variable Attributes): Add xref to GCC
24629 Internals manual to explain mode attribute keywords.
24631 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
24634 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
24635 and Preprocessor Options.
24636 (Options for Linking): Document -pthread here....
24637 (RS/6000 and PowerPC Options): ...not here.
24638 (Solaris 2 Options): ...or here.
24639 * doc/cppopts.texi: Document -pthread.
24641 2017-01-08 Martin Sebor <msebor@redhat.com>
24643 PR middle-end/77708
24644 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
24645 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
24646 New member functions.
24647 (format_directive): Used them.
24649 (pass_sprintf_length::handle_gimple_call): Same.
24650 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
24651 to avoid truncation for any argument.
24652 (extract_affine_mul): Same.
24653 * tree.c (get_file_function_name): Same.
24655 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24657 PR middle-end/77484
24658 * predict.def (PRED_INDIR_CALL): Set to 86.
24660 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24662 PR preprocessor/54124
24663 * doc/cppopts.texi: Reformat -d subtable to list the full name
24664 of the options. Add cross-reference to the docs for the general
24665 compiler -d options.
24666 * doc/invoke.texi (Developer Options): Add cross-reference to the
24667 preprocessor-specific -d option documentation.
24669 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24671 PR preprocessor/13498
24672 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
24673 redudant material, and reflect new command-line options.
24674 (System Headers): Likewise.
24676 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
24678 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
24679 -isystem, and -idirafter. Copy-edit.
24680 * doc/cppopts.texi: Copy-edit. Remove contradiction about
24681 default for -ftrack-macro-expansion. Delete obsolete and
24682 badly-formatted implementation details about -fdebug-cpp output.
24683 * doc/cppwarnopts.texi: Copy-edit.
24685 2017-01-07 David Malcolm <dmalcolm@redhat.com>
24688 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
24689 that the transition from a max line width >= 1<<10 to narrower
24690 lines works correctly.
24692 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
24694 * doc/options.texi (PerFunction): New.
24695 * opt-functions.awk (switch_flags): Map both Optimization and
24696 PerFunction to CL_OPTIMIZATION.
24697 * opth-gen.awk: Test for PerFunction flag along with
24699 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
24700 it only when the latter is present. Skip those that don't in
24701 the hash function generator.
24702 * common.opt (fvar-tracking): Mark as PerFunction instead of
24704 (fvar-tracking-assignments): Likewise.
24705 (fvar-tracking-assignments-toggle): Likewise.
24706 (fvar-tracking-uninit): Likewise.
24708 2017-01-07 Jakub Jelinek <jakub@redhat.com>
24710 PR translation/79018
24711 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
24714 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
24717 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
24718 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
24719 TARGET_LONG_JUMP_TABLE_OFFSETS.
24720 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
24721 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
24722 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
24723 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
24724 * config/m68k/m68k.md (tablejump expander): Likewise.
24725 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
24726 TARGET_LONG_JUMP_TABLE_OFFSETS.
24727 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
24728 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
24730 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
24731 David Holsgrove <david.holsgrove@xilinx.com>
24733 * common/config/microblaze/microblaze-common.c
24734 (TARGET_EXCEPT_UNWIND_INFO): Remove.
24735 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
24737 * config/microblaze/microblaze.c (microblaze_must_save_register)
24738 (microblaze_expand_epilogue, microblaze_return_addr): Handle
24740 (microblaze_eh_return): New function.
24741 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
24742 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
24743 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
24744 * config/microblaze/microblaze.md (eh_return): New pattern.
24746 2017-01-06 Jakub Jelinek <jakub@redhat.com>
24748 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
24749 GCC_DIAGNOSTIC_STRINGIFY): Define.
24751 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
24753 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24755 * config/arm/arm.md (<mcrr>): New.
24757 * config/arm/arm.c (arm_arch5te): New.
24758 (arm_option_override): Set arm_arch5te.
24759 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
24761 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
24762 (arm_mcrr_qualifiers): ... this. New.
24763 (MRRC_QUALIFIERS): Define to...
24764 (arm_mrrc_qualifiers): ... this. New.
24765 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
24767 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
24768 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
24769 (MRRCI, mrrc, MRRC): New.
