1 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
2 Alan Hayward <alan.hayward@arm.com>
3 David Sherwood <david.sherwood@arm.com>
5 * expmed.c (extract_high_half): Use scalar_int_mode and remove
7 (expmed_mult_highpart_optab): Likewise.
8 (expmed_mult_highpart): Likewise.
10 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11 Alan Hayward <alan.hayward@arm.com>
12 David Sherwood <david.sherwood@arm.com>
14 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
15 instead of a machine_mode.
16 (builtin_memset_read_str): Likewise.
17 * builtins.c (c_readstr): Likewise.
18 (builtin_memcpy_read_str): Likewise.
19 (builtin_strncpy_read_str): Likewise.
20 (builtin_memset_read_str): Likewise.
21 (builtin_memset_gen_str): Likewise.
22 (expand_builtin_signbit): Use scalar_int_mode for local variables.
23 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
24 instead of a machine_mode.
25 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
27 (make_extraction): Likewise.
28 (try_widen_shift_mode): Take and return scalar_int_modes instead
30 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
31 a scalar_int_mode instead of a machine_mode.
32 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
33 (avr_addr_space_pointer_mode): Likewise.
34 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
35 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
36 (msp430_unwind_word_mode): Likewise.
37 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
38 (spu_addr_space_pointer_mode): Likewise.
39 (spu_addr_space_address_mode): Likewise.
40 (spu_libgcc_cmp_return_mode): Likewise.
41 (spu_libgcc_shift_count_mode): Likewise.
42 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
43 (rl78_addr_space_pointer_mode): Likewise.
44 (fl78_unwind_word_mode): Likewise.
45 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
47 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
48 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
49 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
50 (mips_valid_pointer_mode): Likewise.
51 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
52 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
53 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
55 (ft32_addr_space_address_mode): Likewise.
56 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
57 scalar_int_mode instead of a machine_mode.
58 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
60 (m32c_addr_space_address_mode): Likewise.
61 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
62 (rs6000_eh_return_filter_mode): Likewise.
63 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
64 (rs6000_eh_return_filter_mode): Likewise.
65 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
66 (s390_libgcc_shift_count_mode): Likewise.
67 (s390_unwind_word_mode): Likewise.
68 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
70 * target.def (mode_rep_extended): Likewise.
71 (valid_pointer_mode): Likewise.
72 (addr_space.valid_pointer_mode): Likewise.
73 (eh_return_filter_mode): Return a scalar_int_mode rather than
75 (libgcc_cmp_return_mode): Likewise.
76 (libgcc_shift_count_mode): Likewise.
77 (unwind_word_mode): Likewise.
78 (addr_space.pointer_mode): Likewise.
79 (addr_space.address_mode): Likewise.
80 * doc/tm.texi: Regenerate.
81 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
83 (do_jump): Use scalar_int_mode for local variables.
84 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
85 rather than a machine_mode.
86 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
87 (scompare_loc_descriptor_wide): Likewise.
88 (scompare_loc_descriptor_narrow): Likewise.
89 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
91 * except.c (sjlj_emit_dispatch_table): Likewise.
92 (expand_builtin_eh_copy_values): Likewise.
93 * explow.c (convert_memory_address_addr_space_1): Likewise.
94 Take a scalar_int_mode rather than a machine_mode.
95 (convert_memory_address_addr_space): Take a scalar_int_mode rather
97 (memory_address_addr_space): Use scalar_int_mode for local variables.
98 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
99 rather than a machine_mode.
100 * expmed.c (mask_rtx): Likewise.
101 (init_expmed_one_conv): Likewise.
102 (expand_mult_highpart_adjust): Likewise.
103 (extract_high_half): Likewise.
104 (expmed_mult_highpart_optab): Likewise.
105 (expmed_mult_highpart): Likewise.
106 (expand_smod_pow2): Likewise.
107 (expand_sdiv_pow2): Likewise.
108 (emit_store_flag_int): Likewise.
109 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
111 (extract_low_bits): Likewise.
112 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
114 * expr.c (pieces_addr::adjust): Likewise.
115 (can_store_by_pieces): Likewise.
116 (store_by_pieces): Likewise.
117 (clear_by_pieces_1): Likewise.
118 (expand_expr_addr_expr_1): Likewise.
119 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
120 (expand_expr_real_1): Likewise.
121 (try_casesi): Likewise.
122 * final.c (shorten_branches): Likewise.
123 * fold-const.c (fold_convert_const_int_from_fixed): Change the
124 type of "mode" to machine_mode.
125 * internal-fn.c (expand_arith_overflow_result_store): Take a
126 scalar_int_mode rather than a machine_mode.
127 (expand_mul_overflow): Use scalar_int_mode for local variables.
128 * loop-doloop.c (doloop_modify): Likewise.
129 (doloop_optimize): Likewise.
130 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
132 (expand_doubleword_shift_condmove): Likewise.
133 (expand_doubleword_shift): Likewise.
134 (expand_doubleword_clz): Likewise.
135 (expand_doubleword_popcount): Likewise.
136 (expand_doubleword_parity): Likewise.
137 (expand_absneg_bit): Use scalar_int_mode for local variables.
138 (prepare_float_lib_cmp): Likewise.
139 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
140 rather than a machine_mode.
141 (convert_memory_address_addr_space): Likewise.
142 (get_mode_bounds): Likewise.
143 (get_address_mode): Return a scalar_int_mode rather than a
145 * rtlanal.c (get_address_mode): Likewise.
146 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
148 * targhooks.c (default_mode_rep_extended): Likewise.
149 (default_valid_pointer_mode): Likewise.
150 (default_addr_space_valid_pointer_mode): Likewise.
151 (default_eh_return_filter_mode): Return a scalar_int_mode rather
153 (default_libgcc_cmp_return_mode): Likewise.
154 (default_libgcc_shift_count_mode): Likewise.
155 (default_unwind_word_mode): Likewise.
156 (default_addr_space_pointer_mode): Likewise.
157 (default_addr_space_address_mode): Likewise.
158 * targhooks.h (default_eh_return_filter_mode): Likewise.
159 (default_libgcc_cmp_return_mode): Likewise.
160 (default_libgcc_shift_count_mode): Likewise.
161 (default_unwind_word_mode): Likewise.
162 (default_addr_space_pointer_mode): Likewise.
163 (default_addr_space_address_mode): Likewise.
164 (default_mode_rep_extended): Take a scalar_int_mode rather than
166 (default_valid_pointer_mode): Likewise.
167 (default_addr_space_valid_pointer_mode): Likewise.
168 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
170 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
171 rather than a machine_mode.
172 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
174 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
175 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
179 Alan Hayward <alan.hayward@arm.com>
180 David Sherwood <david.sherwood@arm.com>
182 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
183 the mode argument to scalar_int_mode.
184 (do_jump_by_parts_zero_rtx): Likewise.
185 (do_jump_by_parts_equality_rtx): Likewise.
186 (do_jump_by_parts_greater): Take a mode argument.
187 (do_jump_by_parts_equality): Likewise.
188 (do_jump_1): Update calls accordingly.
190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
191 Alan Hayward <alan.hayward@arm.com>
192 David Sherwood <david.sherwood@arm.com>
194 * is-a.h (safe_dyn_cast): New function.
195 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
196 (jump_table_for_label): Likewise.
197 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
198 instead of an rtx_insn *.
199 (shorten_branches): Use dyn_cast instead of LABEL_P and
200 JUMP_TABLE_DATA_P. Use jump_table_for_label and
201 rtx_jump_table_data::get_data_mode.
202 (final_scan_insn): Likewise.
204 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
205 Alan Hayward <alan.hayward@arm.com>
206 David Sherwood <david.sherwood@arm.com>
208 * combine.c (try_combine): Use is_a <scalar_int_mode> when
209 trying to combine a full-register integer set with a subreg
212 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
213 Alan Hayward <alan.hayward@arm.com>
214 David Sherwood <david.sherwood@arm.com>
216 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
217 that is always either address_mode or pointer_mode.
219 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
220 Alan Hayward <alan.hayward@arm.com>
221 David Sherwood <david.sherwood@arm.com>
223 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
224 when the two are known to be equal.
226 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
227 Alan Hayward <alan.hayward@arm.com>
228 David Sherwood <david.sherwood@arm.com>
230 * simplify-rtx.c (simplify_const_unary_operation): Use
231 is_a <scalar_int_mode> instead of checking for a nonzero
232 precision. Forcibly convert op_mode to a scalar_int_mode
233 in that case. More clearly differentiate the operand and
234 result modes and use the former when deciding what the value
235 of a count-bits operation should be. Use is_int_mode instead
236 of checking for a MODE_INT. Remove redundant check for whether
237 this mode has a zero precision.
239 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
240 Alan Hayward <alan.hayward@arm.com>
241 David Sherwood <david.sherwood@arm.com>
243 * optabs.c (widen_leading): Change the type of the mode argument
244 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
245 (widen_bswap): Likewise.
246 (expand_parity): Likewise.
247 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
248 (expand_ffs): Likewise.
249 (epand_unop): Check for scalar integer modes before calling the
252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
253 Alan Hayward <alan.hayward@arm.com>
254 David Sherwood <david.sherwood@arm.com>
256 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
258 (expand_expr_real_1): Update call accordingly.
260 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
261 Alan Hayward <alan.hayward@arm.com>
262 David Sherwood <david.sherwood@arm.com>
264 * expmed.c (store_bit_field_using_insv): Add op0_mode and
265 value_mode arguments. Use scalar_int_mode internally.
266 (store_bit_field_1): Rename the new integer mode from imode
267 to op0_mode and use it instead of GET_MODE (op0). Update calls
268 to store_split_bit_field, store_bit_field_using_insv and
269 store_fixed_bit_field.
270 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
271 Use scalar_int_mode internally. Use a bit count rather than a mode
272 when calculating the largest bit size for get_best_mode.
273 Update calls to store_split_bit_field and store_fixed_bit_field_1.
274 (store_fixed_bit_field_1): Add mode and value_mode arguments.
275 Remove assertion that OP0 has a scalar integer mode.
276 (store_split_bit_field): Add op0_mode and value_mode arguments.
277 Update calls to extract_fixed_bit_field.
278 (extract_bit_field_using_extv): Add an op0_mode argument.
279 Use scalar_int_mode internally.
280 (extract_bit_field_1): Rename the new integer mode from imode to
281 op0_mode and use it instead of GET_MODE (op0). Update calls to
282 extract_split_bit_field, extract_bit_field_using_extv and
283 extract_fixed_bit_field.
284 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
285 to extract_split_bit_field and extract_fixed_bit_field_1.
286 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
287 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
289 (extract_split_bit_field): Add an op0_mode argument. Update call
290 to extract_fixed_bit_field.
292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
293 Alan Hayward <alan.hayward@arm.com>
294 David Sherwood <david.sherwood@arm.com>
296 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
297 * explow.c (hard_function_value): Likewise.
298 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
299 convert_to_mode call outside the loop.
300 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
301 for the mode iterator. Require the mode specified by max_pieces
303 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
305 (copy_blkmode_to_reg): Likewise.
306 (set_storage_via_setmem): Likewise.
307 * optabs.c (prepare_cmp_insn): Likewise.
308 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
309 * stor-layout.c (finish_bitfield_representative): Likewise.
311 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
312 Alan Hayward <alan.hayward@arm.com>
313 David Sherwood <david.sherwood@arm.com>
315 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
316 * expr.c (convert_move): Use them.
317 (convert_modes): Likewise.
318 (store_expr_with_bounds): Likewise.
320 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
321 Alan Hayward <alan.hayward@arm.com>
322 David Sherwood <david.sherwood@arm.com>
324 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
325 parameter for the mode of "x". Remove the "known_x", "known_mode"
326 and "known_ret" arguments. Change the type of the mode argument
328 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
329 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
330 (reg_num_sign_bit_copies_for_combine): Likewise.
331 * rtlanal.c (nonzero_bits1): Likewise.
332 (num_sign_bit_copies1): Likewise.
333 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
334 (reg_num_sign_bit_copies_general): Likewise.
335 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
336 (reg_nonzero_bits_general): Likewise.
338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
339 Alan Hayward <alan.hayward@arm.com>
340 David Sherwood <david.sherwood@arm.com>
342 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
343 than in subroutines. Return 1 for non-integer modes.
344 (cached_num_sign_bit_copies): Change the type of the mode parameter
346 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
347 classes. Handle CONST_INT_P first and then check whether X also
348 has a scalar integer mode. Check the same thing for inner registers
349 of a SUBREG and for values that are being extended or truncated.
351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
352 Alan Hayward <alan.hayward@arm.com>
353 David Sherwood <david.sherwood@arm.com>
355 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
356 in subroutines. Return the mode mask for non-integer modes.
357 (cached_nonzero_bits): Change the type of the mode parameter
359 (nonzero_bits1): Likewise. Remove early exit for other mode
360 classes. Handle CONST_INT_P first and then check whether X
361 also has a scalar integer mode.
363 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
364 Alan Hayward <alan.hayward@arm.com>
365 David Sherwood <david.sherwood@arm.com>
367 * combine.c (extract_left_shift): Add a mode argument and update
369 (make_compound_operation_int): Change the type of the mode parameter
370 to scalar_int_mode and update the call to extract_left_shift.
372 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
373 Alan Hayward <alan.hayward@arm.com>
374 David Sherwood <david.sherwood@arm.com>
376 * combine.c (simplify_and_const_int): Change the type of the mode
377 parameter to scalar_int_mode.
378 (simplify_and_const_int_1): Likewise. Update recursive call.
380 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
381 Alan Hayward <alan.hayward@arm.com>
382 David Sherwood <david.sherwood@arm.com>
384 * combine.c (simplify_compare_const): Check that the mode is a
385 scalar_int_mode (rather than VOIDmode) before testing its
387 (simplify_comparison): Move COMPARISON_P handling out of the
388 loop and restrict the latter part of the loop to scalar_int_modes.
389 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
390 and when considering SUBREG_REGs. Use is_int_mode instead of
391 checking GET_MODE_CLASS against MODE_INT.
393 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
394 Alan Hayward <alan.hayward@arm.com>
395 David Sherwood <david.sherwood@arm.com>
397 * combine.c (try_widen_shift_mode): Move check for equal modes to...
398 (simplify_shift_const_1): ...here. Use scalar_int_mode for
399 shift_unit_mode and for modes involved in scalar shifts.
401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
402 Alan Hayward <alan.hayward@arm.com>
403 David Sherwood <david.sherwood@arm.com>
405 * combine.c (force_int_to_mode): New function, split out from...
406 (force_to_mode): ...here. Keep xmode up-to-date and use it
407 instead of GET_MODE (x).
409 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
410 Alan Hayward <alan.hayward@arm.com>
411 David Sherwood <david.sherwood@arm.com>
413 * optabs-query.h (extraction_insn::struct_mode): Change type to
414 opt_scalar_int_mode and update comment.
415 (extraction_insn::field_mode): Change type to scalar_int_mode.
416 (extraction_insn::pos_mode): Likewise.
417 * combine.c (make_extraction): Update accordingly.
418 * optabs-query.c (get_traditional_extraction_insn): Likewise.
419 (get_optab_extraction_insn): Likewise.
420 * recog.c (simplify_while_replacing): Likewise.
421 * expmed.c (narrow_bit_field_mem): Change the type of the mode
422 parameter to opt_scalar_int_mode.
424 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
425 Alan Hayward <alan.hayward@arm.com>
426 David Sherwood <david.sherwood@arm.com>
428 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
429 to a scalar_int_mode instead of a machine_mode.
430 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
431 (get_best_mode): Return a boolean and use a pointer argument to store
432 the selected mode. Replace the limit mode parameter with a bit limit.
433 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
434 for the values returned by bit_field_mode_iterator::next_mode.
435 (store_bit_field): Update call to get_best_mode.
436 (store_fixed_bit_field): Likewise.
437 (extract_fixed_bit_field): Likewise.
438 * expr.c (optimize_bitfield_assignment_op): Likewise.
439 * fold-const.c (optimize_bit_field_compare): Likewise.
440 (fold_truth_andor_1): Likewise.
441 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
442 Update for new type of m_mode.
443 (get_best_mode): As above.
445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
446 Alan Hayward <alan.hayward@arm.com>
447 David Sherwood <david.sherwood@arm.com>
449 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
450 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
451 (store_bit_field): Check is_a <scalar_int_mode> before calling
452 strict_volatile_bitfield_p.
453 (extract_bit_field): Likewise.
455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
456 Alan Hayward <alan.hayward@arm.com>
457 David Sherwood <david.sherwood@arm.com>
459 * target.def (cstore_mode): Return a scalar_int_mode.
460 * doc/tm.texi: Regenerate.
461 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
462 * targhooks.h (default_cstore_mode): Likewise.
463 * targhooks.c (default_cstore_mode): Likewise, using a forced
465 * expmed.c (emit_cstore): Expect the target of the cstore to be
468 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
469 Alan Hayward <alan.hayward@arm.com>
470 David Sherwood <david.sherwood@arm.com>
472 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
473 * builtins.c (expand_builtin_signbit): Use it.
474 * cfgexpand.c (expand_debug_expr): Likewise.
475 * dojump.c (do_jump): Likewise.
476 (do_compare_and_jump): Likewise.
477 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
478 * expmed.c (make_tree): Likewise.
479 * expr.c (expand_expr_real_2): Likewise.
480 (expand_expr_real_1): Likewise.
481 (try_casesi): Likewise.
482 * fold-const-call.c (fold_const_call_ss): Likewise.
483 * fold-const.c (unextend): Likewise.
484 (extract_muldiv_1): Likewise.
485 (fold_single_bit_test): Likewise.
486 (native_encode_int): Likewise.
487 (native_encode_string): Likewise.
488 (native_interpret_int): Likewise.
489 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
490 * internal-fn.c (expand_addsub_overflow): Likewise.
491 (expand_neg_overflow): Likewise.
492 (expand_mul_overflow): Likewise.
493 (expand_arith_overflow): Likewise.
494 * match.pd: Likewise.
495 * stor-layout.c (layout_type): Likewise.
496 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
497 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
498 * tree-ssanames.c (get_range_info): Likewise.
499 * tree-switch-conversion.c (array_value_type) Likewise.
500 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
501 (vect_recog_divmod_pattern): Likewise.
502 (vect_recog_mixed_size_cond_pattern): Likewise.
503 * tree-vrp.c (extract_range_basic): Likewise.
504 (simplify_float_conversion_using_ranges): Likewise.
505 * tree.c (int_fits_type_p): Likewise.
506 * ubsan.c (instrument_bool_enum_load): Likewise.
507 * varasm.c (mergeable_string_section): Likewise.
508 (narrowing_initializer_constant_valid_p): Likewise.
509 (output_constant): Likewise.
511 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
512 Alan Hayward <alan.hayward@arm.com>
513 David Sherwood <david.sherwood@arm.com>
515 * machmode.h (NARROWEST_INT_MODE): New macro.
516 * expr.c (alignment_for_piecewise_move): Use it instead of
517 GET_CLASS_NARROWEST_MODE (MODE_INT).
518 (push_block): Likewise.
519 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
521 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
523 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
524 Alan Hayward <alan.hayward@arm.com>
525 David Sherwood <david.sherwood@arm.com>
527 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
529 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
530 * optabs.c (expand_unop): Likewise.
532 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
533 Alan Hayward <alan.hayward@arm.com>
534 David Sherwood <david.sherwood@arm.com>
536 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
537 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
538 (popcount_loc_descriptor): Likewise.
539 (bswap_loc_descriptor): Likewise.
540 (rotate_loc_descriptor): Likewise.
541 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
542 calling the functions above.
544 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
545 Alan Hayward <alan.hayward@arm.com>
546 David Sherwood <david.sherwood@arm.com>
548 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
550 (try_combine): Likewise.
551 (simplify_if_then_else): Likewise.
552 * cse.c (cse_insn): Likewise.
553 * dwarf2out.c (mem_loc_descriptor): Likewise.
554 * emit-rtl.c (gen_lowpart_common): Likewise.
555 * simplify-rtx.c (simplify_truncation): Likewise.
556 (simplify_binary_operation_1): Likewise.
557 (simplify_const_relational_operation): Likewise.
558 (simplify_ternary_operation): Likewise.
559 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
561 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
562 Alan Hayward <alan.hayward@arm.com>
563 David Sherwood <david.sherwood@arm.com>
565 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
566 * reload.c (push_reload): Likewise.
567 (find_reloads): Likewise.
569 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
570 Alan Hayward <alan.hayward@arm.com>
571 David Sherwood <david.sherwood@arm.com>
573 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
574 (make_compound_operation_int): Likewise.
575 (change_zero_ext): Likewise.
576 * expr.c (convert_move): Likewise.
577 (convert_modes): Likewise.
578 * fwprop.c (forward_propagate_subreg): Likewise.
579 * loop-iv.c (get_biv_step_1): Likewise.
580 * optabs.c (widen_operand): Likewise.
581 * postreload.c (move2add_valid_value_p): Likewise.
582 * recog.c (simplify_while_replacing): Likewise.
583 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
584 (simplify_binary_operation_1): Likewise. Remove redundant
587 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
588 Alan Hayward <alan.hayward@arm.com>
589 David Sherwood <david.sherwood@arm.com>
591 * combine.c (combine_simplify_rtx): Add checks for
592 is_a <scalar_int_mode>.
593 (simplify_if_then_else): Likewise.
594 (make_field_assignment): Likewise.
595 (simplify_comparison): Likewise.
596 * ifcvt.c (noce_try_bitop): Likewise.
597 * loop-invariant.c (canonicalize_address_mult): Likewise.
598 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
600 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
601 Alan Hayward <alan.hayward@arm.com>
602 David Sherwood <david.sherwood@arm.com>
604 * simplify-rtx.c (simplify_binary_operation_1): Use
605 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
607 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
608 Alan Hayward <alan.hayward@arm.com>
609 David Sherwood <david.sherwood@arm.com>
611 * wide-int.h (int_traits<unsigned char>) New class.
612 (int_traits<unsigned short>) Likewise.
613 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
614 Use GET_MODE_UNIT_PRECISION and remove redundant test for
616 * combine.c (set_nonzero_bits_and_sign_copies): Use
617 is_a <scalar_int_mode>.
618 (find_split_point): Likewise.
619 (combine_simplify_rtx): Likewise.
620 (simplify_logical): Likewise.
621 (expand_compound_operation): Likewise.
622 (expand_field_assignment): Likewise.
623 (make_compound_operation): Likewise.
624 (extended_count): Likewise.
625 (change_zero_ext): Likewise.
626 (simplify_comparison): Likewise.
627 * dwarf2out.c (scompare_loc_descriptor): Likewise.
628 (ucompare_loc_descriptor): Likewise.
629 (minmax_loc_descriptor): Likewise.
630 (mem_loc_descriptor): Likewise.
631 (loc_descriptor): Likewise.
632 * expmed.c (init_expmed_one_mode): Likewise.
633 * lra-constraints.c (lra_constraint_offset): Likewise.
634 * optabs.c (prepare_libcall_arg): Likewise.
635 * postreload.c (move2add_note_store): Likewise.
636 * reload.c (operands_match_p): Likewise.
637 * rtl.h (load_extend_op): Likewise.
638 * rtlhooks.c (gen_lowpart_general): Likewise.
639 * simplify-rtx.c (simplify_truncation): Likewise.
640 (simplify_unary_operation_1): Likewise.
641 (simplify_binary_operation_1): Likewise.
642 (simplify_const_binary_operation): Likewise.
643 (simplify_const_relational_operation): Likewise.
644 (simplify_subreg): Likewise.
645 * stor-layout.c (bitwise_mode_for_mode): Likewise.
646 * var-tracking.c (adjust_mems): Likewise.
647 (prepare_call_arguments): Likewise.
649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
650 Alan Hayward <alan.hayward@arm.com>
651 David Sherwood <david.sherwood@arm.com>
653 * machmode.h (is_int_mode): New fuction.
654 * combine.c (find_split_point): Use it.
655 (combine_simplify_rtx): Likewise.
656 (simplify_if_then_else): Likewise.
657 (simplify_set): Likewise.
658 (simplify_shift_const_1): Likewise.
659 (simplify_comparison): Likewise.
660 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
661 * cse.c (notreg_cost): Likewise.
662 (cse_insn): Likewise.
663 * cselib.c (cselib_lookup_1): Likewise.
664 * dojump.c (do_jump_1): Likewise.
665 (do_compare_rtx_and_jump): Likewise.
666 * dse.c (get_call_args): Likewise.
667 * dwarf2out.c (rtl_for_decl_init): Likewise.
668 (native_encode_initializer): Likewise.
669 * expmed.c (emit_store_flag_1): Likewise.
670 (emit_store_flag): Likewise.
671 * expr.c (convert_modes): Likewise.
672 (store_field): Likewise.
673 (expand_expr_real_1): Likewise.
674 * fold-const.c (fold_read_from_constant_string): Likewise.
675 * gimple-ssa-sprintf.c (get_format_string): Likewise.
676 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
677 * optabs.c (expand_binop): Likewise.
678 (expand_unop): Likewise.
679 (expand_abs_nojump): Likewise.
680 (expand_one_cmpl_abs_nojump): Likewise.
681 * simplify-rtx.c (mode_signbit_p): Likewise.
682 (val_signbit_p): Likewise.
683 (val_signbit_known_set_p): Likewise.
684 (val_signbit_known_clear_p): Likewise.
685 (simplify_relational_operation_1): Likewise.
686 * tree.c (vector_type_mode): Likewise.
688 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
689 Alan Hayward <alan.hayward@arm.com>
690 David Sherwood <david.sherwood@arm.com>
692 * machmode.h (smallest_mode_for_size): Fix formatting.
693 (smallest_int_mode_for_size): New function.
694 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
695 instead of smallest_mode_for_size.
696 * combine.c (make_extraction): Likewise.