24770 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
24771 VUNSPEC_MRRC2): New.
24773 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24775 * config/arm/arm.md (<mcr>): New.
24777 * config/arm/arm.c (arm_coproc_builtin_available): Add
24778 support for mcr, mrc, mcr2 and mrc2.
24779 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
24780 (arm_mcr_qualifiers): ... this. New.
24781 (MRC_QUALIFIERS): Define to ...
24782 (arm_mrc_qualifiers): ... this. New.
24783 (MCR_QUALIFIERS): Define to ...
24784 (arm_mcr_qualifiers): ... this. New.
24785 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
24787 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
24788 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
24789 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
24790 VUNSPEC_MRC2): New.
24792 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24794 * config/arm/arm.md (*ldc): New.
24798 * config/arm/arm.c (arm_coproc_builtin_available): Add
24799 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
24800 (arm_coproc_ldc_stc_legitimate_address): New.
24801 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
24802 'qualifier_const_pointer'.
24803 (LDC_QUALIFIERS): Define to...
24804 (arm_ldc_qualifiers): ... this. New.
24805 (STC_QUALIFIERS): Define to...
24806 (arm_stc_qualifiers): ... this. New.
24807 * config/arm/arm-protos.h
24808 (arm_coproc_ldc_stc_legitimate_address): New.
24809 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
24810 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
24811 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
24812 stc2, stcl, stc2l): New.
24813 * config/arm/constraints.md (Uz): New.
24814 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
24815 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
24816 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
24817 VUNSPEC_STC2L): New.
24819 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24821 * config/arm/arm.md (<cdp>): New.
24822 * config/arm/arm.c (neon_const_bounds): Rename this ...
24823 (arm_const_bounds): ... this.
24824 (arm_coproc_builtin_available): New.
24825 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
24826 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
24827 (CDP_QUALIFIERS): Define to...
24828 (arm_cdp_qualifiers): ... this. New.
24830 (arm_expand_builtin_args): Add case for 6 arguments.
24831 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
24832 (arm_const_bounds): ... this.
24833 (arm_coproc_builtin_available): New.
24834 * config/arm/arm_acle.h (__arm_cdp): New.
24836 * config/arm/arm_acle_builtins.def (cdp): New.
24838 * config/arm/iterators.md (CDPI,CDP,cdp): New.
24839 * config/arm/neon.md: Rename all 'neon_const_bounds' to
24840 'arm_const_bounds'.
24841 * config/arm/types.md (coproc): New.
24842 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
24843 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
24844 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
24845 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
24847 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24849 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
24850 (UBINOP_QUALIFIERS): New.
24852 (acle_builtin_data): New. Change comment.
24853 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
24854 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
24855 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
24856 arm_acle_builtins.def.
24857 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
24858 (arm_init_acle_builtins): New.
24859 (CRC32_BUILTIN): Remove.
24860 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
24861 crc32cb, crc32ch and crc32cw.
24862 (arm_init_crc32_builtins): Remove.
24863 (arm_init_builtins): Use arm_init_acle_builtins rather
24864 than arm_init_crc32_builtins.
24865 (arm_expand_acle_builtin): New.
24866 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
24867 * config/arm/arm_acle_builtins.def: New.
24869 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
24871 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
24872 (arm_builtin_datum): ... this.
24873 (arm_init_neon_builtin): Rename to ...
24874 (arm_init_builtin): ... this. Add a new parameters PREFIX
24875 and USE_SIG_IN_NAME.
24876 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
24877 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
24878 'arm_builtin_datum'.
24879 (arm_init_vfp_builtins): Likewise.
24880 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
24881 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
24882 (arm_expand_neon_args): Rename to ...
24883 (arm_expand_builtin_args): ... this. Rename builtin_arg
24884 enum values and differentiate between ARG_BUILTIN_MEMORY
24885 and ARG_BUILTIN_NEON_MEMORY.
24886 (arm_expand_neon_builtin_1): Rename to ...
24887 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
24888 values, arm_expand_builtin_args and add bool parameter NEON.
24889 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
24890 (arm_expand_vfp_builtin): Likewise.
24891 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
24893 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24895 PR middle-end/77484
24896 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
24897 * predict.c (tree_estimate_probability_bb): Reverse direction of
24898 polymorphic call predictor.
24900 2017-01-06 David Malcolm <dmalcolm@redhat.com>
24902 * passes.c (execute_one_pass): Split out pass-skipping logic into...
24903 (determine_pass_name_match): ...this new function and...
24904 (should_skip_pass_p): ...this new function.
24906 2017-01-06 Nathan Sidwell <nathan@acm.org>
24908 * ipa-visibility.c (function_and_variable_visibility): Reformat
24909 comments and long lines. Remove extrneous if.
24910 * symtab.c (symtab_node::make_decl_local): Fix code format.
24911 (symtab_node::set_section_for_node): Fix comment typo.
24913 2017-01-06 Martin Liska <mliska@suse.cz>
24916 * lra-constraints.c: Rename invariant to lra_invariant.
24917 * predict.c (set_even_probabilities): Initialize e to NULL.
24919 2017-01-05 Martin Sebor <msebor@redhat.com>
24921 PR tree-optimization/78910
24922 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
24923 (format_integer): Correct off-by-one error in the handling
24924 of precision with negative numbers in signed conversions..
24926 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
24928 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
24930 2017-01-05 Jakub Jelinek <jakub@redhat.com>
24932 PR tree-optimization/71016
24933 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
24934 factor_out_conditional_conversion. Formatting fix.
24935 (factor_out_conditional_conversion): Add cond_stmt argument.
24936 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
24937 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
24940 2017-01-05 David Malcolm <dmalcolm@redhat.com>
24942 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
24943 read-rtl-function.o, and selftest-rtl.o.
24944 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
24945 (selftest::aarch64_test_loading_full_dump): New function.
24946 (selftest::aarch64_run_selftests): New function.
24947 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
24948 selftest::aarch64_run_selftests.
24949 * config/i386/i386.c
24950 (selftest::ix86_test_loading_dump_fragment_1): New function.
24951 (selftest::ix86_test_loading_call_insn): New function.
24952 (selftest::ix86_test_loading_full_dump): New function.
24953 (selftest::ix86_test_loading_unspec): New function.
24954 (selftest::ix86_run_selftests): Call the new functions.
24955 * emit-rtl.c (maybe_set_max_label_num): New function.
24956 * emit-rtl.h (maybe_set_max_label_num): New decl.
24957 * function.c (instantiate_decls): Guard call to
24958 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
24959 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
24961 * gensupport.c (gen_reader::gen_reader): Pass "false"
24962 for new "compact" param of rtx_reader.
24963 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
24964 rather than an empty string for NULL strings.
24965 * read-md.c: Potentially include config.h rather than bconfig.h.
24966 Wrap include of errors.h with #ifdef GENERATOR_FILE.
24967 (have_error): New global, copied from errors.c.
24968 (md_reader::read_name): Rename to...
24969 (md_reader::read_name_1): ...this, adding "out_loc" param,
24970 and converting "missing name or number" to returning false, rather
24972 (md_reader::read_name): Reimplement in terms of read_name_1.
24973 (md_reader::read_name_or_nil): New function.
24974 (md_reader::read_string): Handle "(nil)" by returning NULL.
24975 (md_reader::md_reader): Add new param "compact".
24976 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
24977 (md_reader::read_file): New method.
24978 * read-md.h (md_reader::md_reader): Add new param "compact".
24979 (md_reader::read_file): New method.
24980 (md_reader::is_compact): New accessor.
24981 (md_reader::read_name): Convert return type from void to file_location.
24982 (md_reader::read_name_or_nil): New decl.
24983 (md_reader::read_name_1): New decl.
24984 (md_reader::m_compact): New field.
24985 (noop_reader::noop_reader): Pass "false" for new "compact" param
24987 (rtx_reader::rtx_reader): Add new "compact" param.
24988 (rtx_reader::read_rtx_operand): Make virtual and convert return
24989 type from void to rtx.
24990 (rtx_reader::read_until): New decl.
24991 (rtx_reader::handle_any_trailing_information): New virtual function.
24992 (rtx_reader::postprocess): New virtual function.