697 * config/arc/arc.c (arc_expand_movmem): Likewise.
698 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
699 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
700 * config/s390/s390.c (s390_expand_insv): Likewise.
701 * config/sparc/sparc.c (assign_int_registers): Likewise.
702 * config/spu/spu.c (spu_function_value): Likewise.
703 (spu_function_arg): Likewise.
704 * coverage.c (get_gcov_type): Likewise.
705 (get_gcov_unsigned_t): Likewise.
706 * dse.c (find_shift_sequence): Likewise.
707 * expmed.c (store_bit_field_1): Likewise.
708 * expr.c (convert_move): Likewise.
709 (store_field): Likewise.
710 * internal-fn.c (expand_arith_overflow): Likewise.
711 * optabs-query.c (get_best_extraction_insn): Likewise.
712 * optabs.c (expand_twoval_binop_libfunc): Likewise.
713 * stor-layout.c (layout_type): Likewise.
714 (initialize_sizetypes): Likewise.
715 * targhooks.c (default_get_mask_mode): Likewise.
716 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
718 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
719 Alan Hayward <alan.hayward@arm.com>
720 David Sherwood <david.sherwood@arm.com>
722 * machmode.h (opt_mode::else_blk): New function.
723 (int_mode_for_mode): Declare.
724 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
725 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
727 * cfgexpand.c (expand_debug_expr): Likewise.
728 * combine.c (gen_lowpart_or_truncate): Likewise.
729 (gen_lowpart_for_combine): Likewise.
730 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
731 * config/avr/avr.c (avr_to_int_mode): Likewise.
732 (avr_out_plus_1): Likewise.
733 (avr_out_plus): Likewise.
734 (avr_out_round): Likewise.
735 * config/i386/i386.c (ix86_split_to_parts): Likewise.
736 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
737 (s390_expand_vcond): Likewise.
738 * config/spu/spu.c (spu_split_immediate): Likewise.
739 (spu_expand_mov): Likewise.
740 * dse.c (get_stored_val): Likewise.
741 * expmed.c (store_bit_field_1): Likewise.
742 (convert_extracted_bit_field): Use int_mode_for_mode instead of
744 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
745 (extract_low_bits): Likewise.
746 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
747 handling rather than repeating the check.
748 (emit_group_store): Likewise.
749 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
750 * optabs.c (expand_absneg_bit): Likewise.
751 (expand_copysign_absneg): Likewise.
752 (expand_copysign_bit): Likewise.
753 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
754 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
755 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
756 * var-tracking.c (prepare_call_arguments): Likewise.
757 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
758 int_mode_for_mode instead of mode_for_size.
759 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
761 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
762 Alan Hayward <alan.hayward@arm.com>
763 David Sherwood <david.sherwood@arm.com>
765 * machmode.h (int_mode_for_size): New function.
766 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
767 instead of mode_for_size.
768 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
770 * combine.c (expand_field_assignment): Use int_mode_for_size
771 instead of mode_for_size.
772 (make_extraction): Likewise.
773 (simplify_shift_const_1): Likewise.
774 (simplify_comparison): Likewise.
775 * dojump.c (do_jump): Likewise.
776 * dwarf2out.c (mem_loc_descriptor): Likewise.
777 * emit-rtl.c (init_derived_machine_modes): Likewise.
778 * expmed.c (flip_storage_order): Likewise.
779 (convert_extracted_bit_field): Likewise.
780 * expr.c (copy_blkmode_from_reg): Likewise.
781 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
782 * internal-fn.c (expand_mul_overflow): Likewise.
783 * lower-subreg.c (simple_move): Likewise.
784 * optabs-libfuncs.c (init_optabs): Likewise.
785 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
786 * tree.c (vector_type_mode): Likewise.
787 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
788 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
789 * tree-vect-generic.c (expand_vector_parallel): Likewise.
790 * tree-vect-stmts.c (vectorizable_load): Likewise.
791 (vectorizable_store): Likewise.
793 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
794 Alan Hayward <alan.hayward@arm.com>
795 David Sherwood <david.sherwood@arm.com>
797 * coretypes.h (pod_mode): New type.
798 (scalar_int_mode_pod): New typedef.
799 * machmode.h (pod_mode): New class.
800 (int_n_data_t::m): Change type to scalar_int_mode_pod.
801 * genmodes.c (emit_mode_int_n): Update accordingly.
802 * lower-subreg.h (target_lower_subreg): Change type to
804 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
807 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
808 Alan Hayward <alan.hayward@arm.com>
809 David Sherwood <david.sherwood@arm.com>
811 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
812 machine_mode to scalar_int_mode.
813 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
814 (rs6000_option_override_internal): Remove cast to int.
815 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
816 machine_mode to scalar_int_mode.
817 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
818 (rs6000_option_override_internal): Remove cast to int.
819 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
820 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
822 * config/s390/s390.c (s390_expand_builtin): Likewise.
823 * coretypes.h (scalar_int_mode): New type.
824 (opt_scalar_int_mode): New typedef.
825 * machmode.h (scalar_int_mode): New class.
826 (scalar_int_mode::includes_p): New function.
827 (byte_mode): Change type to scalar_int_mode.
828 (word_mode): Likewise.
829 (ptr_mode): Likewise.
830 * emit-rtl.c (byte_mode): Likewise.
831 (word_mode): Likewise.
832 (ptr_mode): Likewise.
833 (init_derived_machine_modes): Update accordingly.
834 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
835 and MODE_PARTIAL_INT.
836 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
839 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
840 Alan Hayward <alan.hayward@arm.com>
841 David Sherwood <david.sherwood@arm.com>
843 * target.def (libgcc_floating_mode_supported_p): Take a
845 * doc/tm.texi: Regenerate.
846 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
848 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
849 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
852 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
853 Alan Hayward <alan.hayward@arm.com>
854 David Sherwood <david.sherwood@arm.com>
856 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
857 * doc/tm.texi: Regenerate.
858 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
859 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
860 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
861 * targhooks.h (default_floatn_mode): Likewise.
862 * targhooks.c (default_floatn_mode): Likewise.
863 * tree.c (build_common_tree_nodes): Update accordingly.
865 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
866 Alan Hayward <alan.hayward@arm.com>
867 David Sherwood <david.sherwood@arm.com>
869 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
870 (mode_iterator::iterate_p): Likewise.
871 (mode_iterator::get_wider): Likewise.
872 * expr.c (init_expr_target): Use opt_scalar_float_mode.
874 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
875 Alan Hayward <alan.hayward@arm.com>
876 David Sherwood <david.sherwood@arm.com>
878 * coretypes.h (opt_scalar_float_mode): New typedef.
879 * machmode.h (float_mode_for_size): New function.
880 * emit-rtl.c (double_mode): Delete.
881 (init_emit_once): Use float_mode_for_size.
882 * stor-layout.c (layout_type): Likewise.
883 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
886 Alan Hayward <alan.hayward@arm.com>
887 David Sherwood <david.sherwood@arm.com>
889 * output.h (assemble_real): Take a scalar_float_mode.
890 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
891 * config/arm/arm.md (consttable_4): Likewise.
892 (consttable_8): Likewise.
893 (consttable_16): Likewise.
894 * config/mips/mips.md (consttable_float): Likewise.
895 * config/s390/s390.c (s390_output_pool_entry): Likewise.
896 * varasm.c (assemble_real): Take a scalar_float_mode.
897 (output_constant_pool_2): Update accordingly.
898 (output_constant): Likewise.
900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
901 Alan Hayward <alan.hayward@arm.com>
902 David Sherwood <david.sherwood@arm.com>
904 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
905 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
906 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
907 (native_encode_real): Likewise.
908 (native_interpret_real): Likewise.
909 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
910 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
913 Alan Hayward <alan.hayward@arm.com>
914 David Sherwood <david.sherwood@arm.com>
916 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
917 <scalar_float_mode>. Simplify.
918 (gen_extend_conv_libfunc): Likewise.
920 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
921 Alan Hayward <alan.hayward@arm.com>
922 David Sherwood <david.sherwood@arm.com>
924 * coretypes.h (scalar_float_mode): New type.
925 * machmode.h (mode_traits::from_int): Use machine_mode if
926 USE_ENUM_MODES is defined.
927 (is_a): New function.
929 (dyn_cast): Likewise.
930 (scalar_float_mode): New class.
931 (scalar_float_mode::includes_p): New function.
932 (is_float_mode): Likewise.
933 * gdbhooks.py (MachineModePrinter): New class.
934 (build_pretty_printer): Use it for scalar_float_mode.
935 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
936 (format_helper::format_helper): Turn into a template.
937 * genmodes.c (get_mode_class): New function.
938 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
939 or machine_mode if none.
940 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
941 as_a <scalar_float_mode>.
942 * dwarf2out.c (mem_loc_descriptor): Likewise.
943 (insert_float): Likewise.
944 (add_const_value_attribute): Likewise.
945 * simplify-rtx.c (simplify_immed_subreg): Likewise.
946 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
947 (expand_unop): Update accordingly.
948 (expand_abs_nojump): Likewise.
949 (expand_copysign_absneg): Take a scalar_float_mode.
950 (expand_copysign_bit): Likewise.
951 (expand_copysign): Update accordingly.
953 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
954 Alan Hayward <alan.hayward@arm.com>
955 David Sherwood <david.sherwood@arm.com>
957 * coretypes.h (opt_mode): New class.
958 * machmode.h (opt_mode): Likewise.
959 (opt_mode::else_void): New function.
960 (opt_mode::require): Likewise.
961 (opt_mode::exists): Likewise.
962 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
963 (GET_MODE_2XWIDER_MODE): Likewise.
964 (mode_iterator::get_wider): Update accordingly.
965 (mode_iterator::get_2xwider): Likewise.
966 (mode_iterator::get_known_wider): Likewise, turning into a template.
967 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
968 forcing a wider mode to exist.
969 * config/cr16/cr16.h (LONG_REG_P): Likewise.
970 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
971 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
972 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
973 * lower-subreg.c (init_lower_subreg): Likewise.
974 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
975 on the final iteration.
976 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
977 a wider mode exists before asking for a move pattern.
978 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
979 forcing a wider mode to exist.
980 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
981 returning false if no such mode exists.
982 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
983 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
984 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
985 Avoid checking for a MODE_INT if we already know the mode is not a
987 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
988 forcing a wider mode to exist.
989 (expmed_mult_highpart_optab): Likewise.
990 (expmed_mult_highpart): Likewise.
991 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
993 * lto-streamer-in.c (lto_input_mode_table): Likewise.
994 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
995 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
996 * internal-fn.c (expand_mul_overflow): Update use of
997 GET_MODE_2XWIDER_MODE.
998 * omp-low.c (omp_clause_aligned_alignment): Likewise.
999 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
1000 GET_MODE_WIDER_MODE.
1001 (convert_plusminus_to_widen): Likewise.
1002 * tree-switch-conversion.c (array_value_type): Likewise.
1003 * var-tracking.c (emit_note_insn_var_location): Likewise.
1004 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1005 Return false inside rather than outside the loop if no wider mode
1007 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
1008 and GET_MODE_2XWIDER_MODE
1009 (can_compare_p): Use else_void.
1010 * gdbhooks.py (OptMachineModePrinter): New class.
1011 (build_pretty_printer): Use it for opt_mode.
1013 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1014 Alan Hayward <alan.hayward@arm.com>
1015 David Sherwood <david.sherwood@arm.com>
1017 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
1018 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
1020 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1021 Alan Hayward <alan.hayward@arm.com>
1022 David Sherwood <david.sherwood@arm.com>
1024 * machmode.h (mode_traits): New structure.
1025 (get_narrowest_mode): New function.
1026 (mode_iterator::start): Likewise.
1027 (mode_iterator::iterate_p): Likewise.
1028 (mode_iterator::get_wider): Likewise.
1029 (mode_iterator::get_known_wider): Likewise.
1030 (mode_iterator::get_2xwider): Likewise.
1031 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
1032 (FOR_EACH_MODE): Likewise.
1033 (FOR_EACH_MODE_FROM): Likewise.
1034 (FOR_EACH_MODE_UNTIL): Likewise.
1035 (FOR_EACH_WIDER_MODE): Likewise.
1036 (FOR_EACH_2XWIDER_MODE): Likewise.
1037 * builtins.c (expand_builtin_strlen): Use new mode iterators.
1038 * combine.c (simplify_comparison): Likewise
1039 * config/i386/i386.c (type_natural_mode): Likewise.
1040 * cse.c (cse_insn): Likewise.
1041 * dse.c (find_shift_sequence): Likewise.
1042 * emit-rtl.c (init_derived_machine_modes): Likewise.
1043 (init_emit_once): Likewise.
1044 * explow.c (hard_function_value): Likewise.
1045 * expmed.c (extract_fixed_bit_field_1): Likewise.
1046 (extract_bit_field_1): Likewise.
1047 (expand_divmod): Likewise.
1048 (emit_store_flag_1): Likewise.
1049 * expr.c (init_expr_target): Likewise.
1050 (convert_move): Likewise.
1051 (alignment_for_piecewise_move): Likewise.
1052 (widest_int_mode_for_size): Likewise.
1053 (emit_block_move_via_movmem): Likewise.
1054 (copy_blkmode_to_reg): Likewise.
1055 (set_storage_via_setmem): Likewise.
1056 (compress_float_constant): Likewise.
1057 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1058 * optabs-query.c (get_best_extraction_insn): Likewise.
1059 * optabs.c (expand_binop): Likewise.
1060 (expand_twoval_unop): Likewise.
1061 (expand_twoval_binop): Likewise.
1062 (widen_leading): Likewise.
1063 (widen_bswap): Likewise.
1064 (expand_parity): Likewise.
1065 (expand_unop): Likewise.
1066 (prepare_cmp_insn): Likewise.
1067 (prepare_float_lib_cmp): Likewise.
1068 (expand_float): Likewise.
1069 (expand_fix): Likewise.
1070 (expand_sfix_optab): Likewise.
1071 * postreload.c (move2add_use_add2_insn): Likewise.
1072 * reg-stack.c (reg_to_stack): Likewise.
1073 * reginfo.c (choose_hard_reg_mode): Likewise.
1074 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1075 * stor-layout.c (mode_for_size): Likewise.
1076 (smallest_mode_for_size): Likewise.
1077 (mode_for_vector): Likewise.
1078 (finish_bitfield_representative): Likewise.
1079 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
1080 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
1081 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1082 * var-tracking.c (prepare_call_arguments): Likewise.
1084 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1085 Alan Hayward <alan.hayward@arm.com>
1086 David Sherwood <david.sherwood@arm.com>
1088 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
1089 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
1090 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
1091 * machmode.h (mode_size): Move earlier in file.
1092 (mode_precision): Likewise.
1093 (mode_inner): Likewise.
1094 (mode_nunits): Likewise.
1095 (mode_unit_size): Likewise.
1096 (unit_unit_precision): Likewise.
1097 (mode_wider): Likewise.
1098 (mode_2xwider): Likewise.
1099 (machine_mode): New class.
1100 (mode_to_bytes): New function.
1101 (mode_to_bits): Likewise.
1102 (mode_to_precision): Likewise.
1103 (mode_to_inner): Likewise.
1104 (mode_to_unit_size): Likewise.
1105 (mode_to_unit_precision): Likewise.
1106 (mode_to_nunits): Likewise.
1107 (GET_MODE_SIZE): Use mode_to_bytes.
1108 (GET_MODE_BITSIZE): Use mode_to_bits.
1109 (GET_MODE_PRECISION): Use mode_to_precision.
1110 (GET_MODE_INNER): Use mode_to_inner.
1111 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
1112 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
1113 (GET_MODE_NUNITS): Use mode_to_nunits.
1114 * system.h (ALWAYS_INLINE): New macro.
1115 * config/powerpcspe/powerpcspe-c.c
1116 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
1117 int for arg1_mode and arg2_mode.
1119 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1120 Alan Hayward <alan.hayward@arm.com>
1121 David Sherwood <david.sherwood@arm.com>
1123 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
1124 Prefix mode names with E_ in case statements.
1125 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1126 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
1127 (aarch64_split_simd_move): Likewise.
1128 (aarch64_gen_storewb_pair): Likewise.
1129 (aarch64_gen_loadwb_pair): Likewise.
1130 (aarch64_gen_store_pair): Likewise.
1131 (aarch64_gen_load_pair): Likewise.
1132 (aarch64_get_condition_code_1): Likewise.
1133 (aarch64_constant_pool_reload_icode): Likewise.
1134 (get_rsqrte_type): Likewise.
1135 (get_rsqrts_type): Likewise.
1136 (get_recpe_type): Likewise.
1137 (get_recps_type): Likewise.
1138 (aarch64_gimplify_va_arg_expr): Likewise.
1139 (aarch64_simd_container_mode): Likewise.
1140 (aarch64_emit_load_exclusive): Likewise.
1141 (aarch64_emit_store_exclusive): Likewise.
1142 (aarch64_expand_compare_and_swap): Likewise.
1143 (aarch64_gen_atomic_cas): Likewise.
1144 (aarch64_emit_bic): Likewise.
1145 (aarch64_emit_atomic_swap): Likewise.
1146 (aarch64_emit_atomic_load_op): Likewise.
1147 (aarch64_evpc_trn): Likewise.
1148 (aarch64_evpc_uzp): Likewise.
1149 (aarch64_evpc_zip): Likewise.
1150 (aarch64_evpc_ext): Likewise.
1151 (aarch64_evpc_rev): Likewise.
1152 (aarch64_evpc_dup): Likewise.
1153 (aarch64_gen_ccmp_first): Likewise.
1154 (aarch64_gen_ccmp_next): Likewise.
1155 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1156 (alpha_emit_xfloating_libcall): Likewise.
1157 (emit_insxl): Likewise.
1158 (alpha_arg_type): Likewise.
1159 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
1160 (arc_preferred_simd_mode): Likewise.
1161 (arc_secondary_reload): Likewise.
1162 (get_arc_condition_code): Likewise.
1163 (arc_print_operand): Likewise.
1164 (arc_legitimate_constant_p): Likewise.
1165 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1166 * config/arc/arc.md (casesi_load): Likewise.
1167 (casesi_compact_jump): Likewise.
1168 * config/arc/predicates.md (proper_comparison_operator): Likewise.
1169 (cc_use_register): Likewise.
1170 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1171 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
1172 (arm_init_iwmmxt_builtins): Likewise.
1173 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
1174 (neon_expand_vector_init): Likewise.
1175 (arm_attr_length_move_neon): Likewise.
1176 (maybe_get_arm_condition_code): Likewise.
1177 (arm_emit_vector_const): Likewise.
1178 (arm_preferred_simd_mode): Likewise.
1179 (arm_output_iwmmxt_tinsr): Likewise.
1180 (thumb1_output_casesi): Likewise.
1181 (thumb2_output_casesi): Likewise.
1182 (arm_emit_load_exclusive): Likewise.
1183 (arm_emit_store_exclusive): Likewise.
1184 (arm_expand_compare_and_swap): Likewise.
1185 (arm_evpc_neon_vuzp): Likewise.
1186 (arm_evpc_neon_vzip): Likewise.
1187 (arm_evpc_neon_vrev): Likewise.
1188 (arm_evpc_neon_vtrn): Likewise.
1189 (arm_evpc_neon_vext): Likewise.
1190 (arm_validize_comparison): Likewise.
1191 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
1192 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
1193 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
1194 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
1195 (c6x_preferred_simd_mode): Likewise.
1196 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
1197 (epiphany_rtx_costs): Likewise.
1198 * config/epiphany/predicates.md (proper_comparison_operator):
1200 * config/frv/frv.c (condexec_memory_operand): Likewise.
1201 (frv_emit_move): Likewise.
1202 (output_move_single): Likewise.
1203 (output_condmove_single): Likewise.
1204 (frv_hard_regno_mode_ok): Likewise.
1205 (frv_matching_accg_mode): Likewise.
1206 * config/h8300/h8300.c (split_adds_subs): Likewise.
1207 (h8300_rtx_costs): Likewise.
1208 (h8300_print_operand): Likewise.
1209 (compute_mov_length): Likewise.
1210 (output_logical_op): Likewise.
1211 (compute_logical_op_length): Likewise.
1212 (compute_logical_op_cc): Likewise.
1213 (h8300_shift_needs_scratch_p): Likewise.
1214 (output_a_shift): Likewise.
1215 (compute_a_shift_length): Likewise.
1216 (compute_a_shift_cc): Likewise.
1217 (expand_a_rotate): Likewise.
1218 (output_a_rotate): Likewise.
1219 * config/i386/i386.c (classify_argument): Likewise.
1220 (function_arg_advance_32): Likewise.
1221 (function_arg_32): Likewise.
1222 (function_arg_64): Likewise.
1223 (function_value_64): Likewise.
1224 (ix86_gimplify_va_arg): Likewise.
1225 (ix86_legitimate_constant_p): Likewise.
1226 (put_condition_code): Likewise.
1227 (split_double_mode): Likewise.
1228 (ix86_avx256_split_vector_move_misalign): Likewise.
1229 (ix86_expand_vector_logical_operator): Likewise.
1230 (ix86_split_idivmod): Likewise.
1231 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
1232 (ix86_build_const_vector): Likewise.
1233 (ix86_build_signbit_mask): Likewise.
1234 (ix86_match_ccmode): Likewise.
1235 (ix86_cc_modes_compatible): Likewise.
1236 (ix86_expand_branch): Likewise.
1237 (ix86_expand_sse_cmp): Likewise.
1238 (ix86_expand_sse_movcc): Likewise.
1239 (ix86_expand_int_sse_cmp): Likewise.
1240 (ix86_expand_vec_perm_vpermi2): Likewise.
1241 (ix86_expand_vec_perm): Likewise.
1242 (ix86_expand_sse_unpack): Likewise.
1243 (ix86_expand_int_addcc): Likewise.
1244 (ix86_split_to_parts): Likewise.
1245 (ix86_vectorize_builtin_gather): Likewise.
1246 (ix86_vectorize_builtin_scatter): Likewise.
1247 (avx_vpermilp_parallel): Likewise.
1248 (inline_memory_move_cost): Likewise.
1249 (ix86_tieable_integer_mode_p): Likewise.
1250 (x86_maybe_negate_const_int): Likewise.
1251 (ix86_expand_vector_init_duplicate): Likewise.
1252 (ix86_expand_vector_init_one_nonzero): Likewise.
1253 (ix86_expand_vector_init_one_var): Likewise.
1254 (ix86_expand_vector_init_concat): Likewise.
1255 (ix86_expand_vector_init_interleave): Likewise.
1256 (ix86_expand_vector_init_general): Likewise.
1257 (ix86_expand_vector_set): Likewise.
1258 (ix86_expand_vector_extract): Likewise.
1259 (emit_reduc_half): Likewise.
1260 (ix86_emit_i387_round): Likewise.
1261 (ix86_mangle_type): Likewise.
1262 (ix86_expand_round_sse4): Likewise.
1263 (expand_vec_perm_blend): Likewise.
1264 (canonicalize_vector_int_perm): Likewise.
1265 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
1266 (expand_vec_perm_1): Likewise.
1267 (expand_vec_perm_interleave3): Likewise.
1268 (expand_vec_perm_even_odd_pack): Likewise.
1269 (expand_vec_perm_even_odd_1): Likewise.
1270 (expand_vec_perm_broadcast_1): Likewise.
1271 (ix86_vectorize_vec_perm_const_ok): Likewise.
1272 (ix86_expand_vecop_qihi): Likewise.
1273 (ix86_expand_mul_widen_hilo): Likewise.
1274 (ix86_expand_sse2_abs): Likewise.
1275 (ix86_expand_pextr): Likewise.
1276 (ix86_expand_pinsr): Likewise.
1277 (ix86_preferred_simd_mode): Likewise.
1278 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
1279 * config/i386/sse.md (*andnot<mode>3): Likewise.
1280 (<mask_codefor><code><mode>3<mask_name>): Likewise.
1281 (*<code><mode>3): Likewise.
1282 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
1283 (ia64_expand_atomic_op): Likewise.
1284 (ia64_arg_type): Likewise.
1285 (ia64_mode_to_int): Likewise.
1286 (ia64_scalar_mode_supported_p): Likewise.
1287 (ia64_vector_mode_supported_p): Likewise.
1288 (expand_vec_perm_broadcast): Likewise.
1289 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
1290 (iq2000_function_arg_advance): Likewise.
1291 (iq2000_function_arg): Likewise.
1292 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
1293 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
1294 (m68k_libcall_value): Likewise.
1295 (m68k_function_value): Likewise.
1296 (sched_attr_op_type): Likewise.
1297 * config/mcore/mcore.c (mcore_output_move): Likewise.
1298 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
1300 (microblaze_function_arg): Likewise.
1301 * config/mips/mips.c (mips16_build_call_stub): Likewise.
1302 (mips_print_operand): Likewise.
1303 (mips_mode_ok_for_mov_fmt_p): Likewise.
1304 (mips_vector_mode_supported_p): Likewise.
1305 (mips_preferred_simd_mode): Likewise.
1306 (mips_expand_vpc_loongson_even_odd): Likewise.
1307 (mips_expand_vec_unpack): Likewise.
1308 (mips_expand_vi_broadcast): Likewise.
1309 (mips_expand_vector_init): Likewise.
1310 (mips_expand_vec_reduc): Likewise.
1311 (mips_expand_msa_cmp): Likewise.
1312 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
1313 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
1314 (cc_flags_for_mode): Likewise.
1315 * config/msp430/msp430.c (msp430_print_operand): Likewise.
1316 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
1317 (nds32_output_casesi_pc_relative): Likewise.
1318 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1319 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
1320 (nvptx_gen_unpack): Likewise.
1321 (nvptx_gen_pack): Likewise.
1322 (nvptx_gen_shuffle): Likewise.
1323 (nvptx_gen_wcast): Likewise.
1324 (nvptx_preferred_simd_mode): Likewise.
1325 * config/pa/pa.c (pa_secondary_reload): Likewise.
1326 * config/pa/predicates.md (base14_operand): Likewise.
1327 * config/powerpcspe/powerpcspe-c.c
1328 (altivec_resolve_overloaded_builtin): Likewise.
1329 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
1331 (rs6000_preferred_simd_mode): Likewise.
1332 (output_vec_const_move): Likewise.
1333 (rs6000_expand_vector_extract): Likewise.
1334 (rs6000_split_vec_extract_var): Likewise.
1335 (reg_offset_addressing_ok_p): Likewise.
1336 (rs6000_legitimate_offset_address_p): Likewise.
1337 (rs6000_legitimize_address): Likewise.
1338 (rs6000_emit_set_const): Likewise.
1339 (rs6000_const_vec): Likewise.
1340 (rs6000_emit_move): Likewise.
1341 (spe_build_register_parallel): Likewise.
1342 (rs6000_darwin64_record_arg_recurse): Likewise.
1343 (swap_selector_for_mode): Likewise.
1344 (spe_init_builtins): Likewise.
1345 (paired_init_builtins): Likewise.
1346 (altivec_init_builtins): Likewise.
1347 (do_load_for_compare): Likewise.
1348 (rs6000_generate_compare): Likewise.
1349 (rs6000_expand_float128_convert): Likewise.
1350 (emit_load_locked): Likewise.
1351 (emit_store_conditional): Likewise.
1352 (rs6000_output_function_epilogue): Likewise.
1353 (rs6000_handle_altivec_attribute): Likewise.
1354 (rs6000_function_value): Likewise.
1355 (emit_fusion_gpr_load): Likewise.
1356 (emit_fusion_p9_load): Likewise.
1357 (emit_fusion_p9_store): Likewise.
1358 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
1359 (fusion_gpr_mem_load): Likewise.
1360 (fusion_addis_mem_combo_load): Likewise.
1361 (fusion_addis_mem_combo_store): Likewise.
1362 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
1363 (fusion_gpr_mem_load): Likewise.
1364 (fusion_addis_mem_combo_load): Likewise.
1365 (fusion_addis_mem_combo_store): Likewise.
1366 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1368 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
1369 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
1370 (rs6000_preferred_simd_mode): Likewise.
1371 (output_vec_const_move): Likewise.
1372 (rs6000_expand_vector_extract): Likewise.
1373 (rs6000_split_vec_extract_var): Likewise.
1374 (reg_offset_addressing_ok_p): Likewise.
1375 (rs6000_legitimate_offset_address_p): Likewise.
1376 (rs6000_legitimize_address): Likewise.
1377 (rs6000_emit_set_const): Likewise.
1378 (rs6000_const_vec): Likewise.
1379 (rs6000_emit_move): Likewise.
1380 (rs6000_darwin64_record_arg_recurse): Likewise.
1381 (swap_selector_for_mode): Likewise.
1382 (paired_init_builtins): Likewise.
1383 (altivec_init_builtins): Likewise.
1384 (rs6000_expand_float128_convert): Likewise.
1385 (emit_load_locked): Likewise.
1386 (emit_store_conditional): Likewise.
1387 (rs6000_output_function_epilogue): Likewise.
1388 (rs6000_handle_altivec_attribute): Likewise.
1389 (rs6000_function_value): Likewise.
1390 (emit_fusion_gpr_load): Likewise.
1391 (emit_fusion_p9_load): Likewise.
1392 (emit_fusion_p9_store): Likewise.
1393 * config/rx/rx.c (rx_gen_move_template): Likewise.
1394 (flags_from_mode): Likewise.
1395 * config/s390/predicates.md (s390_alc_comparison): Likewise.
1396 (s390_slb_comparison): Likewise.
1397 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
1398 (s390_vector_mode_supported_p): Likewise.
1399 (s390_cc_modes_compatible): Likewise.
1400 (s390_match_ccmode_set): Likewise.
1401 (s390_canonicalize_comparison): Likewise.
1402 (s390_emit_compare_and_swap): Likewise.
1403 (s390_branch_condition_mask): Likewise.
1404 (s390_rtx_costs): Likewise.
1405 (s390_secondary_reload): Likewise.
1406 (__SECONDARY_RELOAD_CASE): Likewise.
1407 (s390_expand_cs): Likewise.
1408 (s390_preferred_simd_mode): Likewise.
1409 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
1410 * config/sh/sh.c (sh_print_operand): Likewise.
1411 (dump_table): Likewise.
1412 (sh_secondary_reload): Likewise.
1413 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1414 * config/sh/sh.md (casesi_worker_1): Likewise.
1415 (casesi_worker_2): Likewise.
1416 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
1417 (fcc_comparison_operator): Likewise.
1418 * config/sparc/sparc.c (sparc_expand_move): Likewise.
1419 (emit_soft_tfmode_cvt): Likewise.
1420 (sparc_preferred_simd_mode): Likewise.
1421 (output_cbranch): Likewise.
1422 (sparc_print_operand): Likewise.
1423 (sparc_expand_vec_perm_bmask): Likewise.
1424 (vector_init_bshuffle): Likewise.
1425 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1426 (spu_vector_mode_supported_p): Likewise.
1427 (spu_expand_insv): Likewise.
1428 (spu_emit_branch_or_set): Likewise.
1429 (spu_handle_vector_attribute): Likewise.
1430 (spu_builtin_splats): Likewise.
1431 (spu_builtin_extract): Likewise.
1432 (spu_builtin_promote): Likewise.
1433 (spu_expand_sign_extend): Likewise.
1434 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1435 (tilegx_simd_int): Likewise.
1436 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
1437 (tilepro_simd_int): Likewise.
1438 * config/v850/v850.c (const_double_split): Likewise.
1439 (v850_print_operand): Likewise.
1440 (ep_memory_offset): Likewise.
1441 * config/vax/vax.c (vax_rtx_costs): Likewise.
1442 (vax_output_int_move): Likewise.
1443 (vax_output_int_add): Likewise.
1444 (vax_output_int_subtract): Likewise.
1445 * config/visium/predicates.md (visium_branch_operator): Likewise.
1446 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
1447 (visium_print_operand_address): Likewise.
1448 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1449 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
1450 (xtensa_expand_conditional_branch): Likewise.
1451 (xtensa_copy_incoming_a7): Likewise.
1452 (xtensa_output_literal): Likewise.
1453 * dfp.c (decimal_real_maxval): Likewise.
1454 * targhooks.c (default_libgcc_floating_mode_supported_p): 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 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
1461 (mode_nunits_inline): Likewise.
1462 (mode_inner_inline): Likewise.
1463 (mode_unit_size_inline): Likewise.
1464 (mode_unit_precision_inline): Likewise.
1465 (emit_insn_modes_h): Likewise. Also emit a #define of the
1467 (emit_mode_wider): Add an E_ prefix to mode names.
1468 (emit_mode_complex): Likewise.
1469 (emit_mode_inner): Likewise.
1470 (emit_mode_adjustments): Likewise.
1471 (emit_mode_int_n): Likewise.
1472 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
1473 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
1474 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
1475 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
1476 (CRC32_BUILTIN, ENTRY): Likewise.
1477 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
1478 (aarch64_pop_regs): Likewise.
1479 (aarch64_process_components): Likewise.
1480 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
1481 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
1482 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
1483 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
1484 * config/arm/arm.c (arm_init_libfuncs): Likewise.
1485 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
1487 * config/i386/i386-builtin.def (pcmpestr): Likewise.
1488 (pcmpistr): Likewise.
1489 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
1490 * config/mmix/mmix.c (mmix_output_condition): Likewise.
1491 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
1493 * config/rl78/rl78.c (mduc_regs): Likewise.
1494 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
1495 (htm_expand_builtin): Likewise.
1496 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
1497 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
1498 * config/xtensa/xtensa.c (print_operand): Likewise.
1499 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
1500 (NUM_MODE_VECTOR_INT): Likewise.
1501 * genoutput.c (null_operand): Likewise.
1502 (output_operand_data): Likewise.
1503 * genrecog.c (print_parameter_value): Likewise.
1504 * lra.c (debug_operand_data): Likewise.
1506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1508 * dwarf2out.c (scompare_loc_descriptor_wide)
1509 (scompare_loc_descriptor_narrow): New functions, split out from...
1510 (scompare_loc_descriptor): ...here.
1511 * expmed.c (emit_store_flag_int): New function, split out from...
1512 (emit_store_flag): ...here.
1514 2017-08-30 Richard Biener <rguenther@suse.de>
1516 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
1517 (dwarf2out_early_finish): Move setting of AT_pubnames from
1518 early debug output to early finish.
1520 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
1522 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
1523 and -mdata-region to the assembler.
1525 2017-08-30 Richard Biener <rguenther@suse.de>
1527 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
1529 (gen_subprogram_die): Add DW_AT_object_pointer only early.
1530 (dwarf2out_early_global_decl): Only generate a DIE for the
1531 abstract origin if it doesn't already exist or is a declaration DIE.
1532 (resolve_addr): Do not add the linkage name twice when
1533 generating a stub DIE for the DW_TAG_GNU_call_site target.
1535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1537 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1538 Use machine_mode rather than int for arg1_mode.
1540 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
1543 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
1544 that the second argument of the built-in functions to unpack
1545 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
1546 switch statement instead a lot of if statements.
1547 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
1548 Allow 64-bit values to be in Altivec registers as well as
1549 traditional floating point registers.
1550 (pack<mode>, FMOVE128_VSX iterator): Likewise.
1552 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
1554 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
1555 MAX_REGS_PER_ADDRESS == 1.
1557 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
1559 * config/i386/i386.opt (flag_fentry): Do not init to -1.
1560 * config/i386/i386.c (ix86_option_override_internal): Simplify
1561 setting of opts->x_flag_entry.
1563 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1564 Jakub Jelinek <jakub@redhat.com>
1565 Richard Biener <rguenther@suse.de>
1567 PR tree-optimization/81503
1568 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
1569 folded constant fits in the target type; reorder tests for clarity.
1571 2017-08-29 Martin Liska <mliska@suse.cz>
1573 * passes.def: Include pass_lower_switch.
1574 * stmt.c (dump_case_nodes): Remove and move to
1575 tree-switch-conversion.
1576 (case_values_threshold): Likewise.
1577 (expand_switch_as_decision_tree_p): Likewise.
1578 (emit_case_decision_tree): Likewise.
1579 (expand_case): Likewise.
1580 (balance_case_nodes): Likewise.
1581 (node_has_low_bound): Likewise.
1582 (node_has_high_bound): Likewise.
1583 (node_is_bounded): Likewise.
1584 (emit_case_nodes): Likewise.
1585 (struct simple_case_node): New struct.
1586 (add_case_node): Remove.
1587 (emit_case_dispatch_table): Use vector instead of case_list.
1588 (reset_out_edges_aux): Remove.
1589 (compute_cases_per_edge): Likewise.
1590 (expand_case): Build list of simple_case_node.
1591 (expand_sjlj_dispatch_table): Use it.
1592 * tree-switch-conversion.c (struct case_node): Moved from
1593 stmt.c and adjusted.
1594 (emit_case_nodes): Likewise.
1595 (node_has_low_bound): Likewise.
1596 (node_has_high_bound): Likewise.
1597 (node_is_bounded): Likewise.
1598 (case_values_threshold): Likewise.
1599 (reset_out_edges_aux): Likewise.
1600 (compute_cases_per_edge): Likewise.
1601 (add_case_node): Likewise.
1602 (dump_case_nodes): Likewise.
1603 (balance_case_nodes): Likewise.
1604 (expand_switch_as_decision_tree_p): Likewise.
1605 (emit_jump): Likewise.
1606 (emit_case_decision_tree): Likewise.
1607 (try_switch_expansion): Likewise.
1608 (do_jump_if_equal): Likewise.
1609 (emit_cmp_and_jump_insns): Likewise.
1610 (fix_phi_operands_for_edge): New function.
1611 (record_phi_operand_mapping): Likewise.
1612 (class pass_lower_switch): New pass.
1613 (pass_lower_switch::execute): New function.
1614 (make_pass_lower_switch): Likewise.
1615 (conditional_probability):
1616 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
1617 * tree-pass.h: Add make_pass_lower_switch.
1619 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
1622 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
1625 2017-08-29 Richard Biener <rguenther@suse.de>
1627 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
1628 we do not add a DW_AT_inline attribute twice.
1629 (gen_subprogram_die): Remove code setting DW_AT_inline on
1630 DECL_ABSTRACT_P nodes.
1632 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
1634 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
1635 calls to internal functions.
1636 (gimplify_modify_expr): Likewise.
1637 * tree-call-cdce.c (use_internal_fn): Likewise.
1638 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1639 (convert_to_divmod): Set the nothrow flag.
1640 * tree-if-conv.c (predicate_mem_writes): Likewise.
1641 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
1642 (vectorizable_call): Likewise.
1643 (vectorizable_store): Likewise.
1644 (vectorizable_load): Likewise.
1645 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
1646 (vect_recog_mask_conversion_pattern): Likewise.
1648 2017-08-29 Martin Liska <mliska@suse.cz>
1651 * gcc.c (driver_handle_option): Add new argument.
1652 * opts-common.c (handle_option): Pass
1653 target_option_override_hook.
1654 * opts-global.c (lang_handle_option): Add new option.
1655 (set_default_handlers): Add new argument.
1656 (decode_options): Likewise.
1657 * opts.c (target_handle_option): Likewise.
1658 (common_handle_option): Call target_option_override_hook.
1659 * opts.h (struct cl_option_handler_func): Add hook for
1660 target option override.
1661 (struct cl_option_handlers): Likewise.
1662 (set_default_handlers): Add new argument.
1663 (decode_options): Likewise.
1664 (common_handle_option): Likewise.
1665 (target_handle_option): Likewise.
1666 * toplev.c (toplev::main): Pass targetm.target_option.override
1669 2017-08-29 Richard Biener <rguenther@suse.de>
1670 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1672 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
1673 life to the active subtree.
1675 2017-08-28 Jeff Law <law@redhat.com>
1677 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
1678 derive_equivalences.
1679 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
1681 (edge_info::derive_equivalences): New private member function
1683 * tree-ssa-dom.c (class edge_info): Changed from a struct
1684 to a class. Add ctor/dtor, methods and data members.
1685 (edge_info::edge_info): Renamed from allocate_edge_info.
1686 Initialize additional members.
1687 (edge_info::~edge_info): New.
1688 (free_dom_edge_info): Delete the edge info.
1689 (record_edge_info): Use new class & associated member functions.
1690 Tighten forms for testing for edge equivalences.
1691 (record_temporary_equivalences): Iterate over the simple
1692 equivalences rather than assuming there's only one per edge.
1693 (cprop_into_successor_phis): Iterate over the simple
1694 equivalences rather than assuming there's only one per edge.
1695 (optimize_stmt): Use operand_equal_p rather than pointer
1696 equality for mini-DSE code.
1698 2017-08-28 Nathan Sidwell <nathan@acm.org>
1700 * gcc.c (execute): Fold SIGPIPE handling into switch
1701 statement. Adjust internal error message.
1703 2017-08-28 Richard Biener <rguenther@suse.de>
1706 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
1707 Do nothing for removed DIEs.
1709 2017-08-28 Richard Biener <rguenther@suse.de>
1711 PR tree-optimization/81977
1712 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
1715 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
1718 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
1719 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
1720 using targetm.gen_mem_thread_fence.
1722 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
1725 * config/i386/i386.md (*<btsc><mode>): Change operand 2
1726 predicate to register_operand. Reorder operands.
1727 (*btr<mode>): Ditto.
1728 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
1729 (*btr<mode>_mask): Ditto.
1731 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
1733 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
1734 * config/rs6000/xmmintrin.h: New file.
1735 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
1737 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1740 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
1741 parameter and_insn and return it.
1742 (recombine_lvx_pattern): Insert a copy to ensure availability of
1743 the base register of the copied masking operation at the point of
1744 the instruction replacement.
1745 (recombine_stvx_pattern): Likewise.
1747 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
1749 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
1750 undocumented switches.
1751 (-mpower9-dform-vector): Likewise.
1752 (-mpower9-dform): Likewise.
1753 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
1754 comments to delete references to -mpower9-dform* switches.
1755 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
1756 Delete reference to -mpower9-dform* switches, test for
1757 -mpower9-vector instead.
1758 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
1759 (OTHER_P9_VECTOR_MASKS): Likewise.
1760 (POWERPC_MASKS): Likewise.
1761 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
1762 tests against -mpower9-dform* to -mpower9-vector. Delete code
1763 that checked for -mpower9-dform* consistancy with other options.
1764 Add test for -mpower9-misc to enable other power9 switches.
1765 (rs6000_init_hard_regno_mode_ok): Likewise.
1766 (rs6000_option_override_internal): Likewise.
1767 (rs6000_emit_prologue): Likewise.
1768 (rs6000_emit_epilogue): Likewise.
1769 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
1770 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
1771 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
1773 (emit_fusion_p9_store): Likewise.
1774 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
1775 resetting these macros if the assembler does not support ISA 3.0
1777 (TARGET_P9_DFORM_VECTOR): Likewise.
1778 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
1779 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
1781 2017-08-25 Alan Modra <amodra@gmail.com>
1783 PR rtl-optimization/81747
1784 * cse.c (cse_extended_basic_block): Don't attempt to record
1785 equivalences for degenerate conditional branches that branch
1786 to their fall-through.
1788 2017-08-24 Martin Sebor <msebor@redhat.com>
1791 * gimple-fold.c (size_must_be_zero_p): New function.
1792 (gimple_fold_builtin_memory_op): Call it.
1794 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
1796 * config/rs6000/mm_malloc.h: New file.
1798 2017-08-24 Bin Cheng <bin.cheng@arm.com>
1800 PR tree-optimization/81913
1801 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
1802 analysis when either IVs in condition can wrap.
1804 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
1806 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
1807 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
1809 2017-08-24 Richard Biener <rguenther@suse.de>
1812 * targhooks.c (default_target_can_inline_p): Properly
1813 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
1814 is present and always compare.
1815 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
1816 infer -mfpmath=sse from TARGET_SSE_P.
1817 (ix86_can_inline_p): Properly use target_option_default_node when
1818 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
1820 2017-08-24 Richard Biener <rguenther@suse.de>
1823 * dwarf2out.c (output_die): Handle flag_generate_offload like
1825 (output_comp_unit): Likewise.
1826 (gen_array_type_die): Likewise.
1827 (dwarf2out_early_finish): Likewise.
1828 (note_variable_value_in_expr): Likewise.
1829 (dwarf2out_finish): Likewise. Adjust assert.
1830 * cgraphunit.c (symbol_table::compile): Move setting of
1831 flag_generate_offload earlier ...
1832 (symbol_table::finalize_compilation_unit): ... here, before
1833 early debug finalization.
1835 2017-08-24 Richard Biener <rguenther@suse.de>
1837 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
1838 and ipa-fnsummary.h.
1839 (ix86_can_inline_p): When ix86_fpmath flags do not match
1840 check whether the callee uses FP math at all.
1842 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
1845 * tree-ssanames.c (get_nonzero_bits): Use element_precision
1846 instead of TYPE_PRECISION.
1848 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
1849 Alan Hayward <alan.hayward@arm.com>
1850 David Sherwood <david.sherwood@arm.com>
1852 * combine.c (make_extraction): Use subreg_offset_from_lsb.
1854 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
1856 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
1858 (ix86_frame::stack_realign_allocate): New field.
1859 (struct machine_frame_state): Modify comments.
1860 (machine_frame_state::sp_realigned_fp_end): New field.
1861 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
1863 (sp_valid_at): Add assertion to assure no attempt to access invalid
1864 offset of a realigned stack.
1865 (fp_valid_at): Likewise.
1866 (choose_baseaddr): Modify comments.
1867 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
1868 ix86_expand_prologue.
1869 (ix86_expand_prologue): Modify stack realignment and allocation.
1870 (ix86_expand_epilogue): Modify comments.
1871 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
1872 avx2_runtime, avx512f, and avx512f_runtime.
1874 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
1876 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
1877 (mstackrealign): Do not init to -1.
1878 * config/i386/i386.c (ix86_option_override_internal):
1879 Check opts_set, not opts when setting default value of
1880 opts->x_ix86_force_align_arg_pointer.
1882 2017-08-23 Richard Biener <rguenther@suse.de>
1884 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
1885 lang_hooks.decl_printable_name.
1886 * print-rtl-function.c (print_rtx_function): Likewise.
1887 * tree-pretty-print.c (dump_function_header): Likewise.
1889 2017-08-23 Richard Biener <rguenther@suse.de>
1892 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
1893 -g0 at compile-time.
1895 2017-08-23 Tamar Christina <tamar.christina@arm.com>
1898 * doc/sourcebuild.texi (Other hardware attributes):
1901 2017-08-23 Tamar Christina <tamar.christina@arm.com>
1904 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
1905 Fix single-use check.
1907 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
1909 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
1911 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
1913 * doc/install.texi: Modify to add more details on running selected
1916 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
1918 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
1919 is combined with -mabi=ms.
1920 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
1923 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1925 PR tree-optimization/81488
1926 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
1927 and cached_basis fields.
1928 (MAX_SPREAD): New constant.
1929 (alloc_cand_and_find_basis): Initialize new fields.
1930 (clear_visited): New function.
1931 (create_phi_basis_1): Rename from create_phi_basis, set visited
1932 and cached_basis fields.
1933 (create_phi_basis): New wrapper function.
1934 (phi_add_costs_1): Rename from phi_add_costs, add spread
1935 parameter, set visited field, short-circuit when limits reached.
1936 (phi_add_costs): New wrapper function.
1937 (record_phi_increments_1): Rename from record_phi_increments, set
1939 (record_phi_increments): New wrapper function.
1940 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
1941 (phi_incr_cost): New wrapper function.
1942 (all_phi_incrs_profitable_1): Rename from
1943 all_phi_incrs_profitable, set visited field.
1944 (all_phi_incrs_profitable): New wrapper function.
1946 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
1947 Alan Hayward <alan.hayward@arm.com>
1948 David Sherwood <david.sherwood@arm.com>
1950 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
1951 that takes the outer and inner modes.
1952 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
1953 comparison as the canonical test for a paradoxical subreg.
1954 * combine.c (simplify_set): Use paradoxical_subreg_p.
1955 (make_extraction): Likewise.
1956 (force_to_mode): Likewise.
1957 (rtx_equal_for_field_assignment_p): Likewise.
1958 (gen_lowpart_for_combine): Likewise.
1959 (simplify_comparison): Likewise.
1960 * cse.c (equiv_constant): Likewise.
1961 * expmed.c (store_bit_field_1): Likewise.
1962 * final.c (alter_subreg): Likewise.
1963 * fwprop.c (propagate_rtx): Likewise.
1964 (forward_propagate_subreg): Likewise.
1965 * ira-conflicts.c (ira_build_conflicts): Likewise.
1966 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
1967 * lra-constraints.c (curr_insn_transform): Likewise.
1968 (split_reg): Likewise.
1969 * lra-eliminations.c (move_plus_up): Likewise.
1970 (lra_eliminate_regs_1): Likewise.
1971 * recog.c (general_operand): Likewise.
1972 * ree.c (combine_reaching_defs): Likewise.
1973 * reload.c (push_reload): Likewise.
1974 (find_reloads): Likewise.
1975 * reload1.c (elimination_effects): Likewise.
1976 (compute_reload_subreg_offset): Likewise.
1977 (choose_reload_regs): Likewise.
1978 * rtlanal.c (subreg_lsb_1): Likewise.
1979 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1980 (simplify_subreg): Likewise.
1981 * var-tracking.c (track_loc_p): Likewise.
1982 * emit-rtl.c (byte_lowpart_offset): Likewise.
1983 (paradoxical_subreg_p): Delete out-of-line definition.
1985 2017-08-22 Jeff Law <law@redhat.com>
1987 PR tree-optimization/81741
1988 PR tree-optimization/71947
1989 * tree-ssa-dom.c: Include tree-inline.h.
1990 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
1991 equivalences if one is more expensive to compute than the other.
1992 * tree-ssa-scopedtables.h (class const_or_copies): Make
1993 record_const_or_copy_raw method private.
1994 (class avail_exprs_stack): New method simplify_binary_operation.
1995 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
1996 avail_exprs_stack::simplify_binary_operation as needed.
1997 (avail_exprs_stack::simplify_binary_operation): New function.
1999 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
2001 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
2002 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
2003 (DOT_SYMBOLS): Likewise.
2004 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
2005 (RELOCATABLE_NEEDS_FIXUP): Likewise.
2006 (RS6000_ABI_NAME): Likewise.
2007 (TARGET_CMODEL): Likewise.
2008 (TOC_SECTION_ASM_OP): Likewise.
2009 (SET_CMODEL): New macro.
2010 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
2012 2017-08-22 Richard Biener <rguenther@suse.de>
2014 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
2015 to allow for free-lang-data replacements similar to verify_type_variant.
2017 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
2018 Alan Hayward <alan.hayward@arm.com>
2019 David Sherwood <david.sherwood@arm.com>
2021 * config/aarch64/aarch64.md (casesi): Use DImode rather than
2022 VOIDmode for the LABEL_REF.
2024 2017-08-22 Richard Biener <rguenther@suse.de>
2026 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
2028 2017-08-22 Richard Biener <rguenther@suse.de>
2030 * common.opt (feliminate-dwarf2-dups): Ignore.
2031 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
2032 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
2033 same_die_p_wrap, compute_section_prefix,
2034 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
2035 (comdat_symbol_id, comdat_symbol_number): Likewise.
2036 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
2038 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
2039 (output_die): Mark unreachable path unreachable.
2040 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
2041 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
2042 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
2043 (dwarf2out_early_finish): Likewise.