24993 (rtx_reader::finalize_string): New virtual function.
24994 (rtx_reader::m_in_call_function_usage): New field.
24995 (rtx_reader::m_reuse_rtx_by_id): New field.
24996 * read-rtl-function.c: New file.
24997 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
24998 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
24999 (selftest::verify_three_block_rtl_cfg): New decl.
25000 * read-rtl-function.h: New file.
25001 * read-rtl.c: Potentially include config.h rather than bconfig.h.
25002 For host, include function.h, memmodel.h, and emit-rtl.h.
25003 (one_time_initialization): New function.
25004 (struct compact_insn_name): New struct.
25005 (compact_insn_names): New array.
25006 (find_code): Handle insn codes in compact dumps.
25007 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
25008 (bind_subst_iter_and_attr): Likewise.
25009 (add_condition_to_string): Likewise.
25010 (add_condition_to_rtx): Likewise.
25011 (apply_attribute_uses): Likewise.
25012 (add_current_iterators): Likewise.
25013 (apply_iterators): Likewise.
25014 (initialize_iterators): Guard usage of apply_subst_iterator with
25015 #ifdef GENERATOR_FILE.
25016 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
25017 (md_reader::read_mapping): Likewise.
25018 (add_define_attr_for_define_subst): Likewise.
25019 (add_define_subst_attr): Likewise.
25020 (read_subst_mapping): Likewise.
25021 (check_code_iterator): Likewise.
25022 (rtx_reader::read_rtx): Likewise. Move one-time initialization
25024 (one_time_initialization): New function.
25025 (rtx_reader::read_until): New method.
25026 (read_flags): New function.
25027 (parse_reg_note_name): New function.
25028 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
25029 Handle reuse_rtx ids.
25030 Wrap iterator lookup within #ifdef GENERATOR_FILE.
25031 Add parsing support for RTL dumps, mirroring the special-cases in
25032 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
25033 values, and calling handle_any_trailing_information.
25034 (rtx_reader::read_rtx_operand): Convert return type from void
25035 to rtx, returning return_rtx. Handle case 'e'. Call
25036 finalize_string on XSTR and XTMPL fields.
25037 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
25038 "(nil)" values were omitted. Call the postprocess vfunc on the
25040 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
25041 class ctor. Initialize m_in_call_function_usage. Call
25042 one_time_initialization.
25043 * rtl-tests.c (selftest::test_uncond_jump): Call
25044 set_new_first_and_last_insn.
25045 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
25046 * selftest-rtl.c: New file.
25047 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
25048 (selftest::get_insn_by_uid): New decl.
25049 * selftest-run-tests.c (selftest::run_tests): Call
25050 read_rtl_function_c_tests.
25051 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
25052 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
25055 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
25057 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
25058 operands in a special way. Assert that pos+len <= mode precision.
25060 2017-01-05 Jakub Jelinek <jakub@redhat.com>
25062 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
25063 3 argument Alias with unlimited for the negative form.
25064 (fno-vect-cost-model): Removed.
25066 2017-01-05 Martin Liska <mliska@suse.cz>
25068 * hsa-gen.c (gen_hsa_divmod): New function.
25069 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
25071 2017-01-05 Martin Liska <mliska@suse.cz>
25074 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
25077 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25079 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
25080 small constant length operands.
25082 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25084 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
25085 between loop iterations.
25087 2017-01-05 Martin Liska <mliska@suse.cz>
25090 * gimplify.c (gimplify_decl_expr): Compare to
25091 asan_poisoned_variables instread of checking flags.
25092 (gimplify_target_expr): Likewise.
25093 (gimplify_expr): Likewise.
25094 (gimplify_function_tree): Conditionally initialize
25095 asan_poisoned_variables.
25097 2017-01-04 Jeff Law <law@redhat.com>
25099 PR tree-optimizatin/78812
25100 * rtl.h (contains_mem_rtx_p): Prototype.
25101 * ifcvt.c (containts_mem_rtx_p): Move from here to...
25102 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
25103 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
25104 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
25105 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
25107 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25109 * input.c (assert_char_at_range): Default-initialize actual_range.