2045 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
2047 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
2049 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
2052 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
2053 not VAR_P. Filter attribute warnings with OPT_Wattributes.
2054 (avr_attribute_table) <io, io_low, address>: Initialize
2055 .decl_required with true.
2057 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
2059 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
2060 undocumented debugging options.
2061 (-mvsx-scalar-double): Likewise.
2062 (-mallow-df-permute): Likewise.
2063 (-mvectorize-builtins): Likewise.
2064 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2065 (rs6000_builtin_vectorized_function): Likewise.
2066 (rs6000_builtin_md_vectorized_function): Likewise.
2067 (rs6000_opt_vars): Likewise.
2069 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
2072 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
2073 (*btrq_imm): Rename from *btrq.
2074 (*btcq_imm): Rename from *btcq.
2075 (btsc): New code attribute.
2076 (*<btsc><mode>): New insn pattern.
2077 (*btr<mode>): Ditto.
2078 (*<btsc><mode>_mask): New insn_and_split pattern.
2079 (*btr<mode>_mask): Ditto.
2081 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2082 Alan Hayward <alan.hayward@arm.com>
2083 David Sherwood <david.sherwood@arm.com>
2085 * function.c (pad_below): Simplify padding calculation.
2087 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2088 Alan Hayward <alan.hayward@arm.com>
2089 David Sherwood <david.sherwood@arm.com>
2091 * target.def (function_prologue): Remove frame size argument.
2092 (function_epilogue): Likewise.
2093 * doc/tm.texi: Regenerate.
2094 * final.c (final_start_function): Update call to function_prologue.
2095 (final_end_function): Update call to function_epilogue.
2096 (default_function_pro_epilogue): Remove frame size argument.
2097 * output.h (default_function_pro_epilogue): Likewise.
2098 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
2099 (arm_output_function_prologue): Likewise.
2100 * config/frv/frv.c (frv_function_prologue): Likewise.
2101 (frv_function_epilogue): Likewise.
2102 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
2103 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
2104 (ia64_output_function_epilogue): Likewise.
2105 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
2106 (m32r_output_function_epilogue): Likewise.
2107 * config/microblaze/microblaze.c (microblaze_function_prologue)
2108 (microblaze_function_epilogue): Likewise.
2109 * config/mips/mips.c (mips_output_function_prologue): Likewise.
2110 (mips_output_function_epilogue): Likewise.
2111 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
2112 (mmix_target_asm_function_epilogue): Likewise.
2113 * config/msp430/msp430.c (msp430_start_function): Likewise.
2114 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
2115 (nds32_asm_function_epilogue): Likewise.
2116 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
2117 * config/pa/pa.c (pa_output_function_prologue): Likewise.
2118 (pa_output_function_epilogue): Likewise.
2119 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
2120 (rs6000_output_function_epilogue): Likewise.
2121 * config/rl78/rl78.c (rl78_start_function): Likewise.
2122 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
2123 (rs6000_output_function_epilogue): Likewise.
2124 * config/rx/rx.c (rx_output_function_prologue): Likewise.
2125 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
2126 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
2127 (sparc_asm_function_epilogue): Likewise.
2129 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2131 * tree.h (type_has_mode_precision_p): New function.
2132 * convert.c (convert_to_integer_1): Use it.
2133 * expr.c (expand_expr_real_2): Likewise.
2134 (expand_expr_real_1): Likewise.
2135 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
2136 * match.pd: Likewise.
2137 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
2138 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
2139 * tree-tailcall.c (process_assignment): Likewise.
2140 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2141 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
2142 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
2143 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2144 (vectorizable_assignment): Likewise.
2145 (vectorizable_shift): Likewise.
2146 (vectorizable_operation): Likewise.
2147 * tree-vrp.c (register_edge_assert_for_2): Likewise.
2149 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
2151 * match.pd: Add pow (C, x) simplification.
2153 2017-08-21 Richard Biener <rguenther@suse.de>
2155 PR tree-optimization/81900
2156 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
2157 for blocks with abnormal predecessors.
2158 (compute_antic): Do not set visited flag prematurely.
2160 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
2163 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
2165 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2167 * stor-layout.h (vector_type_mode): Move to...
2168 * tree.h (vector_type_mode): ...here.
2169 * stor-layout.c (vector_type_mode): Move to...
2170 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
2172 2017-08-21 Richard Biener <rguenther@suse.de>
2174 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
2175 register_external_die hooks.
2176 (debug_false_tree_charstarstar_uhwistar): Declare.
2177 (debug_nothing_tree_charstar_uhwi): Likewise.
2178 * debug.c (do_nothing_debug_hooks): Adjust.
2179 (debug_false_tree_charstarstar_uhwistar): New do nothing.
2180 (debug_nothing_tree_charstar_uhwi): Likewise.
2181 * dbxout.c (dbx_debug_hooks): Adjust.
2182 (xcoff_debug_hooks): Likewise.
2183 * sdbout.c (sdb_debug_hooks): Likewise.
2184 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2185 * dwarf2out.c (macinfo_label_base): New global.
2186 (dwarf2out_register_external_die): New function for the
2187 register_external_die hook.
2188 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
2189 (dwarf2_debug_hooks): Use them.
2190 (dwarf2_lineno_debug_hooks): Adjust.
2191 (struct die_struct): Add with_offset flag.
2192 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
2193 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
2194 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
2195 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
2196 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
2197 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
2198 defining section names for the early LTO debug variants.
2199 (reset_indirect_string): New helper.
2200 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
2201 (print_dw_val): Add support for offsetted symbol references.
2202 (get_ultimate_context): Split out from is_cxx.
2203 (is_cxx): Use get_ultimate_context.
2204 (is_fortran): Add decl overload.
2205 (compute_comp_unit_symbol): Split out worker from
2206 compute_section_prefix.
2207 (compute_section_prefix): Call compute_comp_unit_symbol and
2208 set comdat_type_p here.
2209 (output_die): Skip DIE symbol output for the LTO added one.
2210 Handle DIE symbol references with offset.
2211 (output_comp_unit): Guard section name mangling properly.
2212 For LTO debug sections emit a symbol at the section beginning
2213 which we use to refer to its DIEs.
2214 (add_abstract_origin_attribute): For DIEs registered via
2215 dwarf2out_register_external_die directly refer to the early
2216 DIE rather than indirectly through the shadow one we created.
2217 Remove obsolete call to dwarf2out_abstract_function for
2218 non-function/block origins.
2219 (gen_array_type_die): When generating early LTO debug do
2220 not emit DW_AT_string_length.
2221 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
2222 late when in LTO. As suggested place a gcc_unreachable for
2223 the DECL_ABSTRACT_P case.
2224 (gen_subprogram_die): Avoid another specification DIE
2225 for early built declarations/definitions for the late LTO case.
2226 (gen_variable_die): Add type references for late duplicated VLA dies
2228 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
2229 we have the abstract instance already.
2230 (process_scope_var): Adjust decl DIE contexts in LTO which
2231 first puts them in limbo.
2232 (gen_decl_die): Do not generate type DIEs late apart from
2233 types for VLAs or for decls we do not yet have a DIE. Do not
2234 call dwarf2out_abstract_function late.
2235 (dwarf2out_early_global_decl): Make sure to create DIEs
2236 for abstract instances of a decl first.
2237 (dwarf2out_late_global_decl): Adjust comment.
2238 (output_macinfo_op): With multiple macro sections use
2239 macinfo_label_base to distinguish labels.
2240 (output_macinfo): Likewise. Update macinfo_label_base.
2241 Pass in the line info label.
2242 (note_variable_value_in_expr): When generating LTO resolve
2243 all variable values here by generating DIEs as needed.
2244 (init_sections_and_labels): Add early LTO debug flag parameter
2245 and generate different sections and names if set. Add generation
2246 counter for the labels so we can have multiple of them.
2247 (reset_dies): Helper to allow DIEs to be output multiple times.
2248 (dwarf2out_finish): When outputting DIEs to the fat part of an
2249 LTO object first reset DIEs.
2250 (dwarf2out_early_finish): Output early DIEs when generating LTO.
2251 (modified_type_die): Check for decl_ultimate_origin being self
2253 (gen_type_die_with_usage): Likewise.
2254 (gen_typedef_die): Allow decl_ultimate_origin being self.
2255 (set_decl_abstract_flags): Remove.
2256 (set_block_abstract_flags): Likewise.
2257 (dwarf2out_abstract_function): Treat the early generated DIEs
2258 as the abstract copy and only add DW_AT_inline and
2259 DW_AT_artificial here and call set_decl_origin_self.
2260 If the DIE has an abstract origin don't do anything.
2261 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
2262 if we have none yet (Go fails to build one, PR78628).
2263 (variably_modified_type_p): Prevent endless recursion for Ada
2264 cyclic pointer types.
2265 * lto-streamer-in.c: Include debug.h.
2266 (dref_queue): New global.
2267 (lto_read_tree_1): Stream in DIE references.
2268 (lto_input_tree): Register DIE references.
2269 (input_function): Stream DECL_DEBUG_ARGS.
2270 * lto-streamer-out.c: Include debug.h.
2271 (lto_write_tree_1): Output DIE references.
2272 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
2273 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
2274 (output_function): Stream DECL_DEBUG_ARGS.
2275 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
2276 Stream DECL_ABSTRACT_ORIGIN.
2277 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
2278 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
2279 DECL_CONTEXT for file-scope decls.
2280 * lto-streamer.h (struct dref_entry): Declare.
2281 (dref_queue): Likewise.
2282 * cfgexpand.c (pass_expand::execute): Do not call the
2283 outlining_inline_function hook here.
2284 * lto-wrapper.c (debug_obj): New global.
2285 (tool_cleanup): Unlink it if required.
2286 (debug_objcopy): New function.
2287 (run_gcc): Handle early debug sections in the IL files by
2288 extracting them to separate files, partially linkin them and
2289 feeding the result back as result to the linker.
2290 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
2291 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
2292 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
2293 sections into a separate segment.
2294 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
2296 (darwin_asm_dwarf_section): Likewise.
2297 (darwin_asm_output_dwarf_offset): Likewise.
2298 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
2300 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2301 Alan Hayward <alan.hayward@arm.com>
2302 David Sherwood <david.sherwood@arm.com>
2304 * read-md.h (md_reader::record_potential_iterator_use): Replace
2305 pointer argument with an rtx and an index.
2306 * read-rtl.c (iterator_group::apply_iterator): Likewise.
2307 (apply_mode_iterator): Likewise.
2308 (apply_code_iterator): Likewise.
2309 (apply_int_iterator): Likewise.
2310 (apply_subst_iterator): Likewise.
2311 (record_iterator_use): Likewise.
2312 (record_attribute_use): Likewise.
2313 (md_reader::record_potential_iterator_use): Likewise. Update calls
2314 to record_iterator_use and apply_iterator.
2315 (iterator_use): Replace ptr with x and index.
2316 (attribute_use): Likewise.
2317 (apply_attribute_uses): Update calls to apply_iterator.
2318 (apply_iterators): Likewise. Update initialization of iterator_use.
2319 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
2320 and record_potential_iterator_use.
2321 (rtx_reader::read_rtx_operand): Likewise.
2323 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2324 Alan Hayward <alan.hayward@arm.com>
2325 David Sherwood <david.sherwood@arm.com>
2327 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
2330 2017-08-21 Richard Biener <rguenther@suse.de>
2333 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
2334 at struct end conservatively when comparing common bases.
2336 2017-08-21 Richard Biener <rguenther@suse.de>
2338 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
2339 (mem_ref_in_stmt): Remove.
2340 (determine_max_movement): Use ref index to get at the reference.
2341 (invariantness_dom_walker::before_dom_children): Deal with
2342 lim data already initialized.
2343 (gather_mem_refs_stmt): Initialize lim data and record ref index.
2345 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2347 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
2348 (TARGET_ISA_ROUND): Ditto.
2349 (TARGET_ROUND): Ditto.
2350 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
2351 * config/i386/i386.md: Ditto.
2352 * config/i386/sse.md: Ditto.
2353 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
2354 with OPTION_MASK_ISA_SSE4_1.
2356 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2359 * doc/extend.texi (x86 Built-in Functions): Correct the name of
2360 __builtin_ia32_lzcnt_u16.
2362 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2365 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
2366 (rs6000_set_current_function): Rewrite function to use it.
2368 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
2371 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
2372 and TYPE_WARN_IF_NOT_ALIGN.
2373 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
2374 (handle_warn_if_not_align): New.
2375 (place_union_field): Call handle_warn_if_not_align.
2376 (place_field): Call handle_warn_if_not_align.
2377 Copy TYPE_WARN_IF_NOT_ALIGN.
2378 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
2379 (layout_type): Likewise.
2380 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
2382 (tree_decl_common): Add warn_if_not_align.
2383 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
2384 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
2385 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
2386 (DECL_WARN_IF_NOT_ALIGN): Likewise.
2387 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
2388 * doc/extend.texi: Document warn_if_not_aligned attribute.
2389 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
2391 2017-08-17 Martin Liska <mliska@suse.cz>
2394 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
2395 (get_data_dependence): Use it as pointer type.
2396 (distribute_loop): Likewise.
2398 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2400 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
2401 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
2402 (p8_vmrgow_<mode>_direct): New define_insn.
2403 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
2404 handle endianness for vmrgew and vmrgow permute patterns.
2406 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2408 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
2409 * config/rs6000/rs6000-cpus.def: Remove comment.
2410 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
2411 (POWERPC_MASKS): Likewise.
2412 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
2413 use of TARGET_VSX_TIMODE.
2414 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
2415 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
2416 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
2417 (rs6000_option_override_internal): Remove dead code.
2418 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
2419 (rs6000_legitimize_reload_address): Likewise.
2420 (rs6000_legitimate_address_p): Likewise.
2421 (rs6000_opt_masks): Delete "vsx-timode".
2422 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
2423 from function comment.
2424 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
2425 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
2426 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
2428 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
2429 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
2430 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
2433 (define_peephole2): Likewise.
2435 2017-08-17 Martin Sebor <msebor@redhat.com>
2438 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
2439 past the end of an array.
2440 (test_pp_format): Add test cases.
2442 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
2444 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
2445 missing ECF_NOTHROW flags.
2447 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2450 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
2451 operands residing in integer registers.
2452 (*vsx_le_perm_load_<mode>): Likewise.
2453 (*vsx_le_perm_store_<mode>): Likewise.
2454 (define_peephole2): Add peepholes to optimize the above.
2456 2017-08-17 Marek Polacek <polacek@redhat.com>
2459 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
2460 to mimic what shorten_compare did. Change the return type to bool.
2461 (fold_cond_expr_with_comparison): Update call to
2462 operand_equal_for_comparison_p.
2463 (fold_ternary_loc): Likewise.
2465 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
2467 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
2469 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
2471 2017-08-17 Richard Biener <rguenther@suse.de>
2473 * tree-ssa-structalias.c (solve_graph): When propagating
2474 to successors update the graphs succ edges and avoid duplicate work.
2476 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
2479 * config/i386/i386.c (ix86_option_override_internal): Save target
2480 specific options after ix86_stack_protector_guard_reg was changed.
2482 2017-08-17 Richard Biener <rguenther@suse.de>
2484 PR tree-optimization/81827
2485 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
2486 (new_var_info): Initialize it conservatively.
2487 (get_call_vi): Mark register vars.
2488 (new_scalar_tmp_constraint_exp): Likewise.
2489 (handle_rhs_call): Likewise.
2490 (handle_const_call): Likewise.
2491 (create_function_info_for): Likewise.
2492 (solve_constraints): Sort varinfos to separate register from
2493 non-register vars to pack points-to solution bitmaps during
2496 2017-08-17 Marek Polacek <polacek@redhat.com>
2498 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
2500 2017-08-17 Richard Biener <rguenther@suse.de>
2502 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
2503 to true when overflow is undefined and we saturated the result.
2505 2017-08-17 Alan Modra <amodra@gmail.com>
2508 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
2509 Don't use store multiple if only one reg needs saving.
2510 (interesting_frame_related_regno): New function.
2511 (rs6000_frame_related): Don't emit frame info for regs that
2513 (rs6000_emit_epilogue): Likewise.
2515 2017-08-16 Nathan Sidwell <nathan@acm.org>
2517 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
2518 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
2519 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
2520 * tree.c (free_lang_data_in_type): Use else-if chain. Always
2521 clear TYPE_LANG_1. Remove obsolete member-function stripping.
2522 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
2523 (verify_type): Adjust for TYPE_BINFO move.
2524 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
2525 process TYPE_BINFO directly.
2526 (hash_tree): Likewise.
2527 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
2529 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
2532 2017-08-16 David Malcolm <dmalcolm@redhat.com>
2534 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
2536 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
2539 * config/i386/i386.md (*anddi_1_btr): Change predicates of
2540 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
2541 Add ix86_binary_operator_ok to insn constraint.
2542 (*iordi_1_bts): Ditto.
2543 (*xordi_1_btc): Ditto.
2544 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
2545 Update corresponding peephole2 pattern.
2549 2017-08-16 Bin Cheng <bin.cheng@arm.com>
2551 PR tree-optimization/81832
2552 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
2553 copy loop header which has IFN_LOOP_DIST_ALIAS call.
2555 2017-08-16 Marek Polacek <polacek@redhat.com>
2558 * fold-const.c (fold_indirect_ref_1): Restore original behavior
2559 regarding size_zero_node.
2561 2017-08-16 Martin Liska <mliska@suse.cz>
2564 * config.gcc: Respect previously set extra_objs in case
2567 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
2569 PR tree-optimization/81835
2570 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
2571 the phi SSA_NAME. Check that the condition in a COND_EXPR does
2572 not depend on the phi.
2574 2017-08-16 Alan Modra <amodra@gmail.com>
2576 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
2579 2017-08-16 Alan Modra <amodra@gmail.com>
2581 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
2582 (save_reg_p): ..into this. Update all callers.
2583 (first_reg_to_save): Simplify.
2585 2017-08-16 Alan Modra <amodra@gmail.com>
2587 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
2590 2017-08-15 Joseph Myers <joseph@codesourcery.com>
2594 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
2595 constant count if that count is less than 32.
2597 2017-08-15 Nathan Sidwell <nathan@acm.org>
2599 * gcc.c (execute): Emit friendlier message if inferior is killed
2600 by an external cause.
2602 2017-08-15 Richard Biener <rguenther@suse.de>
2604 PR tree-optimization/81790
2605 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
2606 CONSTRUCTORs from simplifying and VN.
2608 2017-08-14 Martin Sebor <msebor@redhat.com>
2610 * builtin-attrs.def: Add comments.
2612 2017-08-14 Martin Sebor <msebor@redhat.com>
2615 * doc/extend.texi (attribute nonstring): Document new attribute.
2617 2017-08-14 Martin Sebor <msebor@redhat.com>
2620 * tree-diagnostic.c (default_tree_printer): Handle %G.
2621 * gimple-pretty-print.h (percent_G_format): Declare new function.
2622 * gimple-pretty-print.c (percent_G_format): Define.
2623 * tree-pretty-print.c (percent_K_format): Add argument.
2625 2017-08-14 Martin Sebor <msebor@redhat.com>
2627 PR translation/79998
2628 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
2629 Remove a stray space.
2631 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
2634 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
2635 (*iordi_1_bts): Ditto.
2636 (*xordi_1_btc): Ditto.
2638 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2641 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
2642 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2644 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
2645 quoted strings, and make more translator-friendly.
2646 (darwin_rs6000_override_options): Likewise.
2647 (rs6000_option_override_internal): Likewise.
2648 (rs6000_return_in_memory): Fix overlong line.
2649 (init_cmulative_args): Use quoted strings, and make more
2650 translator-friendly.
2651 (rs6000_pass_by_reference): Fix overlong line.
2652 (def_builtin): Use quoted strings.
2653 (altivec_expand_predicate_builtin): Use quoted strings, and make
2654 more translator-friendly.
2655 (htm_expand_builtin): Use quoted strings.
2656 (cpu_expand_builtin): Use quoted strings, and make more
2657 translator-friendly.
2658 (altivec_expand_builtin): Likewise.
2659 (paired_expand_predicate_builtin): Likewise.
2660 (rs6000_invalid_builtin): Likewise.
2661 (builtin_function_type): Use quoted strings.
2662 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
2663 more translator-friendly.
2664 (rs6000_trampoline_init): Likewise.
2665 (rs6000_handle_altivec_attribute): Likewise.
2666 (rs6000_inner_target_options): Use quoted strings.
2667 (rs6000_disable_incompatible_switches): Likewise.
2668 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
2669 strings, and make more translator-friendly.
2670 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
2672 2017-08-14 Bin Cheng <bin.cheng@arm.com>
2674 PR tree-optimization/81799
2675 * tree-loop-distribution.c (version_loop_by_alias_check): Force
2676 cond_expr to simple gimple operand.
2678 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
2681 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
2683 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
2687 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
2688 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
2689 TARGET_GASISR_PROLOGUES.
2690 * config/avr/avr.c (avr_option_override): Same.
2691 (avr_pass_pre_proep::execute): Same.
2693 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
2696 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
2697 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
2699 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
2701 * config/i386/i386.md (*load_tp_<mode>): Redefine as
2702 define_insn_and_split. Split to a memory load from 0 in
2703 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
2704 using PTR mode iterator.
2705 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
2706 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
2707 (*add_tp_<mode>): Redefine as define_insn_and_split.
2708 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
2709 address space. Merge with *add_tp_x32 using PTR mode iterator.
2710 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
2711 Split to an add with a memory load from 0 in
2712 DEFAULT_TLS_SEG_REG address space.
2714 2017-08-12 Andrew Pinski <apinski@cavium.com>
2716 * config/aarch64/aarch64-option-extensions.def (rdma):
2717 Fix feature string to what Linux prints out in /proc/cpuinfo.
2719 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
2722 * dwarf2out.c (modified_type_die): For C typedef types that have
2723 an ultimate origin, process the ultimate origin instead of the
2725 (gen_typedef_die): Assert that input DECLs have no ultimate
2727 (gen_type_die_with_usage): For typedef variants that have an
2728 ultimate origin, just call gen_decl_die on the original DECL.
2729 (process_scope_var): Avoid creating DIEs for local typedefs and
2730 concrete static variables.
2732 2017-08-12 Alan Modra <amodra@gmail.com>
2736 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
2737 match gnu-user.h startfile.
2738 (ENDFILE_LINUX_SPEC): Similarly.
2740 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
2743 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
2745 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
2747 2017-08-11 Tamar Christina <tamar.christina@arm.com>
2748 * config/aarch64/aarch64.md (mov<mode>): Change.
2749 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
2750 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
2751 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
2753 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
2755 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
2756 for storage order barriers.
2758 2017-08-11 Martin Liska <mliska@suse.cz>
2761 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
2762 variables of void type.
2764 2017-08-11 Martin Liska <mliska@suse.cz>
2766 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
2767 TARGET_SUPPORTS_ALIASES.
2768 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
2769 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
2770 (optimize_weakref): Likewise.
2771 * symtab.c (symtab_node::noninterposable_alias): Likewise.
2772 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
2773 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
2775 2017-08-11 Martin Liska <mliska@suse.cz>
2778 * config/i386/i386.c (make_resolver_func): Do complete
2779 refactoring of the function.
2781 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
2784 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
2785 * config/i386/i386.c (ix86_stack_protect_guard): Use
2786 ix86_stack_protect_guard_symbol_str to generate varible declaration.
2787 * doc/invoke.texi (x86 Options): Document
2788 -mstack-protector-guard-symbol= option.
2790 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
2792 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
2793 * config/i386/i386.c (ix86_split_stack_guard): New function.
2794 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
2795 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
2796 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
2797 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
2798 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
2799 (split_stack_space_check): Call ix86_split_stack_guard.
2801 2017-08-10 Martin Sebor <msebor@redhat.com>
2803 * print-tree.c (print_node): Print location using the established
2805 Replace spaces with colons.
2806 (debug_raw, debug): Ditto.
2808 2017-08-10 Martin Sebor <msebor@redhat.com>
2811 * pretty-print.c (pp_format): Correct the handling of %s precision.
2813 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
2816 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
2818 (ix86_finalize_stack_frame_flags): This. Also clear
2819 frame_pointer_needed if -fno-omit-frame-pointer is used without
2821 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
2822 with ix86_finalize_stack_frame_flags.
2823 (ix86_expand_epilogue): Likewise.
2824 (ix86_expand_split_stack_prologue): Likewise.
2825 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
2827 2017-08-10 Martin Liska <mliska@suse.cz>
2830 * c-attribs.c (handle_target_attribute):
2831 Report warning for an empty string argument of target attribute.
2833 2017-08-09 Jakub Jelinek <jakub@redhat.com>
2836 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
2838 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
2839 or DECL_NONLOCAL labels.
2840 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
2841 or DECL_NONLOCAL labels here.
2843 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
2845 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
2846 to indicate when early gimple folding has been disabled.
2847 (rs6000_gimple_fold_builtin): Add debug content.
2848 (rs6000_invalid_builtin): Fix whitespace.
2849 (rs6000_expand_builtin): Fix whitespace.
2850 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
2852 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
2855 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
2856 SAVE_MULTIPLE if not all the registers that saves, should be saved.
2858 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
2860 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
2861 (qdf24xx): Likewise.
2862 * config/aarch64/aarch64.md: Include falkor.md.
2863 * config/aarch64/falkor.md: New.
2865 2017-08-09 Marek Polacek <polacek@redhat.com>
2868 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
2869 * diagnostic.c (emit_diagnostic): Add a comment.
2870 (emit_diagnostic_valist): New function.
2872 2017-08-09 Marek Polacek <polacek@redhat.com>
2875 * input.c (make_location): New overload.
2876 * input.h (make_location): Declare.
2878 2017-08-08 Alan Modra <amodra@gmail.com>
2879 H.J. Lu <hongjiu.lu@intel.com>
2882 * gcc.c (NO_PIE_SPEC): Delete.
2883 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
2885 (LINK_PIE_SPEC): ..to here.
2886 (LINK_COMMAND_SPEC): Support -no-pie.
2887 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
2888 chain of crtbegin*.o selection, update for PIE_SPEC changes and
2890 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
2891 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
2892 (ENDFILE_CRTEND_SPEC): Similarly.
2894 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
2897 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
2898 (mstack-protector-guard-offset=): Ditto.
2899 * config/i386/i386.c (ix86_option_override): Handle
2900 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
2902 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
2903 ix86_stack_protect_guard_offset variables.
2904 (TARGET_STACK_PROTECT_GUARD): Always define.
2905 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
2906 and -mstack-protector-guard-offset= options.
2908 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2910 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
2911 boundary case for the last candidate.
2913 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2915 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
2916 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
2918 2017-08-08 Tamar Christina <tamar.christina@arm.com>
2921 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
2922 * config/aarch64/aarch64-builtins.c
2923 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
2924 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
2925 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
2927 2017-08-08 Tamar Christina <tamar.christina@arm.com>
2928 Andrew Pinski <pinskia@gmail.com>
2931 * internal-fn.def (XORSIGN): New.
2932 * optabs.def (xorsign_optab): New.
2933 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
2934 (convert_expand_mult_copysign): New.
2935 (pass_optimize_widening_mul::execute): Call
2936 convert_expand_mult_copysign.
2938 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2940 PR tree-optimization/81354
2941 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
2942 Insert on edges rather than explicitly creating landing pads.
2943 (analyze_candidates_and_replace): Commit edge inserts.
2945 2017-08-08 Richard Biener <rguenther@suse.de>
2948 * tree-ssa-loop-niter.c: Include tree-dfa.h.
2949 (expand_simple_operations): Also look through ADDR_EXPRs with
2950 MEM_REF bases treating them as POINTER_PLUS_EXPR.
2952 2017-08-08 Richard Biener <rguenther@suse.de>
2954 PR tree-optimization/81723
2955 * tree-vect-slp.c (struct bst_traits): New hash traits.
2956 (bst_fail): New global.
2957 (vect_build_slp_tree_2): New worker, split out from ...
2958 (vect_build_slp_tree): ... this now wrapping it with using
2959 bst_fail set to cache SLP tree build fails. Properly handle
2961 (vect_analyze_slp_instance): Allocate and free bst_fail.
2963 2017-08-08 Martin Liska <mliska@suse.cz>
2966 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
2967 LABEL_DECLs that can be from a different function.
2969 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2971 PR tree-optimization/81744
2972 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
2973 loop's number of iterations.
2975 2017-08-08 Martin Liska <mliska@suse.cz>
2977 * asan.c: Include header files.
2978 * attribs.c (build_decl_attribute_variant): New function moved
2980 (build_type_attribute_qual_variant): Likewise.
2981 (cmp_attrib_identifiers): Likewise.
2982 (simple_cst_list_equal): Likewise.
2983 (omp_declare_simd_clauses_equal): Likewise.
2984 (attribute_value_equal): Likewise.
2985 (comp_type_attributes): Likewise.
2986 (build_type_attribute_variant): Likewise.
2987 (lookup_ident_attribute): Likewise.
2988 (remove_attribute): Likewise.
2989 (merge_attributes): Likewise.
2990 (merge_type_attributes): Likewise.
2991 (merge_decl_attributes): Likewise.
2992 (merge_dllimport_decl_attributes): Likewise.
2993 (handle_dll_attribute): Likewise.
2994 (attribute_list_equal): Likewise.
2995 (attribute_list_contained): Likewise.
2996 * attribs.h (lookup_attribute): New function moved from tree.[ch].
2997 (lookup_attribute_by_prefix): Likewise.
2998 * bb-reorder.c: Include header files.
2999 * builtins.c: Likewise.
3000 * calls.c: Likewise.
3001 * cfgexpand.c: Likewise.
3002 * cgraph.c: Likewise.
3003 * cgraphunit.c: Likewise.
3004 * convert.c: Likewise.
3005 * dwarf2out.c: Likewise.
3006 * final.c: Likewise.
3007 * fold-const.c: Likewise.
3008 * function.c: Likewise.
3009 * gimple-expr.c: Likewise.
3010 * gimple-fold.c: Likewise.
3011 * gimple-pretty-print.c: Likewise.
3012 * gimple.c: Likewise.
3013 * gimplify.c: Likewise.
3014 * hsa-common.c: Likewise.
3015 * hsa-gen.c: Likewise.
3016 * internal-fn.c: Likewise.
3017 * ipa-chkp.c: Likewise.
3018 * ipa-cp.c: Likewise.
3019 * ipa-devirt.c: Likewise.
3020 * ipa-fnsummary.c: Likewise.
3021 * ipa-inline.c: Likewise.
3022 * ipa-visibility.c: Likewise.
3024 * lto-cgraph.c: Likewise.
3025 * omp-expand.c: Likewise.
3026 * omp-general.c: Likewise.
3027 * omp-low.c: Likewise.
3028 * omp-offload.c: Likewise.
3029 * omp-simd-clone.c: Likewise.
3030 * opts-global.c: Likewise.
3031 * passes.c: Likewise.
3032 * predict.c: Likewise.
3033 * sancov.c: Likewise.
3034 * sanopt.c: Likewise.
3035 * symtab.c: Likewise.
3036 * toplev.c: Likewise.
3037 * trans-mem.c: Likewise.
3038 * tree-chkp.c: Likewise.
3039 * tree-eh.c: Likewise.
3040 * tree-into-ssa.c: Likewise.
3041 * tree-object-size.c: Likewise.
3042 * tree-parloops.c: Likewise.
3043 * tree-profile.c: Likewise.
3044 * tree-ssa-ccp.c: Likewise.
3045 * tree-ssa-live.c: Likewise.
3046 * tree-ssa-loop.c: Likewise.
3047 * tree-ssa-sccvn.c: Likewise.
3048 * tree-ssa-structalias.c: Likewise.
3049 * tree-ssa.c: Likewise.
3050 * tree-streamer-in.c: Likewise.
3051 * tree-vectorizer.c: Likewise.
3052 * tree-vrp.c: Likewise.
3054 * ubsan.c: Likewise.
3055 * varasm.c: Likewise.
3056 * varpool.c: Likewise.
3057 * tree.c: Remove functions moved to attribs.[ch].
3059 * config/aarch64/aarch64.c: Add attrs.h header file.
3060 * config/alpha/alpha.c: Likewise.
3061 * config/arc/arc.c: Likewise.
3062 * config/arm/arm.c: Likewise.
3063 * config/avr/avr.c: Likewise.
3064 * config/bfin/bfin.c: Likewise.
3065 * config/c6x/c6x.c: Likewise.
3066 * config/cr16/cr16.c: Likewise.
3067 * config/cris/cris.c: Likewise.
3068 * config/darwin.c: Likewise.
3069 * config/epiphany/epiphany.c: Likewise.
3070 * config/fr30/fr30.c: Likewise.
3071 * config/frv/frv.c: Likewise.
3072 * config/ft32/ft32.c: Likewise.
3073 * config/h8300/h8300.c: Likewise.
3074 * config/i386/winnt.c: Likewise.
3075 * config/ia64/ia64.c: Likewise.
3076 * config/iq2000/iq2000.c: Likewise.
3077 * config/lm32/lm32.c: Likewise.
3078 * config/m32c/m32c.c: Likewise.
3079 * config/m32r/m32r.c: Likewise.
3080 * config/m68k/m68k.c: Likewise.
3081 * config/mcore/mcore.c: Likewise.
3082 * config/microblaze/microblaze.c: Likewise.
3083 * config/mips/mips.c: Likewise.
3084 * config/mmix/mmix.c: Likewise.
3085 * config/mn10300/mn10300.c: Likewise.
3086 * config/moxie/moxie.c: Likewise.
3087 * config/msp430/msp430.c: Likewise.
3088 * config/nds32/nds32-isr.c: Likewise.
3089 * config/nds32/nds32.c: Likewise.
3090 * config/nios2/nios2.c: Likewise.
3091 * config/nvptx/nvptx.c: Likewise.
3092 * config/pa/pa.c: Likewise.
3093 * config/pdp11/pdp11.c: Likewise.
3094 * config/powerpcspe/powerpcspe.c: Likewise.
3095 * config/riscv/riscv.c: Likewise.
3096 * config/rl78/rl78.c: Likewise.
3097 * config/rx/rx.c: Likewise.
3098 * config/s390/s390.c: Likewise.
3099 * config/sh/sh.c: Likewise.
3100 * config/sol2.c: Likewise.
3101 * config/sparc/sparc.c: Likewise.
3102 * config/spu/spu.c: Likewise.
3103 * config/stormy16/stormy16.c: Likewise.
3104 * config/tilegx/tilegx.c: Likewise.
3105 * config/tilepro/tilepro.c: Likewise.
3106 * config/v850/v850.c: Likewise.
3107 * config/vax/vax.c: Likewise.
3108 * config/visium/visium.c: Likewise.
3109 * config/xtensa/xtensa.c: Likewise.
3111 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3114 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
3115 constraints since the -mupper-regs-* switches have been
3117 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
3118 into a vector from a double word element that was extracted from
3119 another vector, and eliminate extra XXPERMDI instructions.
3120 (vsx_concat_<mode>_2): Likewise.
3121 (vsx_concat_<mode>_3): Likewise.
3122 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
3123 concat to allow optimizing inserts from previous extracts.
3125 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
3127 * config/i386/i386.c (ix86_stack_protect_guard): Generate
3128 memory reference to a SSP offset in TLS address space.
3129 (ix86_print_operand) <case '@'>: Remove.
3130 (ix86_print_operand_punct_valid_p): Remove '@' code.
3131 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
3133 (stack_tls_protect_set_<mode>): Remove.
3134 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
3135 (stack_tls_protect_test_<mode>): Remove.
3136 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
3138 2017-08-07 Olivier Hainque <hainque@adacore.com>
3141 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
3143 2017-08-07 Douglas Rupp <rupp@adacore.com>
3145 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
3146 variable was referenced as multidir in command.
3148 2017-08-07 Jakub Jelinek <jakub@redhat.com>
3151 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
3154 2017-08-07 Martin Liska <mliska@suse.cz>
3156 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
3157 * config/rl78/rl78.c: Add include of attribs.h.
3158 * config/sh/sh.c: Likewise.
3159 * config/v850/v850.c: Likewise.
3161 2017-08-07 Tom de Vries <tom@codesourcery.com>
3164 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
3166 2017-08-07 Martin Liska <mliska@suse.cz>
3168 * config/mips/mips.c: Include attribs.h.
3170 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
3173 * doc/invoke.texi: Document change in behvaior for -Ofast for
3176 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
3178 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
3180 (aarch64_pop_regs): Likewise.
3181 (aarch64_gen_load_pair): Likewise.
3182 (aarch64_save_callee_saves): Likewise.
3183 (aarch64_restore_callee_saves): Likewise.
3185 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3187 * config/i386/i386.c: Revert the last change.
3189 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3192 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
3194 (ix86_finalize_stack_frame_flags): This. Also clear
3195 frame_pointer_needed if -fno-omit-frame-pointer is used without
3197 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
3198 with ix86_finalize_stack_frame_flags.
3199 (ix86_expand_epilogue): Likewise.
3200 (ix86_expand_split_stack_prologue): Likewise.
3202 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3205 * config/i386/i386.c (get_builtin_code_for_version): Set priority
3206 to P_AES for Westmere.
3208 2017-08-07 Jonathan Yong <10walls@gmail.com>
3210 * config/i386/mingw.opt (fset-stack-executable): Removed.
3211 * config/i386/cygming.opt (fset-stack-executable): Moved
3213 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
3215 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
3217 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
3219 2017-08-07 Marek Polacek <polacek@redhat.com>
3222 * fold-const.c (fold_indirect_ref_1): Check type_domain.
3224 2017-08-07 Martin Liska <mliska@suse.cz>
3226 * attribs.h (canonicalize_attr_name): New function.
3227 (cmp_attribs): Move from c-format.c and adjusted.
3228 (is_attribute_p): Moved from tree.h.
3229 * tree-inline.c: Add new includes.
3230 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
3231 (private_is_attribute_p): Remove.
3232 (private_lookup_attribute): Likewise.
3233 (private_lookup_attribute_by_prefix): Simplify.
3234 (remove_attribute): Use is_attribute_p.
3235 * tree.h: Remove removed declarations.
3237 2017-08-07 Jakub Jelinek <jakub@redhat.com>
3240 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
3241 instead of computing it in the function. Formatting fix.
3242 (expand_case): Don't rely on default_edge being the first edge,
3243 clear it if removing it, pass default_edge to
3244 emit_case_dispatch_table.
3245 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
3248 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
3250 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
3251 insn in the function, emit NOP after the insn.
3253 2017-08-06 Tom de Vries <tom@codesourcery.com>
3255 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
3258 2017-08-06 Tom de Vries <tom@codesourcery.com>
3260 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
3263 2017-08-04 Yury Gribov <tetra2005@gmail.com>
3265 PR tree-optimization/57371
3266 * match.pd: New pattern.
3268 2017-08-04 Marek Polacek <polacek@redhat.com>
3271 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
3272 perform the computation in offset_int.
3274 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3276 PR tree-optimization/81136
3277 * tree-vectorizer.h: Include tree-hash-traits.h.
3278 (vec_base_alignments): New typedef.
3279 (vec_info): Add a base_alignments field.
3280 (vect_record_base_alignments): Declare.
3281 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
3283 (DR_IS_CONDITIONAL_IN_STMT): New macro.
3284 (create_data_ref): Add an is_conditional_in_stmt argument.
3285 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
3286 the is_conditional_in_stmt field.
3287 (data_ref_loc): Add an is_conditional_in_stmt field.
3288 (get_references_in_stmt): Set the is_conditional_in_stmt field.
3289 (find_data_references_in_stmt): Update call to create_data_ref.
3290 (graphite_find_data_references_in_stmt): Likewise.
3291 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
3292 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
3293 (vect_record_base_alignment): New function.
3294 (vect_record_base_alignments): Likewise.
3295 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
3296 for nested statements even if we fail to compute a misalignment.
3297 Use pooled base alignments for unconditional references.
3298 (vect_find_same_alignment_drs): Compare base addresses instead
3300 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
3301 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
3303 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3305 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
3306 Add an explicit name for the enum. Use auto_vec for slp_instances
3308 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
3310 (_bb_vec_info): Add a constructor and destructor.
3311 (vinfo_for_stmt): Return NULL for uids of -1 as well.
3312 (destroy_loop_vec_info): Delete.
3313 (vect_destroy_datarefs): Likewise.
3314 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
3315 (vec_info::vec_info): New function.
3316 (vec_info::~vec_info): Likewise.
3317 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
3318 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
3319 destroy_loop_vec_info.
3320 * tree-vect-loop.c (new_loop_vec_info): Replace with...
3321 (_loop_vec_info::_loop_vec_info): ...this.
3322 (destroy_loop_vec_info): Replace with...
3323 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
3324 the stmt_vec_infos. Leave handling of vec_info information to its
3325 destructor. Remove explicit vector releases.
3326 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
3327 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
3328 * tree-vect-slp.c (new_bb_vec_info): Replace with...
3329 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
3330 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
3331 (destroy_bb_vec_info): Replace with...
3332 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
3333 information to its destructor.
3334 (vect_slp_analyze_bb_1): Use new and delete instead of
3335 new_bb_vec_info and destroy_bb_vec_info.
3336 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
3339 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3341 * tree-data-ref.h (subscript): Add access_fn field.
3342 (data_dependence_relation): Add could_be_independent_p.
3343 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
3344 (same_access_functions): Move to tree-data-ref.c.
3345 * tree-data-ref.c (ref_contains_union_access_p): New function.
3346 (access_fn_component_p): Likewise.
3347 (access_fn_components_comparable_p): Likewise.
3348 (dr_analyze_indices): Add a reference to access_fn_component_p.
3349 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
3351 (constant_access_functions): Likewise.
3352 (add_other_self_distances): Likewise.
3353 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
3354 (initialize_data_dependence_relation): Use XCNEW and remove
3355 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
3356 of access functions that have the same type. Allow the
3357 subsequence to end with different bases in some circumstances.
3358 Record the chosen access functions in SUB_ACCESS_FN.
3359 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
3360 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
3361 (subscript_dependence_tester_1): Likewise dra and drb.
3362 (build_classic_dist_vector): Update calls accordingly.
3363 (subscript_dependence_tester): Likewise.
3364 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
3365 DDR_COULD_BE_INDEPENDENT_P.
3366 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
3367 comp_alias_ddrs instead of may_alias_ddrs.
3368 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
3370 (vect_analyze_data_ref_dependence): Use it if
3371 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
3372 distance vectors if that fails.
3373 (dependence_distance_ge_vf): New function.
3374 (vect_prune_runtime_alias_test_list): Use it. Don't clear
3375 LOOP_VINFO_MAY_ALIAS_DDRS.
3377 2017-08-04 Richard Biener <rguenther@suse.de>
3380 * fold-const.c (fold_binary_loc): Properly restrict
3381 minus_var0 && minus_var1 case when associating undefined overflow
3384 2017-08-04 Tom de Vries <tom@codesourcery.com>
3386 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
3388 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3390 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3391 Don't start diagnostic messages with a capital letter.
3392 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3394 (rs6000_invalid_builtin): Likewise.
3395 (rs6000_trampoline_init): Likewise.
3397 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3400 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
3401 after setting changeable df flags.
3403 2017-08-03 Richard Biener <rguenther@suse.de>
3405 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
3406 up if the use is in USE - X.
3408 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
3410 * toplev.c (dumpfile.h): New include.
3411 (internal_error_reentered): New static function. Use it...
3412 (internal_error_function): ...here to handle reentered internal_error.
3414 2017-08-03 Richard Biener <rguenther@suse.de>
3417 * fold-const.c (split_tree): Add minus_var and minus_con
3418 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
3419 here but always use minus_*.
3420 (associate_trees): Assert we never associate with MINUS_EXPR
3421 and NULL first operand. Do not recurse for PLUS_EXPR operands
3422 when associating as MINUS_EXPR either.
3423 (fold_binary_loc): Track minus_var and minus_con.
3425 2017-08-03 Tom de Vries <tom@codesourcery.com>
3428 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
3429 ACCEL_COMPILER, apply finish_options on
3430 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
3432 2017-08-03 Tom de Vries <tom@codesourcery.com>
3435 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
3436 function_entry_patch_area_size > 0.
3438 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3441 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
3442 instead of 10??LU, perform unit multiplication in wide_int,
3443 don't change alloc_object_size_limit if the limit is larger
3446 PR tree-optimization/81655
3447 PR tree-optimization/81588
3448 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
3449 the case when ranges[i].low and high are 1 for unsigned type with
3453 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
3454 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
3456 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3458 * tree-vrp.h: Add include guard.
3460 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
3463 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
3464 (ud2): New insn pattern.
3465 * config/i386/i386.c (ix86_expand_epilogue):
3466 For naked functions, generate ud2 instead of trap insn.
3468 2017-08-02 Marek Polacek <polacek@redhat.com>
3471 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
3473 2017-08-02 Tom de Vries <tom@codesourcery.com>
3474 Cesar Philippidis <cesar@codesourcery.com>
3476 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
3477 Add missing edge probabilities.
3479 2017-08-02 Tamar Christina <tamar.christina@arm.com>
3481 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
3484 2017-08-02 Jakub Jelinek <jakub@redhat.com>
3487 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
3488 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
3489 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
3490 prologue_seq sequences - if any.
3492 2017-08-02 Richard Biener <rguenther@suse.de>
3494 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
3495 via vectors if supported, integer extracts via punning if supported
3496 or otherwise vector extracts.
3498 2017-08-02 Richard Biener <rguenther@suse.de>
3500 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
3502 (bitmap_insert_into_set): ... this.
3504 2017-08-02 Richard Biener <rguenther@suse.de>
3506 PR tree-optimization/81633
3508 2015-08-17 Alan Hayward <alan.hayward@arm.com>
3510 PR tree-optimization/71752
3511 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
3513 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
3515 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
3516 (machine_function::call_ms2sysv_pad_out): Remove field.
3517 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
3518 (ix86_compute_frame_layout): Likewise.
3520 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
3523 * config/i386/i386.c (ix86_set_func_type): Disallow naked
3524 attribute with interrupt attribute.
3526 2017-08-01 Andrew Pinski <apinski@cavium.com>
3528 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
3529 BIT_INSERT_EXPR's operand 1
3530 to see if the types precision matches.
3532 2017-08-01 Martin Liska <mliska@suse.cz>
3535 * builtins.c (expand_builtin_memcpy_args): Remove.
3536 (expand_builtin_memcpy): Call newly added function
3537 expand_builtin_memory_copy_args.
3538 (expand_builtin_memcpy_with_bounds): Likewise.
3539 (expand_builtin_mempcpy): Remove last argument.
3540 (expand_builtin_mempcpy_with_bounds): Likewise.
3541 (expand_builtin_memory_copy_args): New function created from
3542 expand_builtin_mempcpy_args with small modifications.
3543 (expand_builtin_mempcpy_args): Remove.
3544 (expand_builtin_stpcpy): Remove unused argument.
3545 (expand_builtin): Likewise.
3546 (expand_builtin_with_bounds): Likewise.
3548 2017-08-01 Martin Liska <mliska@suse.cz>
3551 Make mempcpy more optimal (PR middle-end/70140).
3553 2017-08-01 Jakub Jelinek <jakub@redhat.com>
3556 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
3557 __builtin_vec_cmpne verify both arguments are compatible vectors
3558 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
3559 verify arg1_type is a pointer or array type. For __builtin_vec_st,
3560 move computation of aligned to after checking the argument types.
3564 * config/rs6000/vsx.md (vextract_fp_from_shorth,
3565 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
3568 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
3569 Doug Rupp <rupp@adacore.com>
3570 Olivier Hainque <hainque@adacore.com>
3572 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
3573 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
3575 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
3576 for TARGET_OS_CPP_BUILTIN.
3577 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
3578 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
3580 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
3581 passing required abi options to the assembler for EABI configurations.
3582 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
3583 of .text.hot and .text.unlikely sections for kernel modules when
3584 using ARM style exceptions.
3585 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
3586 options. Add EXTRA_CC1_SPEC.
3587 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
3589 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
3591 (ARM_TARGET2_DWARF_FORMAT): Define.
3592 * config/arm/t-vxworks: Adjust multilib control to removal of the
3593 Diab command line options.
3595 2017-08-01 Martin Liska <mliska@suse.cz>
3597 PR gcov-profile/81561
3598 * gcov.c (unblock): Make unblocking safe as we need to preserve
3599 index correspondence of blocks and block_lists.
3601 2017-08-01 Richard Biener <rguenther@suse.de>
3603 PR tree-optimization/81181
3604 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3605 (compute_antic): ... end of iteration here.
3607 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
3609 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
3610 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
3611 (ftree-slp-vectorize): Likewise.
3612 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
3613 can no longer be set independent of flag_tree_loop_vectorize.
3614 * omp-general.c (emp_max_vf): Likewise.
3615 * opts.c (enable_fdo_optimizations): Remove references to
3616 flag_tree_vectorize, these are now implicit.
3617 (common_handle_option): Remove handling for OPT_ftree_vectorize,
3618 and leave it for the options machinery.
3620 2017-08-01 Martin Liska <mliska@suse.cz>
3623 * builtins.c (expand_builtin_memcpy_args): Remove.
3624 (expand_builtin_memcpy): Call newly added function
3625 expand_builtin_memory_copy_args.
3626 (expand_builtin_memcpy_with_bounds): Likewise.
3627 (expand_builtin_mempcpy): Remove last argument.
3628 (expand_builtin_mempcpy_with_bounds): Likewise.
3629 (expand_builtin_memory_copy_args): New function created from
3630 expand_builtin_mempcpy_args with small modifications.
3631 (expand_builtin_mempcpy_args): Remove.
3632 (expand_builtin_stpcpy): Remove unused argument.
3633 (expand_builtin): Likewise.
3634 (expand_builtin_with_bounds): Likewise.
3636 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
3639 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
3640 print "ds:" only for immediates in generic address space.
3642 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
3645 * config/i386/i386.c (ix86_funciton_naked): New prototype.
3646 (ix86_function_ok_for_sibcall): Return false for naked functions.
3648 2017-08-01 Richard Biener <rguenther@suse.de>
3650 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
3651 (compute_antic): Seed worklist with exit block predecessors.
3652 * cfganal.c (dfs_find_deadend): For a cycle return the source
3653 of the edge closing it.
3655 2017-08-01 Tamar Christina <tamar.christina@arm.com>
3657 * config/aarch64/aarch64.c
3658 (aarch64_can_const_movi_rtx_p): Move 0 check.
3660 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3662 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
3663 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
3665 * match.pd: Ditto in address comparison pattern.
3667 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3669 PR tree-optimization/81627
3670 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
3671 closed ssa form for store-store chain.
3673 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3675 PR tree-optimization/81620
3676 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
3677 for store-store chain.
3679 2017-08-01 Jakub Jelinek <jakub@redhat.com>
3681 PR tree-optimization/81588
3682 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
3683 ranges[i].in_p, invert comparison code ccode. For >/>=,
3684 swap rhs1 and rhs2 and comparison code unconditionally,
3685 for </<= don't do that. Don't swap rhs1/rhs2 again if
3686 ranges[i].in_p, instead invert comparison code ccode if
3687 opcode or oe->rank is BIT_IOR_EXPR.
3690 * optabs.def (vec_extract_optab, vec_init_optab): Change from
3691 a direct optab to conversion optab.
3692 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
3693 with GET_MODE_INNER as last argument instead of optab_handler.