25111 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25113 * df-scan.c (df_ref_create_structure): Make regno unsigned,
25114 to match the caller.
25116 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25118 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
25119 insns after final jump in test to emit dummy move.
25121 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25123 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
25124 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
25126 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
25128 * multiple_target.c (create_dispatcher_calls): Init e_next.
25129 * tree-ssa-loop-split.c (split_loop): Init border.
25130 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
25133 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
25138 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
25139 (altivec_register_operand): Do not return true if the operand
25140 contains a SUBREG mixing SImode and SFmode.
25141 (vsx_register_operand): Likewise.
25142 (vsx_reg_sfsubreg_ok): New predicate.
25143 (vfloat_operand): Do not return true if the operand contains a
25144 SUBREG mixing SImode and SFmode.
25145 (vint_operand): Likewise.
25146 (vlogical_operand): Likewise.
25147 (gpc_reg_operand): Likewise.
25148 (int_reg_operand): Likewise.
25149 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
25150 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
25151 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
25153 (rs6000_emit_move_si_sf_subreg): New helper function.
25154 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
25155 fixup SUBREGs involving SImode and SFmode.
25156 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
25157 numbers for the new peephole2 optimization.
25158 (peephole2 for SFmode unions): New peephole2 to optimize cases in
25159 the GLIBC math library that do AND/IOR/XOR operations on single
25160 precision floating point.
25161 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
25162 target macros to say whether we need to avoid SUBREGs mixing
25164 (TARGET_ALLOW_SF_SUBREG): Likewise.
25165 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
25166 (UNSPEC_SI_FROM_SF): Likewise.
25167 (iorxor): Change spacing.
25168 (and_ior_xor): New iterator for AND, IOR, and XOR.
25169 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
25170 (movdi_from_sf_zero_ext): Likewise.
25171 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
25172 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
25173 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
25174 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
25175 (fms<mode>4): Likewise.
25176 (fnma<mode>4): Likewise.
25177 (fnms<mode>4): Likewise.
25178 (nfma<mode>4): Likewise.
25179 (nfms<mode>4): Likewise.
25181 2017-01-04 Marek Polacek <polacek@redhat.com>
25184 * doc/invoke.texi: Document -Wpointer-compare.
25186 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25188 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
25191 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
25192 descriptions for -gdwarf-5 and emit them as uleb128 instead of
25195 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
25198 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
25199 documentation of the powerpc_popcntb_ok attribute.
25200 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
25201 code to issue warning messages if a requested CPU configuration is
25202 not supported by the binary (assembler and loader) toolchain.
25203 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
25204 made to define a built-in function that has been disabled.
25205 (paired_init_builtins): Add assertion to prevent ICE if attempt is
25206 made to define a built-in function that has been disabled.
25207 (altivec_init_builtins): Add comment explaining why definition
25208 of the DST built-in functions is not preceded by an assertion
25209 check. Add assertions to prevent ICE if attempts are made to
25210 define an altivec predicate or an abs* built-in function that has
25212 (htm_init_builtins): Add comment explaining why definition of the
25213 htm built-in functions is not preceded by an assertion check.
25215 2017-01-04 Jeff Law <law@redhat.com>
25217 PR tree-optimizatin/67955
25218 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
25219 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
25220 the points-to solution does not include pt_null. Use DECL_PT_UID
25223 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
25225 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
25226 Use gen_int_mode instead of gen_lopwart for const_int operands.
25228 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25230 PR tree-optimization/71563
25231 * match.pd: Simplify X << Y into X if Y is known to be 0 or
25232 out of range value - has low bits known to be zero.
25234 2017-01-04 Alan Modra <amodra@gmail.com>
25236 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
25237 * configure: Regenerate.
25238 * config.in: Regenerate.
25240 2017-01-04 Jakub Jelinek <jakub@redhat.com>
25243 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
25244 a substring of the message, but strcmp with the whole message. Ifdef
25245 ENABLE_NLS, translate the message first using dgettext.
25247 2017-01-03 Jeff Law <law@redhat.com>
25249 PR tree-optimizatin/78856
25250 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
25251 (mark_threaded_blocks): Remove code to truncate thread paths that
25252 cross multiple loop headers. Instead invalidate the cached loop
25253 iteration information and handle case of a thread path walking
25254 into an irreducible region.