3694 * expmed.c (extract_bit_field_1): Likewise. Use vector from
3695 vector extraction if possible and optab is available.
3696 * expr.c (store_constructor): Use convert_optab_handler instead
3697 of optab_handler. Use vector initialization from smaller
3698 vectors if possible and optab is available.
3699 * tree-vect-stmts.c (vectorizable_load): Likewise.
3700 * doc/md.texi (vec_extract, vec_init): Document that the optabs
3702 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
3703 of vec_init from half-sized vectors with the same element mode.
3704 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
3705 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
3706 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
3707 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
3708 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
3709 after mode in gen_vec_extract* calls.
3710 (vec_extract<mode>): Renamed to ...
3711 (vec_extract<mode><ssescalarmodelower>): ... this.
3712 (vec_extract<mode><ssehalfvecmodelower>): New expander.
3713 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
3714 element mode after mode in gen_vec_init* calls.
3715 (VEC_INIT_HALF_MODE): New mode iterator.
3716 (vec_init<mode>): Renamed to ...
3717 (vec_init<mode><ssescalarmodelower>): ... this.
3718 (vec_init<mode><ssehalfvecmodelower>): New expander.
3719 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
3720 (vec_extractv2sfsf): ... this.
3721 (vec_initv2sf): Renamed to ...
3722 (vec_initv2sfsf): ... this.
3723 (vec_extractv2si): Renamed to ...
3724 (vec_extractv2sisi): ... this.
3725 (vec_initv2si): Renamed to ...
3726 (vec_initv2sisi): ... this.
3727 (vec_extractv4hi): Renamed to ...
3728 (vec_extractv4hihi): ... this.
3729 (vec_initv4hi): Renamed to ...
3730 (vec_initv4hihi): ... this.
3731 (vec_extractv8qi): Renamed to ...
3732 (vec_extractv8qiqi): ... this.
3733 (vec_initv8qi): Renamed to ...
3734 (vec_initv8qiqi): ... this.
3735 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
3736 (vec_init<mode>): Renamed to ...
3737 (vec_init<mode><VEC_base_l>): ... this.
3738 (vec_extract<mode>): Renamed to ...
3739 (vec_extract<mode><VEC_base_l>): ... this.
3740 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
3741 (vec_initv2sfsf): ... this.
3742 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
3743 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
3744 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
3745 element mode after mode in gen_vec_init* calls.
3746 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
3747 (vec_init<mode><Vel>): ... this.
3748 (vec_extract<mode>): Renamed to ...
3749 (vec_extract<mode><Vel>): ... this.
3750 * config/aarch64/iterators.md (Vel): New mode attribute.
3751 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
3752 Add element mode after mode in gen_vec_extract* calls.
3753 * config/s390/vector.md (non_vec_l): New mode attribute.
3754 (vec_extract<mode>): Renamed to ...
3755 (vec_extract<mode><non_vec_l>): ... this.
3756 (vec_init<mode>): Renamed to ...
3757 (vec_init<mode><non_vec_l>): ... this.
3758 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
3759 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
3761 * config/arm/iterators.md (V_elem_l): New mode attribute.
3762 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
3763 (vec_extract<mode><V_elem_l>): ... this.
3764 (vec_extractv2di): Renamed to ...
3765 (vec_extractv2didi): ... this.
3766 (vec_init<mode>): Renamed to ...
3767 (vec_init<mode><V_elem_l>): ... this.
3768 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
3769 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
3770 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
3771 Add element mode after gen_vec_extract* calls.
3772 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
3773 (vec_init<mode><unitmode>): ... this.
3774 (vec_extract<mode>): Renamed to ...
3775 (vec_extract<mode><unitmode>): ... this.
3776 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
3777 (vec_init<mode><unitmode>): ... this.
3778 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
3779 (vec_initv2sfsf): ... this.
3780 (vec_extractv2sf): Renamed to ...
3781 (vec_extractv2sfsf): ... this.
3782 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
3783 Add element mode after gen_vec_extract* calls.
3784 * config/mips/mips.md (unitmode): New mode iterator.
3785 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
3786 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
3787 * config/spu/spu.md (inner_l): New mode attribute.
3788 (vec_init<mode>): Renamed to ...
3789 (vec_init<mode><inner_l>): ... this.
3790 (vec_extract<mode>): Renamed to ...
3791 (vec_extract<mode><inner_l>): ... this.
3792 * config/sparc/sparc.md (veltmode): New mode iterator.
3793 (vec_init<VMALL:mode>): Renamed to ...
3794 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
3795 * config/ia64/vect.md (vec_initv2si): Renamed to ...
3796 (vec_initv2sisi): ... this.
3797 (vec_initv2sf): Renamed to ...
3798 (vec_initv2sfsf): ... this.
3799 (vec_extractv2sf): Renamed to ...
3800 (vec_extractv2sfsf): ... this.
3801 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
3802 (vec_init<mode>): Renamed to ...
3803 (vec_init<mode><VEC_base_l>): ... this.
3804 (vec_extract<mode>): Renamed to ...
3805 (vec_extract<mode><VEC_base_l>): ... this.
3806 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
3807 (vec_initv2sfsf): ... this.
3808 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
3809 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
3810 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
3811 gen_vec_init* calls.
3813 2017-08-01 Richard Biener <rguenther@suse.de>
3815 PR tree-optimization/81297
3816 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
3817 TREE_OVERFLOW from INTEGER_CSTs.
3819 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
3821 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
3823 2017-07-31 Carl Love <cel@us.ibm.com>
3825 * config/rs6000/rs6000-c: Add support for built-in functions
3826 vector signed char vec_xl_be (signed long long, signed char *);
3827 vector unsigned char vec_xl_be (signed long long, unsigned char *);
3828 vector signed int vec_xl_be (signed long long, signed int *);
3829 vector unsigned int vec_xl_be (signed long long, unsigned int *);
3830 vector signed long long vec_xl_be (signed long long, signed long long *);
3831 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
3832 vector signed short vec_xl_be (signed long long, signed short *);
3833 vector unsigned short vec_xl_be (signed long long, unsigned short *);
3834 vector double vec_xl_be (signed long long, double *);
3835 vector float vec_xl_be (signed long long, float *);
3836 * config/rs6000/altivec.h (vec_xl_be): Add #define.
3837 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
3838 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
3840 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
3841 (altivec_expand_builtin): Add switch statement to call
3842 altivec_expand_xl_be for each builtin.
3843 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
3844 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
3845 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
3846 __builtin_vsx_le_be_v16qi.
3847 * doc/extend.texi: Update the built-in documentation file for the
3848 new built-in functions.
3850 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
3853 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
3855 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
3857 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3859 * config.gcc: Add z14.
3860 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
3861 CPU model numbers for z13s and z14.
3862 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
3864 * config/s390/s390-opts.h (enum processor_type): Rename
3865 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
3866 * config/s390/s390.c (processor_table): Add field for CPU name to
3867 be passed to Binutils.
3868 (s390_asm_output_machine_for_arch): Use the new field in
3869 processor_table for Binutils.
3870 (s390_expand_builtin): Replace arch12 with z14.
3871 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
3872 (s390_get_sched_attrmask): Likewise.
3873 (s390_get_unit_mask): Likewise.
3874 * config/s390/s390.opt: Add z14 to processor_type enum.
3876 2017-07-31 Martin Jambor <mjambor@suse.cz>
3879 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
3880 regardless of optimization level.
3882 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
3883 Martin Liska <mliska@suse.cz>
3885 * predict.def: Remove old comment and adjust probability.
3886 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
3889 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
3892 * config/i386/i386.c (ix86_function_naked): New function.
3893 (ix86_can_use_return_insn_p): Return false for naked functions.
3894 (ix86_expand_prologue): Skip prologue for naked functions.
3895 (ix86_expand_epilogue): Skip epilogue for naked functions
3896 and emit trap instruction.
3897 (ix86_warn_func_return): New function.
3898 (ix86_attribute_table): Add "naked" attribute specification.
3899 (TARGET_WARN_FUNC_RETURN): Define.
3900 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
3902 2017-07-31 Martin Liska <mliska@suse.cz>
3904 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
3905 (dump_gimple_bb_header): Always dump BB info.
3906 (pp_cfg_jump): Do not append info about BB when dumping a jump.
3908 2017-07-31 Martin Liska <mliska@suse.cz>
3911 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
3912 also with current_function_decl non-null equality.
3914 2017-07-31 Jakub Jelinek <jakub@redhat.com>
3917 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
3918 change type to the element type, instead add eltype variable and
3919 use it where we are interested in the element type.
3921 PR tree-optimization/81603
3922 * ipa-polymorphic-call.c
3923 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
3924 offset arithmetic in offset_int, bail out if the resulting bit offset
3925 doesn't fit into shwi.
3927 2017-07-31 Martin Liska <mliska@suse.cz>
3929 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
3930 (gimplify_save_expr): Fix comment.
3932 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
3935 * config/i386/i386.c (ix86_function_arg): Update arguments for
3937 (ix86_compute_frame_layout): Set the initial stack offset to
3938 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
3939 INCOMING_FRAME_SP_OFFSET.
3940 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
3941 stack before exception handler returns.
3942 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
3943 the 'ERROR_CODE' for exception handler.
3945 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
3947 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
3948 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
3949 (ASM_OUTPUT_REG_POP): Ditto.
3950 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
3951 instead of asm_fprintf to output pure string.
3953 2017-07-29 Jakub Jelinek <jakub@redhat.com>
3955 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
3956 to imported_module_or_decl hook.
3957 (debug_nothing_tree_tree_tree_bool): Remove.
3958 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
3959 * debug.c (do_nothing_debug_hooks): Use
3960 debug_nothing_tree_tree_tree_bool_bool instead of
3961 debug_nothing_tree_tree_tree_bool.
3962 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3963 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3964 * sdbout.c (sdb_debug_hooks): Likewise.
3965 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
3966 (gen_namespace_die): Add DW_AT_export_symbols attribute if
3968 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
3969 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
3970 attribute, don't add anything.
3972 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3974 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
3975 (fold_build2_stat_loc): Likewise.
3976 (fold_build3_stat_loc): Likewise.
3977 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
3978 (fold_build1_loc): Remove macro.
3979 (fold_build2_loc): Likewise.
3980 (fold_build3_loc): Likewise.
3982 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3984 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
3985 (gimple_build_debug_bind_source_stat): Likewise.
3986 * gimple.h (gimple_build_debug_bind): Remove macro.
3987 (gimple_build_debug_bind_source): Likewise.
3989 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3991 * bitmap.c (bitmap_alloc): Adjust.
3992 (bitmap_gc_alloc): Likewise.
3993 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
3995 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3997 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
3998 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
3999 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
4000 (bitmap_gc_alloc_stat): Likewise.
4001 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
4003 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4005 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
4006 * rtl.h (shallow_copy_rtx): Remove macro.
4008 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4010 * emit-rtl.c (gen_raw_REG): Adjust.
4011 * gengenrtl.c (gendef): Likewise.
4012 * rtl.c (rtx_alloc_stat): Remove _stat from name.
4013 * rtl.h (rtx_alloc): Remove macro.
4015 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4017 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
4018 (build_tree_list_stat): Likewise.
4019 * tree.h (build_tree_list): Remove macro.
4020 (build_tree_list_vec): Likewise.
4022 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4024 * tree.c (make_vector_stat): Remove _stat from name.
4025 (build_vector_stat): Likewise.
4026 * tree.h (make_vector_stat): Remove macro.
4027 (build_vector_stat): Likewise.
4029 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4031 * tree.h (build_var_debug_value): Remove prototype.
4033 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4035 * tree.c (tree_cons_stat): Remove _stat from name.
4036 * tree.h (tree_cons): Remove macro.
4038 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4040 * tree.c (build_vl_exp_stat): Remove _stat from name.
4041 * tree.h (build_vl_exp): Remove macro.
4043 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4045 * tree.c (build_decl_stat): Remove _stat from name.
4046 * tree.h (build_decl): Remove macro.
4048 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4050 * gimple.c (gimple_build_with_ops_stat): Adjust.
4051 (gimple_alloc_stat): Remove _stat from name.
4052 * gimple.h (gimple_alloc): Remove macro.
4054 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4056 * tree.c (make_tree_vec_stat): Remove _stat from name.
4057 (grow_tree_vec_stat): Likewise.
4058 * tree.h (make_tree_vec_stat): Adjust prototype.
4059 (grow_tree_vec_stat): Likewise.
4060 (make_tree_vec): Remove macro.
4061 (grow_tree_vec): Likewise.
4063 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4065 * fold-const.c (fold_build1_stat_loc): Adjust.
4066 (fold_build2_stat_loc): Likewise.
4067 (fold_build3_stat_loc): Likewise.
4068 * tree.c (build0_stat): Remove _stat from name.
4069 (build1_stat): Likewise.
4070 (build2_stat): Likewise.
4071 (build3_stat): Likewise.
4072 (build4_stat): Likewise.
4073 (build5_stat): Likewise.
4074 * tree.h (build1_loc): Remove macro, and rename _stat function
4076 (build2_loc): Likewise.
4077 (build3_loc): Likewise.
4078 (build4_loc): Likewise.
4079 (build5_loc): Likewise.
4081 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4083 * tree.c (make_int_cst_stat): Remove _stat from name.
4084 * tree.h (make_int_cst_stat): Adjust prototype.
4085 (make_int_cst): Remove macro.
4087 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4089 * tree.c (make_tre_binfo_stat): Remove _stat from name.
4090 * tree.h (make_tree_binfo_stat): Adjust prototype.
4091 (make_tree_binfo): Remove.
4093 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4095 * tree.c (copy_node_stat): Rename to copy_node.
4096 (build_distinct_type_copy): Adjust.
4097 * tree.h (copy_node_stat): Adjust prototype.
4098 (copy_node): Remove macro.
4100 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4102 * tree.c (make_node_stat): rename to make_node.
4103 (build_tree_list_stat): Adjust.
4104 (build0_stat): Likewise.
4105 (build2_stat): Likewise.
4106 (build3_stat): Likewise.
4107 (build4_stat): Likewise.
4108 (build5_stat): Likewise.
4109 (build_decl_stat): Likewise.
4110 * tree.h (make_node_stat): Adjust prototype.
4111 (make_node): remove macro.
4113 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
4115 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
4116 (PPC_FEATURE2_SCV): Likewise.
4117 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
4119 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4121 * config/aarch64/aarch64.c
4122 (aarch64_internal_mov_immediate): Add new special pattern.
4123 * config/aarch64/aarch64.md (*movdi_aarch64):
4124 Add reg/32bit const mov case.
4126 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4127 Richard Sandiford <richard.sandiford@linaro.org>
4129 * config/aarch64/aarch64.md (mov<mode>): Generalize.
4130 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4131 Add integer and movi cases.
4132 (movi-split-hf-df-sf split, fp16): New.
4133 (enabled): Added TARGET_FP_F16INST.
4134 * config/aarch64/iterators.md (GPF_HF): New.
4135 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
4137 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4139 * config/aarch64/aarch64.c
4140 (aarch64_simd_container_mode): Add prototype.
4141 (aarch64_expand_mov_immediate): Add HI support.
4142 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
4143 (aarch64_can_const_movi_rtx_p): New.
4144 (aarch64_preferred_reload_class):
4145 Remove restrictions of using FP registers for certain SIMD operations.
4146 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
4147 (aarch64_valid_floating_const): Add integer move validation.
4148 (aarch64_simd_imm_scalar_p): Remove.
4149 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
4150 (aarch64_legitimate_constant_p): Expand list of supported cases.
4151 * config/aarch64/aarch64-protos.h
4152 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
4153 (aarch64_reinterpret_float_as_int): New.
4154 (aarch64_simd_imm_scalar_p): Remove.
4155 * config/aarch64/constraints.md (Uvi): New.
4156 (Dd): Split into Ds and new Dd.
4157 * config/aarch64/aarch64.md (*movsi_aarch64):
4159 (*movdi_aarch64): Add SIMD mov case.
4161 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4163 * tree-predcom.c: (struct chain): Handle store-store chain in which
4164 stores for elimination only store loop invariant values.
4165 (execute_pred_commoning_chain): Ditto.
4166 (prepare_initializers_chain_store_elim): Ditto.
4167 (prepare_finalizers): Ditto.
4168 (is_inv_store_elimination_chain): New function.
4169 (initialize_root_vars_store_elim_1): New function.
4171 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4173 * tree-predcom.c: Revise general description of the pass.
4174 (enum chain_type): New enum type for store elimination.
4175 (struct chain): New field supporting store elimination.
4176 (struct component): Ditto.
4177 (dump_chain): Dump store-stores chain.
4178 (release_chain): Release resources.
4179 (split_data_refs_to_components): Compute and create component
4180 contains only stores for elimination.
4181 (get_chain_last_ref_at): New function.
4182 (make_invariant_chain): Initialization.
4183 (make_rooted_chain): Specify chain type in parameter and record it.
4184 (add_looparound_copies): Skip for store-stores chain.
4185 (determine_roots_comp): Compute type of chain and pass it to
4187 (initialize_root_vars_store_elim_2): New function.
4188 (finalize_eliminated_stores): New function.
4189 (remove_stmt): Handle store for elimination.
4190 (execute_pred_commoning_chain): Execute predictive commoning on
4192 (determine_unroll_factor): Skip unroll for store-stores chain.
4193 (prepare_initializers_chain_store_elim): New function.
4194 (prepare_initializers_chain): Hanlde store-store chain.
4195 (prepare_finalizers_chain, prepare_finalizers): New function.
4196 (tree_predictive_commoning_loop): Return integer value indicating
4197 if loop is unrolled or lcssa form is corrupted.
4198 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
4200 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4202 * tree-predcom.c (initialize_root): Delete.
4203 (execute_pred_commoning_chain): Initialize root vars and replace
4204 reference of non-combined chain directly, rather than call above
4207 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4209 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
4210 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
4212 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4214 * tree-predcom.c (struct chain): New field init_seq.
4215 (release_chain): Release init_seq.
4216 (prepare_initializers_chain): Record intialization stmts in above
4218 (insert_init_seqs): New function.
4219 (tree_predictive_commoning_loop): Call insert_init_seqs.
4221 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4223 * tree-predcom.c (determine_roots_comp): Skip trivial components.
4225 2017-07-28 Richard Biener <rguenther@suse.de>
4227 * match.pd: Remove superfluous :c.
4228 * genmatch.c (simplify::id): Add member.
4229 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
4231 (current_id): New global.
4232 (dt_node::parent): Move from ...
4233 (dt_operand::parent): ... here. Add for_id member.
4234 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
4235 (decision_tree::find_node): Relax order requirement when
4236 merging DT_TRUE nodes to ones inbetween the current simplify
4237 and the one we try to merge with. Add diagnostic whenever
4238 we need to enforce pattern order by not merging.
4239 (decision_tree::insert): Set current_id.
4240 (decision_tree::print_node): Dump parent node and for_id.
4241 (parser::last_id): Add member.
4242 (parser::push_simplify): Assign unique id.
4243 (parser::parser): Initialize last_id.
4245 2017-07-28 Martin Liska <mliska@suse.cz>
4248 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
4249 gimple_build_debug_bind.
4251 2017-07-28 Richard Biener <rguenther@suse.de>
4253 PR tree-optimization/81502
4254 * match.pd: Add pattern combining BIT_INSERT_EXPR with
4256 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
4258 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
4259 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
4260 for BIT_FIELD_REF args.
4261 * fold-const.c (make_bit_field_ref): Likewise.
4262 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
4264 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4267 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
4268 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
4269 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
4270 Or it into SANITIZER_UNDEFINED.
4271 * ubsan.c: Include gimple-fold.h and varasm.h.
4272 (ubsan_expand_ptr_ifn): New function.
4273 (instrument_pointer_overflow): New function.
4274 (maybe_instrument_pointer_overflow): New function.
4275 (instrument_object_size): Formatting fix.
4276 (pass_ubsan::execute): Call instrument_pointer_overflow
4277 and maybe_instrument_pointer_overflow.
4278 * internal-fn.c (expand_UBSAN_PTR): New function.
4279 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
4280 * sanitizer.def (__ubsan_handle_pointer_overflow,
4281 __ubsan_handle_pointer_overflow_abort): New builtins.
4282 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
4283 * internal-fn.def (UBSAN_PTR): New internal function.
4284 * opts.c (sanitizer_opts): Add pointer-overflow.
4285 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
4286 * fold-const.c (build_range_check): Compute pointer range check in
4287 integral type if pointer arithmetics would be needed. Formatting
4290 2017-07-28 Martin Liska <mliska@suse.cz>
4293 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
4294 parameters that are of a variable-length.
4296 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
4298 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
4300 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
4301 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
4302 (CRT_CALL_STATIC_FUNCTION): Likewise.
4303 (ASM_DEFAULT_SPEC): New define.
4304 (ASM_SPEC32): Likewise.
4305 (ASM_SPEC64): Likewise.
4306 (ASM_SPEC_COMMON): Likewise.
4307 (ASM_SPEC): Likewise.
4308 (INVALID_64BIT): Likewise.
4309 (LINK_OS_DEFAULT_SPEC): Likewise.
4310 (LINK_OS_SPEC32): Likewise.
4311 (LINK_OS_SPEC64): Likewise.
4312 (POWERPC_LINUX): Likewise.
4313 (PTRDIFF_TYPE): Likewise.
4314 (RESTORE_FP_PREFIX): Likewise.
4315 (RESTORE_FP_SUFFIX): Likewise.
4316 (SAVE_FP_PREFIX): Likewise.
4317 (SAVE_FP_SUFFIX): Likewise.
4318 (SIZE_TYPE): Likewise.
4319 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
4320 (TARGET_64BIT): Likewise.
4321 (TARGET_64BIT): Likewise.
4322 (TARGET_AIX): Likewise.
4323 (WCHAR_TYPE_SIZE): Likewise.
4324 (WCHAR_TYPE): Undefine.
4325 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
4326 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
4327 (CPP_OS_RTEMS_SPEC): Delete.
4328 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
4329 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
4331 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
4333 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4335 PR tree-optimization/81578
4336 * tree-parloops.c (build_new_reduction): Bail out if
4337 reduction_code isn't one of the standard OpenMP reductions.
4338 Move the details printing after that decision.
4340 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4342 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
4343 related to reload_in_progress.
4344 (splat_input_operand): Likewise.
4345 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
4347 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
4349 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
4350 (TARGET_INSTANTIATE_DECLS): Likewise.
4351 (legitimate_indexed_address_p): Delete reload_in_progress code.
4352 (rs6000_debug_legitimate_address_p): Likewise.
4353 (rs6000_eliminate_indexed_memrefs): Likewise.
4354 (rs6000_emit_le_vsx_store): Likewise.
4355 (rs6000_emit_move_si_sf_subreg): Likewise.
4356 (rs6000_emit_move): Likewise.
4357 (register_to_reg_type): Likewise.
4358 (rs6000_pre_atomic_barrier): Likewise.
4359 (rs6000_machopic_legitimize_pic_address): Likewise.
4360 (rs6000_allocate_stack_temp): Likewise.
4361 (rs6000_address_for_fpconvert): Likewise.
4362 (rs6000_address_for_altivec): Likewise.
4363 (rs6000_secondary_memory_needed_rtx): Delete function.
4364 (rs6000_check_sdmode): Likewise.
4365 (rs6000_alloc_sdmode_stack_slot): Likewise.
4366 (rs6000_instantiate_decls): Likewise.
4367 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
4368 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
4369 Delete reload_in_progress.
4370 (*vec_reload_and_plus_<mptrsize>): Likewise.
4371 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
4372 (vsx_div_v2di): Likewise.
4373 (vsx_udiv_v2di): Likewise.
4375 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4377 * config/rs6000/rs6000.opt (mlra): Replace with stub.
4378 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
4379 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
4380 (rs6000_debug_reg_global): Delete print of LRA status.
4381 (rs6000_option_override_internal): Delete dead LRA related code.
4382 (rs6000_lra_p): Delete function.
4383 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
4385 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
4387 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
4388 * config/riscv/rtems.h: New file.
4390 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4391 Sudakshina Das <sudi.das@arm.com>
4393 * config/aarch64/aarch64.md
4394 (define_split for and<mode>3nr_compare): Move
4395 non aarch64_logical_operand to a register.
4396 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
4397 register immediate operand to a register.
4398 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
4400 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4403 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
4405 2017-07-27 Richard Biener <rguenther@suse.de>
4407 PR tree-optimization/81573
4408 PR tree-optimization/81494
4409 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
4410 multi defuse cycle case.
4412 2017-07-27 Richard Biener <rguenther@suse.de>
4414 PR tree-optimization/81571
4415 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
4418 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
4420 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
4421 earlier and only if MASK_FPU is set. Adjust formatting.
4423 2017-07-27 Martin Liska <mliska@suse.cz>
4425 * opt-functions.awk: Add validation of value of Init.
4426 * optc-gen.awk: Pass new argument.
4428 2017-07-27 Martin Liska <mliska@suse.cz>
4430 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
4431 Fix wrong condition.
4433 2017-07-27 Martin Liska <mliska@suse.cz>
4435 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
4436 BBs and edges seen by autoFDO.
4438 2017-07-27 Richard Biener <rguenther@suse.de>
4440 PR tree-optimization/81502
4441 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
4442 with incompatible but same sized type.
4443 (execute_update_addresses_taken): Likewise.
4445 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
4447 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
4448 flag_tree_loop_vectorize rather than flag_tree_vectorize.
4450 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4453 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
4454 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
4455 Change s_operand to memory_operand.
4457 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
4459 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
4460 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
4461 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
4462 Emit instructions rather than returning an expression. Handle TFmode
4463 and KFmode by casting to TImode.
4464 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
4465 (rs6000_emit_le_vsx_store): Likewise.
4466 * config/rs6000/vsx.md (VSX_TI): New iterator.
4467 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
4468 (*vsx_le_undo_permute_<mode>): Likewise.
4469 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
4470 emit the split sequence.
4471 (*vsx_le_perm_store_<mode>): Likewise.
4473 2017-07-27 Jakub Jelinek <jakub@redhat.com>
4475 PR tree-optimization/81555
4476 PR tree-optimization/81556
4477 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
4478 if true, force CHANGED for the recursive invocation.
4479 (reassociate_bb): Remember original length of ops array, pass
4480 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
4482 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
4483 attributes for noipa attribute. For naked attribute use
4484 lookup_attribute first before lookup_attribute_spec.
4485 * final.c (rest_of_handle_final): Disable IPA RA for functions with
4487 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
4488 for functions with noipa attribute.
4489 (cgraph_externally_visible_p): Return true for functions with noipa
4491 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
4492 for functions with noipa attribute.
4493 * doc/extend.texi: Document noipa function attribute.
4494 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
4495 also for functions with noipa attribute.
4496 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
4498 2017-07-26 Andrew Pinski <apinski@cavium.com>
4500 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
4501 vec_unalign_load_cost and vec_unalign_store_cost.
4503 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4505 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
4506 -mvsx-small-integer option.
4507 (ISA_3_0_MASKS_IEEE): Likewise.
4508 (OTHER_VSX_VECTOR_MASKS): Likewise.
4509 (POWERPC_MASKS): Likewise.
4510 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
4511 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
4512 code, only testing for DImode being allowed in non-VSX floating
4514 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
4515 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
4517 (rs6000_option_override_internal): Delete -mvsx-small-integer.
4518 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
4519 TARGET_P8_VECTOR test.
4520 (rs6000_secondary_reload_simple_move): Likewise.
4521 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
4522 since TARGET_P9_VECTOR was already tested.
4523 (rs6000_opt_masks): Remove -mvsx-small-integer.
4524 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
4525 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4527 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
4528 test for TARGET_VEXTRACTUB was used, and that uses
4530 (p9 extract splitter): Likewise.
4531 (vsx_extract_<mode>_di_p9): Likewise.
4532 (vsx_extract_<mode>_store_p9): Likewise.
4533 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
4534 for TARGET_P9_VECTOR was used. Delete code that is now dead with
4535 the elimination of TARGET_VSX_SMALL_INTEGER.
4536 (vsx_extract_<mode>_p8): Likewise.
4537 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
4538 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
4539 (vsx_set_<mode>_p9): Likewise.
4540 (vsx_set_v4sf_p9): Likewise.
4541 (vsx_set_v4sf_p9_zero): Likewise.
4542 (vsx_insert_extract_v4sf_p9): Likewise.
4543 (vsx_insert_extract_v4sf_p9_2): Likewise.
4544 * config/rs6000/rs6000.md (sign extend splitter): Change
4545 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
4546 (floatsi<mode>2_lfiwax_mem): Likewise.
4547 (floatunssi<mode>2_lfiwzx_mem): Likewise.
4548 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
4549 since a test for TARGET_P9_VECTOR was used.
4550 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4551 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
4552 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4553 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
4554 TARGET_P8_VECTOR test.
4555 (fix_trunc<mode>si2_stfiwx): Likewise.
4556 (fix_trunc<mode>si2_internal): Likewise.
4557 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
4558 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4560 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
4561 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
4562 TARGET_P8_VECTOR test.
4563 (fixuns_trunc<mode>si2_stfiwx): Likewise.
4564 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
4565 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4567 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
4568 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
4569 since a test for TARGET_P9_VECTOR was used.
4570 (splitter for loading small constants): Likewise.
4572 2017-07-26 Andrew Pinski <apinski@cavium.com>
4574 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
4577 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
4580 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
4581 (fp_valid_at): Likewise.
4583 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
4585 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
4586 (qdf24xx_addrcost_table): Likewise.
4587 (cortexa57_tunings): Update to use generic_branch_cost.
4588 (cortexa72_tunings): Likewise.
4589 (cortexa73_tunings): Likewise.
4590 (qdf24xx_tunings): Likewise.
4592 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
4594 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
4595 (thunderx2t99_branch_cost): Likewise.
4596 (cortexa35_tunings): Update to use generic_branch_cost.
4597 (cortexa53_tunings): Likewise.
4598 (cortexa57_tunings): Likewise.
4599 (cortexa72_tunings): Likewise.
4600 (cortexa73_tunings): Likewise.
4601 (thunderx2t99_tunings): Likewise.
4603 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
4605 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
4606 (sparc_option_override): Honour MASK_FSMULD.
4607 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
4608 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
4609 * config/sparc/sparc.opt (mfsmuld): New option.
4610 * doc/invoke.texi (mfsmuld): Document option.
4612 2017-07-26 Marek Polacek <polacek@redhat.com>
4615 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
4617 2017-07-26 Richard Biener <rguenther@suse.de>
4619 * gimple-match-head.c (do_valueize): Return OP if valueize
4621 (get_def): New helper to get at the def stmt of a SSA name
4623 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
4624 do_valueize to get at the def stmt.
4625 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
4627 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
4630 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
4632 2017-07-26 Martin Liska <mliska@suse.cz>
4635 * function.c (expand_function_start): Make expansion of
4636 nonlocal_goto_save_area after parm_birth_insn.
4638 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
4640 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
4641 from all CPU target flags enable members.
4643 2017-07-26 Richard Biener <rguenther@suse.de>
4645 * genmatch.c (dt_simplify::gen): Make iterator vars const.
4646 (decision_tree::gen): Make 'type' const.
4647 (write_predicate): Likewise.
4649 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
4651 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
4652 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
4653 (rs6000_option_override_internal): Likewise.
4654 (rs6000_expand_vector_set): Likewise.
4655 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
4656 (TARGET_UPPER_REGS_SF): Likewise.
4657 (TARGET_UPPER_REGS_DI): Likewise.
4658 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
4659 (TARGET_DIRECT_MOVE_64BIT): Likewise.
4660 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
4661 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4662 (Splitters for DI constants in Altivec registers): Likewise.
4663 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
4664 (vsx_set_v4sf_p9): Likewise.
4665 (vsx_set_v4sf_p9_zero): Likewise.
4666 (vsx_insert_extract_v4sf_p9): Likewise.
4667 (vsx_insert_extract_v4sf_p9_2): Likewise.
4669 2017-07-25 Carl Love <cel@us.ibm.com>
4671 * doc/extend.texi: Update the built-in documentation file for the
4672 existing built-in functions
4673 vector signed char vec_cnttz (vector signed char);
4674 vector unsigned char vec_cnttz (vector unsigned char);
4675 vector signed short vec_cnttz (vector signed short);
4676 vector unsigned short vec_cnttz (vector unsigned short);
4677 vector signed int vec_cnttz (vector signed int);
4678 vector unsigned int vec_cnttz (vector unsigned int);
4679 vector signed long long vec_cnttz (vector signed long long);
4680 vector unsigned long long vec_cnttz (vector unsigned long long);
4682 2017-07-25 Andrew Pinski <apinski@cavium.com>
4684 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
4685 accesses where the use is for the first operand of a BIT_INSERT.
4687 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
4690 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
4691 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
4693 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
4695 * config/i386/gstabs.h: Delete.
4696 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
4698 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
4700 * config/i386/i386.c (ix86_decompose_address): Do not check for
4701 register RTX when looking at index_reg or base_reg.
4702 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
4704 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
4706 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
4707 to update EH info here.
4709 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4711 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
4713 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4715 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
4717 2017-07-25 Torsten Duwe <duwe@suse.de>
4719 * common.opt: Introduce -fpatchable-function-entry
4720 command line option, and its variables function_entry_patch_area_size
4721 and function_entry_patch_area_start.
4722 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
4723 including a two-value parser.
4724 * target.def (print_patchable_function_entry): New target hook.
4725 * targhooks.h (default_print_patchable_function_entry): New function.
4726 * targhooks.c (default_print_patchable_function_entry): Likewise.
4727 * toplev.c (process_options): Switch off IPA-RA if
4728 patchable function entries are being generated.
4729 * varasm.c (assemble_start_function): Look at the
4730 patchable-function-entry command line switch and current
4731 function attributes and maybe generate NOP instructions by
4732 calling the print_patchable_function_entry hook.
4733 * doc/extend.texi: Document patchable_function_entry attribute.
4734 * doc/invoke.texi: Document -fpatchable_function_entry
4735 command line option.
4736 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
4738 * doc/tm.texi: Re-generate.
4740 2017-07-25 Jakub Jelinek <jakub@redhat.com>
4743 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
4744 TARGET_AVX512DQ rather than TARGET_AVX512BW.
4746 2017-07-25 Tamar Christina <tamar.christina@arm.com>
4748 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
4750 2017-07-25 Richard Biener <rguenther@suse.de>
4752 PR tree-optimization/81455
4753 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
4754 not walk in cycles when looking for guards.
4756 2017-07-25 Richard Biener <rguenther@suse.de>
4758 PR tree-optimization/81529
4759 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
4760 when optimizing backedge uses.
4762 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
4764 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
4766 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
4767 to dl_section_ref. On AIX, append an expression to subtract
4768 the size of the section length to dl_section_ref.
4770 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
4772 * configure.ac: If any of the config.* scripts fail, exit 1.
4773 * configure: Regenerate.
4775 2017-07-25 Richard Biener <rguenther@suse.de>
4778 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
4779 of immediate uses, be more verbose on errors.
4781 2017-07-25 Richard Biener <rguenther@suse.de>
4783 PR tree-optimization/81510
4784 * tree-vect-loop.c (vect_is_simple_reduction): When the
4785 reduction stmt is not inside the loop bail out.
4787 2017-07-25 Richard Biener <rguenther@suse.de>
4789 PR tree-optimization/81303
4790 * tree-vect-loop-manip.c (vect_loop_versioning): Build
4791 profitability check against LOOP_VINFO_NITERSM1.
4793 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4795 * domwalk.c (cmp_bb_postorder): Simplify.
4796 (sort_bbs_postorder): New function. Use it...
4797 (dom_walker::walk): ...here to optimize common cases.
4799 2017-07-25 Martin Liska <mliska@suse.cz>
4802 * ipa-visibility.c (function_and_variable_visibility): Make the
4803 redirection just on target that supports aliasing.
4804 Fix GNU coding style.
4806 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
4809 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
4811 * config/arm/rtems.h: Likewise.
4812 * config/bfin/rtems.h: Likewise.
4813 * config/i386/rtemself.h: Likewise.
4814 * config/lm32/rtems.h: Likewise.
4815 * config/m32c/rtems.h: Likewise.
4816 * config/m68k/rtemself.h: Likewise.
4817 * config/microblaze/rtems.h: Likewise.
4818 * config/mips/rtems.h: Likewise.
4819 * config/moxie/rtems.h: Likewise.
4820 * config/nios2/rtems.h: Likewise.
4821 * config/powerpcspe/rtems.h: Likewise.
4822 * config/rs6000/rtems.h: Likewise.
4823 * config/rtems.h: Likewise.
4824 * config/sh/rtems.h: Likewise.
4825 * config/sh/rtemself.h: Likewise.
4826 * config/sparc/rtemself.h: Likewise.
4828 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
4831 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
4832 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
4833 * tree-ssa-structalias.c (alias_get_name): Same.
4835 2017-07-25 Bin Cheng <bin.cheng@arm.com>
4838 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
4839 instructions if no du chain is found.
4841 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
4843 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
4845 2017-07-25 Richard Biener <rguenther@suse.de>
4848 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
4851 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
4853 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
4855 (ISA_2_7_MASKS_SERVER): Likewise.
4856 (ISA_3_0_MASKS_IEEE): Likewise.
4857 (OTHER_P8_VECTOR_MASKS): Likewise.
4858 (OTHER_VSX_VECTOR_MASKS): Likewise.
4859 (POWERPC_MASKS): Likewise.
4860 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
4861 duplicate list of options.
4862 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
4863 explicit -mupper-regs options.
4864 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
4865 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
4866 alias for -mupper-regs-df.
4867 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4868 (rs6000_init_hard_regno_mode_ok): Likewise.
4869 (rs6000_option_override_internal): Likewise.
4870 (rs6000_opt_masks): Likewise.
4871 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
4872 options in terms of whether -mvsx or -mpower8-vector was used.
4873 (TARGET_UPPER_REGS_DI): Likewise.
4874 (TARGET_UPPER_REGS_SF): Likewise.
4875 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
4876 -mupper-regs-* options.
4878 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
4880 * passes.c (emergency_dump_function): Print some empty lines and a
4881 header before the RTL dump.
4883 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
4885 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
4887 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
4890 * config/aarch64/aarch64.c (aarch64_classify_symbol):
4891 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
4893 2017-07-24 Carl Love <cel@us.ibm.com>
4895 * config/rs6000/rs6000-c.c: Add support for built-in functions
4896 vector float vec_extract_fp32_from_shorth (vector unsigned short);
4897 vector float vec_extract_fp32_from_shortl (vector unsigned short);
4898 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
4899 vec_extract_fp_from_shortl): Add defines for the two builtins.
4900 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
4901 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
4903 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
4904 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
4905 * doc/extend.texi: Update the built-in documentation file for the
4906 new built-in function.
4908 2017-07-24 Jakub Jelinek <jakub@redhat.com>
4911 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
4913 * doc/generic.texi: Likewise.
4914 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
4915 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
4917 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
4919 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
4920 (aarch64_mls_elt_merge<mode>): Likewise.
4922 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
4924 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
4925 having __cxa_atexit.
4927 2017-07-23 Michael Collison <michael.collison@arm.com>
4929 * config/arm/arm.c (arm_option_override): Deprecate
4930 use of -mstructure-size-boundary.
4931 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
4932 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
4934 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4937 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
4938 Reduce cost estimate for direct moves.
4940 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
4943 * config/i386/i386.c (ix86_option_override_internal): Disable
4944 BMI, BMI2 and TBM instructions for -m16.
4946 2017-07-21 Carl Love <cel@us.ibm.com>
4948 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4949 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4950 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4951 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4952 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4953 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4954 VMULOSW): New enum "unspec" values.
4955 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4956 altivec_vmulosw): New patterns.
4957 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4958 VMULOSW): Add definitions.
4960 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
4962 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
4963 (qdf24xx): Likewise.
4964 * config/aarch64/aarch64-options-extensions.def (rdma); New.
4965 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
4966 (AARCH64_FL_V8_1): Renumber.
4967 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
4968 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
4969 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
4970 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
4971 rdma to feature modifiers list.
4973 2017-07-21 Yury Gribov <tetra2005@gmail.com>
4976 * ipa-visibility (function_and_variable_visibility): Convert
4977 recursive PLT call to direct call if appropriate.
4979 2017-07-21 Andrew Pinski <apinski@cavium.com>
4981 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
4982 operand 1 to see if the types precision matches.
4983 * fold-const.c (operand_equal_p): Likewise.
4985 2017-07-21 Richard Biener <rguenther@suse.de>
4987 PR tree-optimization/81303
4988 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
4989 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
4990 (vect_peeling_hash_get_lowest_cost): Adjust.
4991 (vect_enhance_data_refs_alignment): Likewise. Use
4992 vect_get_peeling_costs_all_drs to compute the penalty for no
4993 peeling to match up costs.
4995 2017-07-21 Richard Biener <rguenther@suse.de>
4997 PR tree-optimization/81500
4998 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
4999 we didn't identify a reduction path.
5001 2017-07-21 Tom de Vries <tom@codesourcery.com>
5002 Cesar Philippidis <cesar@codesourcery.com>
5004 PR gcov-profile/81442
5005 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
5008 2017-07-21 Tom de Vries <tom@codesourcery.com>
5011 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
5013 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
5014 nvptx_override_options_after_change.
5016 2017-07-21 Ulrich Drepper <drepper@redhat.com>
5018 * dwarf2out.c (output_file_names): Avoid double testing for
5021 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
5023 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
5025 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5027 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
5029 (try_crossjump_to_edge): Do not punt on partitioned functions.
5031 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5033 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
5034 Put all BBs reachable only via paths crossing cold region to cold
5036 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
5038 2016-07-21 Richard Biener <rguenther@suse.de>
5040 PR tree-optimization/81303
5041 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
5042 into account prologue and epilogue iterations when raising
5043 min_profitable_iters to sth at least covering one vector iteration.
5045 2017-07-21 Tamar Christina <tamar.christina@arm.com>
5047 * config/arm/arm.c (arm_test_cpu_arch_dat):
5050 2017-07-20 Nathan Sidwell <nathan@acm.org>
5052 Remove TYPE_METHODS.
5053 * tree.h (TYPE_METHODS): Delete.
5054 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
5055 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
5056 (dbxout_type_methods): Scan TYPE_FIELDS.
5057 (dbxout_type): Don't check TYPE_METHODS here.
5058 * function.c (use_register_for_decl): Always ignore register for
5059 class types when not optimizing.
5060 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
5061 * tree.c (free_lang_data_in_type): Stitch out member functions and
5062 templates from TYPE_FIELDS.
5063 (build_distinct_type_copy, verify_type_variant,
5064 verify_type): Member fns are on TYPE_FIELDS.
5065 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
5066 * tree-pretty-print.c (dump_generic_node): Likewise.
5068 2017-07-20 Jakub Jelinek <jakub@redhat.com>
5071 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
5072 V2TImode and V4TImode.
5073 (ix86_expand_vector_extract): Likewise.
5074 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
5075 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
5076 (ssescalarmode): Handle V4TImode and V2TImode.
5077 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
5078 (*vec_extractv2ti, *vec_extractv4ti): New insns.
5079 (VEXTRACTI128_MODE): New mode iterator.
5080 (splitter for *vec_extractv?ti first element): New.
5081 (VEC_INIT_MODE): New mode iterator.
5082 (vec_init<mode>): Consolidate 3 expanders into one using
5083 VEC_INIT_MODE mode iterator.
5085 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
5087 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
5088 non_spilled_static_chain_regno_p.
5090 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
5092 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
5094 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
5096 * bb-reorder.c (connect_traces): Allow copying of blocks within
5099 2017-07-20 Richard Biener <rguenther@suse.de>
5101 * gimple.h (gimple_phi_result): Add gphi * overload.
5102 (gimple_phi_result_ptr): Likewise.
5103 (gimple_phi_arg): Likewise. Adjust index assert to only
5104 allow actual argument accesses rather than all slots available
5106 (gimple_phi_arg_def): Add gphi * overload.
5107 * tree-phinodes.c (make_phi_node): Initialize only actual
5109 (resize_phi_node): Clear memory not covered by old node,
5110 do not initialize excess argument slots.
5111 (reserve_phi_args_for_new_edge): Initialize new argument slot
5114 2017-07-20 Bin Cheng <bin.cheng@arm.com>
5116 PR tree-optimization/81388
5118 2016-07-21 Bin Cheng <bin.cheng@arm.com>
5120 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
5121 by removing computation of may_be_zero.
5123 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5124 Tom de Vries <tom@codesourcery.com>
5127 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
5128 when gimple level profile disagrees with what RTL expander did.
5130 2017-07-20 Richard Biener <rguenther@suse.de>
5132 PR tree-optimization/61171
5133 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
5134 (vect_analyze_stmt): Add slp instance parameter.
5135 (vectorizable_reduction): Likewise.
5136 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
5137 (vect_is_simple_reduction): Deal with chains not detected
5138 as SLP reduction chain, specifically not properly associated
5139 chains containing a mix of plus/minus.
5140 (get_reduction_op): Remove.
5141 (get_initial_defs_for_reduction): Simplify, pass in whether
5142 this is a reduction chain, pass in the SLP node for the PHIs.
5143 (vect_create_epilog_for_reduction): Get the SLP instance as
5145 (vectorizable_reduction): Get the SLP instance as arg.
5146 During analysis remember the SLP node with the PHIs in the
5147 instance. Simplify getting at the vectorized reduction PHIs.
5148 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
5149 through SLP instance.
5150 (vect_slp_analyze_operations): Likewise.
5151 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
5152 (vect_transform_stmt): Likewise.
5154 2017-07-20 Tom de Vries <tom@codesourcery.com>
5156 PR tree-optimization/81489
5157 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
5158 read of phi arg location to before loop that modifies phi.
5160 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5162 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
5165 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5168 * except.c (execute): Fix ordering issue.
5170 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
5172 PR rtl-optimization/81423
5173 * combine.c (make_compound_operation_int): Don't try to optimize
5174 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
5176 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
5178 PR rtl-optimization/81423
5179 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
5180 with a constant that is -1 in the truncated to mode.
5182 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5184 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
5185 (determine_unlikely_bbs): ... here.
5186 * predict.h (propagate_unlikely_bbs_forward): Declare.
5187 * cfgexpand.c (pass_expand::execute): Use it.
5188 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
5190 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
5191 propagate_unlikely_bbs_forward.
5193 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5196 * except.c (maybe_add_nop_after_section_switch): New function.
5199 2017-07-19 Tom de Vries <tom@codesourcery.com>
5201 * gimple.h (gimple_phi_set_arg): Make assert more strict.
5203 2017-07-19 Tom de Vries <tom@codesourcery.com>
5205 * gimple.h (gimple_phi_arg): Make assert more strict.
5207 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
5209 * config.gcc (powerpc*-*-*): Add mmintrin.h.
5210 * config/rs6000/mmintrin.h: New file.
5211 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
5213 2017-07-19 Jakub Jelinek <jakub@redhat.com>
5215 PR tree-optimization/81346
5216 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
5218 2017-07-19 Tom de Vries <tom@codesourcery.com>
5220 * config/nvptx/nvptx.md (VECIM): Add V2DI.
5222 2017-07-19 Tom de Vries <tom@codesourcery.com>
5224 * config/nvptx/nvptx-modes.def: Add V2DImode.
5225 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
5226 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
5227 (nvptx_output_mov_insn): Handle lack of mov.b128.
5228 (nvptx_print_operand): Handle 'H' and 'L' codes.
5229 (nvptx_vector_mode_supported): Allow V2DImode.
5230 (nvptx_preferred_simd_mode): New function.
5231 (nvptx_data_alignment): New function.
5232 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
5233 nvptx_preferred_simd_mode.
5234 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
5236 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
5238 2017-07-19 Tom de Vries <tom@codesourcery.com>
5240 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
5241 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
5242 (nvptx_vector_mode_supported): New function. Allow V2SImode.
5243 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
5244 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
5245 (mov<VECIM>_insn): New define_insn.
5246 (define_expand "mov<VECIM>): New define_expand.
5248 2017-07-19 Tom de Vries <tom@codesourcery.com>
5250 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
5252 2017-07-19 Jakub Jelinek <jakub@redhat.com>
5254 PR tree-optimization/81346
5255 * fold-const.h (fold_div_compare, range_check_type): Declare.
5256 * fold-const.c (range_check_type): New function.
5257 (build_range_check): Use range_check_type.
5258 (fold_div_compare): No longer static, rewritten into
5259 a match.pd helper function.
5260 (fold_comparison): Don't call fold_div_compare here.
5261 * match.pd (X / C1 op C2): New optimization using fold_div_compare
5264 2017-07-19 Nathan Sidwell <nathan@acm.org>
5266 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
5267 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
5268 * tree.c (find_decls_types_r, verify_type): Use
5269 TYPE_{MIN,MAX}_VALUE_RAW.
5270 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
5271 (hash_tree): Likewise.
5272 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5274 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5277 2017-07-18 Tom de Vries <tom@codesourcery.com>
5280 * omp-expand.c (expand_omp_for_static_chunk): Handle
5281 equal-argument loop exit phi.
5283 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
5286 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
5287 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
5288 operand 2 predicate.
5289 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
5290 operand 2 predicate.
5291 (ror,rol -> rorx splitters): Use const_int_operand as
5292 operand 2 predicate.
5294 2017-06-18 Richard Biener <rguenther@suse.de>
5296 PR tree-optimization/81410
5297 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
5298 the gap in the ! slp_perm SLP case after each group.
5300 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5303 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
5306 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5309 * predict.c (set_even_probabilities): Cleanup; do not affect
5310 probabilities that are already known.
5311 (combine_predictions_for_bb): Call even when count is set.
5313 2017-07-18 Nathan Sidwell <nathan@acm.org>
5315 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
5318 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5321 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
5322 optimization for loop niter analysis.
5324 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
5327 * config/avr/avr.c (avr_optimize_casesi): Don't use
5328 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
5330 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5332 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
5333 body_cost_vec from _vect_peel_extended_info.
5334 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
5335 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
5338 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5340 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
5342 2017-07-18 Richard Biener <rguenther@suse.de>
5344 PR tree-optimization/80620
5345 PR tree-optimization/81403
5346 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
5347 info when re-using a VN table entry.
5349 2017-07-18 Richard Biener <rguenther@suse.de>
5351 PR tree-optimization/81418
5352 * tree-vect-loop.c (vectorizable_reduction): Properly compute
5353 vectype_in. Verify that with lane-reducing reduction operations
5354 we have a single def-use cycle.
5356 2017-07-17 Carl Love <cel@us.ibm.com>
5358 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
5360 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5361 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5362 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5363 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5364 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5365 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5366 VMULOSW): New enum "unspec" values.
5367 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
5368 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
5369 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5370 altivec_vmulosw): New patterns.
5371 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5372 VMULOSW): Add definitions.
5373 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
5375 * config/alpha/alpha.c: Include predict.h.
5377 2017-07-17 Yury Gribov <tetra2005@gmail.com>
5379 * tree-vrp.c (compare_assert_loc): Fix comparison function
5380 to return predictable results.
5382 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5384 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
5387 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
5388 * doc/invoke.texi (mexpand-adddi): Update text.
5390 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5392 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
5393 that also clobbers the CC register. The old expand code is moved
5395 (*arc_clzsi2): ... here.
5396 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
5397 the CC register. The old expand code is moved to ...