25256 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
25259 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
25260 assertions. Add support for doing the signbit if the IEEE 128-bit
25261 floating point value is in a GPR.
25262 * config/rs6000/rs6000.md (Fsignbit): Delete.
25263 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
25264 Update the length attribute if the value is in a GPR.
25265 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
25266 the sign or zero extension instruction, since the value is always 0/1.
25267 (signbit<mode>2_dm2): Delete using <Fsignbit>.
25270 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
25271 extracting SImode to a GPR register so that we can generate a
25272 store, limit the vector to be in a traditional Altivec register
25273 for the vextuwrx instruction.
25275 2017-01-03 Ian Lance Taylor <iant@google.com>
25277 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
25279 2017-01-03 Martin Sebor <msebor@redhat.com>
25281 PR tree-optimization/78696
25282 * gimple-ssa-sprintf.c (format_floating): Correct handling of
25283 precision. Use MPFR for %f for greater fidelity. Correct handling
25285 (pass_sprintf_length::compute_format_length): Set width and precision
25286 specified by asrerisk to void_node for vararg functions.
25287 (try_substitute_return_value): Adjust dump output.
25289 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
25291 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
25293 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
25295 * doc/invoke.texi (SPARC options): Document -mlra as the default.
25296 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
25297 -mlra/-mno-lra was passed to the compiler.
25299 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
25301 PR rtl-optimization/65618
25302 * emit-rtl.c (try_split): Move initialization of "before" and
25303 "after" to just before the call to emit_insn_after_setloc.
25305 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
25307 * doc/md.texi (Standard Names): Remove reference to Java frontend.
25309 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
25311 * dwarf2out.c (gen_enumeration_type_die): When
25312 -gno-strict-dwarf, add a DW_AT_encoding attribute.
25314 2017-01-03 Jakub Jelinek <jakub@redhat.com>
25316 PR tree-optimization/78965
25317 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
25318 Change first argument from const call_info & to call_info &. For %n
25319 set info.nowrite to false.
25321 PR middle-end/78901
25322 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
25323 possibly throwing calls.
25325 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
25326 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
25327 and fns handling, rather than in a separate case SSA_NAME.
25329 2017-01-02 Jeff Law <law@redhat.com>
25331 * config/darwin-driver.c (darwin_driver_init): Const-correctness
25332 fixes for first_period and second_period variables.
25334 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
25337 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
25338 (*insvqi_1): New insn pattern.
25339 (*insvqi_1_mem_rex64): Ditto.
25340 (*insvqi_2): Ditto.
25341 (*insvqi_3): Rename from *insvqi.
25343 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
25345 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
25347 * doc/cfg.texi (Edges): Remove reference to Java.
25348 (Maintaining the CFG): Ditto.
25350 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25352 PR middle-end/77674
25353 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
25354 transparent aliases.
25356 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
25358 PR middle-end/77484
25359 * predict.def (PRED_CALL): Update hitrate.
25360 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
25361 * predict.c (tree_estimate_probability_bb): Split CALL predictor
25362 into direct/indirect/polymorphic variants.
25364 2017-01-01 Jakub Jelinek <jakub@redhat.com>
25366 Update copyright years.
25368 * gcc.c (process_command): Update copyright notice dates.
25369 * gcov-dump.c (print_version): Ditto.
25370 * gcov.c (print_version): Ditto.
25371 * gcov-tool.c (print_version): Ditto.
25372 * gengtype.c (create_file): Ditto.
25373 * doc/cpp.texi: Bump @copying's copyright year.
25374 * doc/cppinternals.texi: Ditto.
25375 * doc/gcc.texi: Ditto.
25376 * doc/gccint.texi: Ditto.
25377 * doc/gcov.texi: Ditto.
25378 * doc/install.texi: Ditto.
25379 * doc/invoke.texi: Ditto.
25381 Copyright (C) 2017 Free Software Foundation, Inc.
25383 Copying and distribution of this file, with or without modification,
25384 are permitted in any medium without royalty provided the copyright
25385 notice and this notice are preserved.