5398 (arc_ctzsi2): ... here.
5400 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5402 * config/arc/arc.opt (mindexed-loads): Use initial value
5403 TARGET_INDEXED_LOADS_DEFAULT.
5404 (mauto-modify-reg): Use initial value
5405 TARGET_AUTO_MODIFY_REG_DEFAULT.
5406 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5407 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5408 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5409 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5411 2017-07-17 Martin Liska <mliska@suse.cz>
5414 * opts.c (finish_options): Do not allow -fgnu-tm
5415 w/ -fsanitize={kernel-,}address. Say sorry.
5417 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5420 * tree-loop-distribution.c (classify_partition): Only assert on
5421 numer of iterations.
5422 (merge_dep_scc_partitions): Delete prameter. Update function call.
5423 (distribute_loop): Remove code handling loop with unknown niters.
5424 (pass_loop_distribution::execute): Skip loop with unknown niters.
5426 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5429 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
5430 function sort_partitions_by_post_order.
5432 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5434 PR tree-optimization/81374
5435 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
5436 the max index of basic blocks, rather than number of basic blocks.
5438 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5440 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
5442 (arc_legitimate_pic_operand_p): Likewise.
5443 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
5445 (arc_needs_pcl_p): Likewise.
5446 (arc_legitimate_pc_offset_p): Likewise.
5447 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
5448 function is also used in constrains.md.
5449 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
5450 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
5451 PLUS. Only return true/false in known cases, otherwise assert.
5452 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
5453 is already called in arc_legitimate_constant_p.
5454 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
5456 (LEGITIMATE_PIC_OPERAND_P): Use
5457 arc_raw_symbolic_reference_mentioned_p function.
5458 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
5463 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5464 Andrew Burgess <andrew.burgess@embecosm.com>
5466 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
5467 (arc_return_address_register): New function.
5468 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
5469 (arc_handle_fndecl_attribute): Add naked attribute.
5470 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5471 (TARGET_WARN_FUNC_RETURN): Likewise.
5472 (arc_allocate_stack_slots_for_args): New function.
5473 (arc_warn_func_return): Likewise.
5474 (machine_function): Change type fn_type.
5475 (arc_compute_function_type): Consider new naked function type,
5476 change function return type.
5477 (arc_must_save_register): Adapt to handle new
5478 arc_compute_function_type's return type.
5479 (arc_expand_prologue): Likewise.
5480 (arc_expand_epilogue): Likewise.
5481 (arc_return_address_regs): Delete.
5482 (arc_return_address_register): New function.
5483 (arc_epilogue_uses): Use above function.
5484 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
5485 (arc_function_type): Change encoding, add naked type.
5486 (ARC_INTERRUPT_P): Change to handle the new encoding.
5487 (ARC_FAST_INTERRUPT_P): Likewise.
5488 (ARC_NORMAL_P): Define.
5489 (ARC_NAKED_P): Likewise.
5490 (arc_compute_function_type): Delete prototype.
5491 * config/arc/arc.md (in_ret_delay_slot): Use
5492 arc_return_address_register function.
5493 (simple_return): Likewise.
5494 (p_return_i): Likewise.
5496 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5498 PR tree-optimization/81428
5499 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
5500 can't be built for those types.
5502 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
5504 Remove stuff dead since r239246.
5506 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
5507 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
5508 (avr_inform_devices): Remove dead stuff.
5510 2017-07-17 Tamar Christina <tamar.christina@arm.com>
5512 * config/arm/arm_neon.h: Fix softp typo.
5514 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5516 PR tree-optimization/81365
5517 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
5518 aggregate moves onto bb predecessor edges, make sure there are no
5519 loads that could alias the lhs in between the start of bb and the
5522 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
5525 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
5526 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
5527 [LSHIFTRT, outer_code = TRUNCATE]: Same.
5529 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5531 PR tree-optimization/81396
5532 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
5533 (init_symbolic_number): Initialize it to 1.
5534 (perform_symbolic_merge): Add n_ops from both operands into the new
5536 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
5537 without base_addr as useless if they need more than one operation.
5538 (bswap_replace): Handle !bswap case for NULL base_addr.
5540 2017-07-17 Tom de Vries <tom@codesourcery.com>
5543 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
5546 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
5548 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
5549 conditional builtin define __FIX_LEON3FT_B2BST.
5551 2017-07-17 Daniel Cederman <cederman@gaisler.com>
5553 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
5554 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
5557 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
5559 PR rtl-optimization/81424
5560 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
5561 to remove potential trapping from operands if -fnon-call-exceptions.
5563 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5565 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
5566 profile_proability for scalling.
5567 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
5569 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5571 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
5573 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5575 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
5576 fixpoint arithmetics.
5578 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5580 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
5581 fixpoint arithmetics.
5583 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5585 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
5586 fixpoint arithmetics.
5588 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5590 * profile-count.h (profile_probability::from_reg_br_prob_note,
5591 profile_probability::to_reg_br_prob_note): New functions.
5592 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
5593 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
5594 * predict.c (probability_reliable_p): Update.
5595 (edge_probability_reliable_p): Update.
5596 (br_prob_note_reliable_p): Update.
5597 (invert_br_probabilities): Update.
5598 (add_reg_br_prob_note): New function.
5599 (combine_predictions_for_insn): Update.
5600 * asan.c (asan_clear_shadow): Update.
5601 * cfgbuild.c (compute_outgoing_frequencies): Update.
5602 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
5603 (update_br_prob_note): Update.
5604 (rtl_verify_edges): Update.
5605 (purge_dead_edges): Update.
5606 (fixup_reorder_chain): Update.
5607 * emit-rtl.c (try_split): Update.
5608 * ifcvt.c (cond_exec_process_insns): Update.
5609 (cond_exec_process_if_block): Update.
5610 (dead_or_predicable): Update.
5611 * internal-fn.c (expand_addsub_overflow): Update.
5612 (expand_neg_overflow): Update.
5613 (expand_mul_overflow): Update.
5614 * loop-doloop.c (doloop_modify): Update.
5615 * loop-unroll.c (compare_and_jump_seq): Update.
5616 * optabs.c (emit_cmp_and_jump_insn_1): Update.
5617 * predict.h: Update.
5618 * reorg.c (mostly_true_jump): Update.
5620 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
5621 * config/alpha/alpha.c (emit_unlikely_jump): Update.
5622 * config/arc/arc.c: (emit_unlikely_jump): Update.
5623 * config/arm/arm.c: (emit_unlikely_jump): Update.
5624 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
5625 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
5626 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
5627 (ix86_print_operand): Update.
5628 (ix86_split_fp_branch): Update.
5629 (predict_jump): Update.
5630 * config/ia64/ia64.c (ia64_print_operand): Update.
5631 * config/mmix/mmix.c (mmix_print_operand): Update.
5632 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
5633 (rs6000_expand_split_stack_prologue): Update.
5634 * config/rs6000/rs6000.c: Update.
5635 * config/s390/s390.c (s390_expand_vec_strlen): Update.
5636 (s390_expand_vec_movstr): Update.
5637 (s390_expand_cs_tdsi): Update.
5638 (s390_expand_split_stack_prologue): Update.
5639 * config/sh/sh.c (sh_print_operand): Update.
5640 (expand_cbranchsi4): Update.
5641 (expand_cbranchdi4): Update.
5642 * config/sparc/sparc.c (output_v9branch): Update.
5643 * config/spu/spu.c (get_branch_target): Update.
5644 (ea_load_store_inline): Update.
5645 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
5646 * config/tilepro/tilepro.c: Update.
5648 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
5650 * gimplify.c (mostly_copy_tree_r): Revert latest change.
5651 (gimplify_save_expr): Likewise.
5653 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5655 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
5657 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5659 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
5661 * timevar.def (TV_IPA_FNSUMMARY): Define.
5663 2017-07-16 Daniel Cederman <cederman@gaisler.com>
5665 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
5666 to back store errata sensitive sequence from being generated.
5667 (sqrtdf2_fix): Likewise.
5669 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5671 * tree-ssa-threadupdate.c (compute_path_counts,
5672 update_joiner_offpath_counts): Use profile_probability.
5674 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
5677 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5679 * config/arm/arm-c.c (arm_cpu_builtins): Define
5680 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
5682 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
5684 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5685 array entries to represent __ieee128 versions of the
5686 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
5687 scalar_extract_sig, and scalar_insert_exp built-in functions.
5688 (altivec_resolve_overloaded_builtin): Add special case handling
5689 for the __builtin_scalar_insert_exp function, as represented by
5690 the P9V_BUILTIN_VEC_VSIEDP constant.
5691 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
5692 exponent support for __ieee128 argument.
5693 (VSESQP): Add scalar extract signature support for __ieee128
5695 (VSTDCNQP): Add scalar test negative support for __ieee128
5697 (VSIEQP): Add scalar insert exponent support for __int128 argument
5698 with __ieee128 result.
5699 (VSIEQPF): Add scalar insert exponent support for __ieee128
5700 argument with __ieee128 result.
5701 (VSTDCQP): Add scalar test data class support for __ieee128
5703 (VSTDCNQP): Add overload support for scalar test negative with
5705 (VSTDCQP): Add overload support for scalar test data class
5707 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
5709 (UNSPEC_VSX_SIEXPQP): New constant.
5710 (xsxexpqp): New insn for VSX scalar extract exponent quad
5712 (xsxsigqp): New insn for VSX scalar extract significand quad
5714 (xsiexpqpf): New insn for VSX scalar insert exponent quad
5715 precision with floating point argument.
5716 (xststdcqp): New expand for VSX scalar test data class quad
5718 (xststdcnegqp): New expand for VSX scalar test negative quad
5720 (xststdcqp): New insn to match expansions for VSX scalar test data
5721 class quad precision and VSX scalar test negative quad precision.
5722 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
5723 special case operand checking to enforce that second operand of
5724 VSX scalar test data class with quad precision argument is a 7-bit
5726 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
5727 prototypes and descriptions of __ieee128 versions of
5728 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
5729 scalar_test_data_class, and scalar_test_neg built-in functions.
5731 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5733 PR tree-optimization/81162
5734 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
5735 replace a negate with an add.
5737 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
5739 * doc/invoke.texi (arm/-mcpu): Document +crypto.
5741 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5743 * config/arm/arm-c.c (arm_cpu_builtins): Define
5744 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
5746 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5748 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
5749 (armv8-r): Set ARM Cortex-R52 as default CPU.
5750 * config/arm/arm-tables.opt: Regenerate.
5751 * config/arm/arm-tune.md: Regenerate.
5752 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
5754 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
5755 extension for -mcpu=cortex-r52.
5757 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5759 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
5760 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
5761 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
5762 (fp-armv8): Define it as FP_ARMv8 only.
5763 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
5764 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
5766 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
5767 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
5768 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
5769 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
5770 than TARGET_FPU_ARMV8.
5771 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
5772 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
5773 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
5775 * config/arm/neon.md (neon_vrint): Likewise.
5776 (neon_vcvt): Likewise.
5777 (neon_<fmaxmin_op><mode>): Likewise.
5778 (<fmaxmin><mode>3): Likewise.
5779 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
5780 * config/arm/predicates.md (arm_cond_move_operator): Check against
5781 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
5783 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
5785 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
5788 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5790 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
5791 loop in comment with memset.
5793 2017-07-14 Martin Liska <mliska@suse.cz>
5795 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
5796 * dwarf2out.c (is_java): Remove the function.
5797 (output_pubname): Remove usage of the function.
5798 (lower_bound_default): Remove usage of DW_LANG_Java.
5799 (gen_compile_unit_die): Likewise.
5800 * gcc.c: Remove compiler defaults for .java and .zip files.
5801 * gimple-expr.c (remove_suffix): Change as there's no longer
5802 extension than 4-letter one.
5803 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
5804 (gimplify_save_expr): Likewise.
5805 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
5806 as it's possible even for other languages than Java.
5807 * langhooks.h (struct lang_hooks): Remove Java from a comment.
5808 * lto-opts.c (lto_write_options): Remove reference to Java.
5809 * opts.c (strip_off_ending): Update file extension handling.
5810 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
5811 * tree-eh.c (lower_resx): Likewise.
5812 * tree.c (free_lang_data_in_type): Remove dead code.
5813 (find_decls_types_r): Likewise.
5814 (build_common_builtin_nodes): Remove Java from a comment.
5815 (verify_type): Remove dead code.
5816 * varasm.c (assemble_external): Remove Java from a comment.
5818 2017-07-14 Martin Liska <mliska@suse.cz>
5820 * opts.c (finish_options): Add quotes.
5821 (common_handle_option): Likewise.
5823 2017-07-14 Martin Liska <mliska@suse.cz>
5825 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
5826 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
5828 * dwarf2out.c (lower_bound_default): Do not handle
5830 (gen_compile_unit_die): Likewise.
5831 * gcc.c: Remove default extension binding for GNU Pascal.
5832 * stmt.c: Remove Pascal language from a comment.
5833 * xcoffout.c: Likewise.
5835 2017-07-13 David Malcolm <dmalcolm@redhat.com>
5838 * diagnostic-show-locus.c (fixit_cmp): New function.
5839 (layout::layout): Sort m_fixit_hints.
5840 (column_range::column_range): Assert that the values are valid.
5841 (struct char_span): New struct.
5842 (correction::overwrite): New method.
5843 (struct source_line): New struct.
5844 (line_corrections::add_hint): Add assertions. Reimplement memcpy
5845 calls in terms of classes source_line and char_span, and
5846 correction::overwrite.
5847 (selftest::test_overlapped_fixit_printing_2): New function.
5848 (selftest::diagnostic_show_locus_c_tests): Call it.
5850 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
5852 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
5853 early if there is no lhs.
5855 2017-07-13 Martin Liska <mliska@suse.cz>
5857 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
5858 (gen_reference_type_die): Likewise.
5859 * stor-layout.c: Remove Pascal-related comment.
5861 2017-07-13 Martin Liska <mliska@suse.cz>
5863 * opts.c (finish_options): Add quotes to error messages.
5864 (parse_sanitizer_options): Likewise.
5866 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
5868 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
5870 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
5872 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
5874 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
5876 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
5878 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
5880 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5883 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
5884 provides the hardware capability bits, define the macro
5885 __BUILTIN_CPU_SUPPORTS__.
5886 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
5887 if GLIBC does not provide the hardware capability bits. Add a
5888 gcc_unreachable call if the built-in cpu function is neither
5889 __builtin_cpu_is nor __builtin_cpu_supports.
5890 (rs6000_get_function_versions_dispatcher): Change the warning
5891 that an old GLIBC is used which does not export the capability
5892 bits to be an error.
5893 * doc/extend.texi (target_clones attribute): Document the
5894 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
5895 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
5896 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
5897 the macros defined by GCC if the newer GLIBC is available.
5899 2017-07-12 Jeff Law <law@redhat.com>
5901 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
5902 remaining includes slightly.
5903 * config/riscv/riscv-builtins.c: Include profile-count.h.
5905 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
5908 * config/avr/avr.c (avr_set_current_function): In diagnostic
5909 messages: Quote keywords and (parts of) identifiers.
5910 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
5913 2017-07-12 Carl Love <cel@us.ibm.com>
5915 * config/rs6000/rs6000-c.c: Add support for built-in functions
5916 vector bool char vec_revb (vector bool char);
5917 vector bool short vec_revb (vector short char);
5918 vector bool int vec_revb (vector bool int);
5919 vector bool long long vec_revb (vector bool long long);
5920 * doc/extend.texi: Update the built-in documentation file for the
5921 new built-in functions.
5923 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5925 * config/s390/s390.md: Remove movcc splitter.
5927 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5929 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
5930 load/store on condition.
5932 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
5935 * config/avr/avr.c (avr_encode_section_info)
5936 [progmem && !TREE_READONLY]: Error if progmem object needs
5939 2017-07-11 Michael Collison <michael.collison@arm.com>
5941 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
5944 2017-07-11 Carl Love <cel@us.ibm.com>
5946 * config/rs6000/rs6000-c.c: Add support for builtins
5947 vector unsigned int vec_parity_lsbb (vector signed int);
5948 vector unsigned int vec_parity_lsbb (vector unsigned int);
5949 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
5950 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
5951 vector unsigned long long vec_parity_lsbb (vector signed long long);
5952 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
5953 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
5954 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
5955 * doc/extend.texi: Update the built-in documentation file for the
5956 new built-in functions.
5958 2017-07-11 David Malcolm <dmalcolm@redhat.com>
5960 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
5961 (layout::m_primary_loc): New field.
5962 (layout::layout): Initialize new field. Move location filtering
5963 logic from here to...
5964 (layout::maybe_add_location_range): ...this new method. Add
5965 support for filtering to just the lines already specified by other
5967 (layout::will_show_line_p): New method.
5968 (gcc_rich_location::add_location_if_nearby): New method.
5969 (selftest::test_add_location_if_nearby): New test function.
5970 (selftest::diagnostic_show_locus_c_tests): Call it.
5971 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
5974 2017-07-11 Tom de Vries <tom@codesourcery.com>
5976 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
5977 (bb_first_real_insn): New function.
5978 (nvptx_single): Add extra initialization of broadcasted condition
5981 2017-07-11 Nathan Sidwell <nathan@acm.org>
5983 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
5985 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
5987 * doc/extend.texi (AVR Function Attributes): Remove weblink to
5988 Binutils doc as TEXI will mess them up.
5989 * doc/invoke.texi (AVR Options): Same here.
5991 2017-07-11 Daniel Cederman <cederman@gaisler.com>
5993 * config/sparc/sparc.opt (mfix-ut700): New option.
5994 (mfix-gr712rc): Likewise.
5995 (sparc_fix_b2bst): New variable.
5996 * doc/invoke.texi (SPARC options): Document them.
5997 (ARM options): Fix warnings.
5998 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
5999 instructions to prevent sequences that can trigger the store-store
6000 errata for certain LEON3FT processors.
6001 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
6002 (sparc_option_override): Set sparc_fix_b2bst appropriately.
6003 * config/sparc/sparc.md (fix_b2bst): New attribute.
6004 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
6006 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
6009 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
6011 (*rsqrtsf2_sse): Ditto.
6013 (div<mode>3): Macroize insn from divdf3 and divsf3
6014 using MODEF mode iterator.
6016 2017-07-10 Martin Sebor <msebor@redhat.com>
6018 PR tree-optimization/80397
6019 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
6020 instead of testing for equality to INTEGER_TYPE.
6022 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
6024 * config.gcc: Remove uclibc from arc target spec.
6026 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
6028 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
6030 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6033 * lto-wrapper.c (remove_option): New function.
6034 (merge_and_complain): Merge PIC/PIE options more realistically.
6036 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
6038 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
6042 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
6043 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
6044 * config.in: Regenerate.
6045 * configure: Regenerate.
6046 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
6047 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
6048 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
6049 (TARGET_GASISR_PROLOGUES): ...target mask.
6050 * common/config/avr/avr-common.c
6051 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
6052 Set -mgas-isr-prologues.
6053 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
6054 INSERT_PASS_BEFORE for it.
6055 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
6056 * config/avr/avr.c (avr_option_override)
6057 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
6058 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
6059 (avr_attribute_table) <no_gccisr>: Add new function attribute.
6060 (avr_set_current_function) <is_no_gccisr>: Init machine field.
6061 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
6063 (make_avr_pass_pre_proep): New function.
6064 (emit_push_sfr) <treg>: Add argument to function and use it
6066 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
6067 and set machine->gasisr.yes.
6068 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
6069 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
6070 __gcc_isr.n_pushed to .L__stack_usage.
6071 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
6072 (avr_asm_final_postscan_insn): ...this new static function.
6073 * config/avr/avr.h (machine_function)
6074 <is_no_gccisr, use_L__stack_usage>: New fields.
6075 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
6076 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
6077 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
6078 (gasisr, *gasisr): New expander and insn.
6079 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
6080 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
6081 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
6083 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
6085 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
6088 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
6090 Move jump-tables out of .text again.
6093 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
6094 (ASM_OUTPUT_ADDR_VEC): New function.
6095 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
6096 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
6097 INSN_ADDRESSes as asm comment.
6098 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
6099 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
6100 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
6101 * config/avr/avr.md (*tablejump): Adjust comment.
6102 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
6103 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
6105 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
6106 (avr_output_addr_vec): New proto.
6107 (avr_log_t) <insn_addresses>: New field.
6109 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
6112 * config/i386/i386.c (ix86_function_arg_advance): Set
6113 outgoing_args_on_stack to true if there are outgoing arguments
6115 (ix86_function_arg): Likewise.
6116 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
6117 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
6118 * config/i386/i386.h (machine_function): Add
6119 outgoing_args_on_stack.
6121 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
6123 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
6125 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
6127 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
6129 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
6130 (REAL_H): Remove $(MACHMODE_H).
6131 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
6133 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
6134 $(MACHMODE_H) and double-int.h.
6135 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
6137 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
6140 2017-07-07 Andrew Pinski <apinski@cavium.com>
6142 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
6143 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
6145 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6147 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
6148 Add warning if GCC was not configured to link against a GLIBC that
6149 exports the hardware capability bits.
6150 (make_resolver_func): Make resolver function private and not a
6151 COMDAT function. Create the name with clone_function_name instead
6152 of make_unique_name.
6155 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
6156 correct operand in doing the split.
6158 2017-07-07 Carl Love <cel@us.ibm.com>
6160 * config/rs6000/rs6000-c: Add support for built-in function
6161 vector unsigned short vec_pack_to_short_fp32 (vector float,
6163 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
6164 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
6165 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
6166 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
6167 (convert_4f32_8i16): Add define_expand.
6168 * doc/extend.texi: Update the built-in documentation file for the
6169 new built-in function.
6171 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6173 * config/sparc/m8.md: New file.
6174 * config/sparc/sparc.md: Include m8.md.
6176 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6178 * config/sparc/sparc.opt: New option -mvis4b.
6179 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
6180 (sparc_option_override): Handle VIS4B.
6181 (enum sparc_builtins): Define
6182 SPARC_BUILTIN_DICTUNPACK{8,16,32},
6183 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
6184 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
6185 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
6186 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
6187 (check_constant_argument): New function.
6188 (sparc_vis_init_builtins): Define builtins
6189 __builtin_vis_dictunpack{8,16,32},
6190 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
6191 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
6192 __builtin_vis_fpcmpde{8,16,32}shl and
6193 __builtin_vis_fpcmpur{8,16,32}shl.
6194 (sparc_expand_builtin): Check that the constant operands to
6195 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
6196 constant and in range.
6197 * config/sparc/sparc-c.c (sparc_target_macros): Handle
6199 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
6200 (SPARC_IMM5_P): Likewise.
6201 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
6202 (enabled): Handle vis4b.
6203 (UNSPEC_DICTUNPACK): New unspec.
6204 (UNSPEC_FPCMPSHL): Likewise.
6205 (UNSPEC_FPUCMPSHL): Likewise.
6206 (UNSPEC_FPCMPDESHL): Likewise.
6207 (UNSPEC_FPCMPURSHL): Likewise.
6208 (cpu_feature): New CPU feature `vis4b'.
6209 (dictunpack{8,16,32}): New insns.
6210 (FPCSMODE): New mode iterator.
6211 (fpcscond): New code iterator.
6212 (fpcsucond): Likewise.
6213 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
6214 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
6215 (fpcmpde{8,16,32}{si,di}shl): Likewise.
6216 (fpcmpur{8,16,32}{si,di}shl): Likewise.
6217 * config/sparc/constraints.md: Define constraints `q' for unsigned
6218 2-bit integer constants and `t' for unsigned 5-bit integer
6220 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
6222 (imm5_operand_dictunpack16): Likewise.
6223 (imm5_operand_dictunpack32): Likewise.
6224 (imm2_operand): Likewise.
6225 * doc/invoke.texi (SPARC Options): Document -mvis4b.
6226 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
6227 ditunpack* and fpcmp*shl builtins.
6229 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6231 * config.gcc: Handle m8 in --with-{cpu,tune} options.
6232 * config.in: Add HAVE_AS_SPARC6 define.
6233 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
6235 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
6237 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
6238 (CPP_CPU_SPEC): Handle m8.
6239 (ASM_CPU_SPEC): Likewise.
6240 * config/sparc/sparc-opts.h (enum processor_type): Add
6242 * config/sparc/sparc.c (m8_costs): New struct.
6243 (sparc_option_override): Handle TARGET_CPU_m8.
6244 (sparc32_initialize_trampoline): Likewise.
6245 (sparc64_initialize_trampoline): Likewise.
6246 (sparc_issue_rate): Likewise.
6247 (sparc_register_move_cost): Likewise.
6248 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
6249 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
6250 (ASM_CPU64_DEFAULT_SPEC): Likewise.
6251 (CPP_CPU_SPEC): Handle M8.
6252 (ASM_CPU_SPEC): Likewise.
6253 (AS_M8_FLAG): Define.
6254 * config/sparc/sparc.md: Add m8 to the cpu attribute.
6255 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
6256 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
6258 * configure: Regenerate.
6259 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
6262 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6264 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
6266 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
6267 ("*movdi_insn_sp32"): Do not set v3pipe.
6268 ("*movsi_insn"): Likewise.
6269 ("*movdi_insn_sp64"): Likewise.
6270 ("*movsf_insn"): Likewise.
6271 ("*movdf_insn_sp32"): Likewise.
6272 ("*movdf_insn_sp64"): Likewise.
6273 ("*zero_extendsidi2_insn_sp64"): Likewise.
6274 ("*sign_extendsidi2_insn"): Likewise.
6275 ("*mov<VM32:mode>_insn"): Likewise.
6276 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6277 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6278 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6279 ("<vlop:code><VL:mode>3"): Likewise.
6280 ("*not_<vlop:code><VL:mode>3"): Likewise.
6281 ("*nand<VL:mode>_vis"): Likewise.
6282 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
6283 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
6284 ("one_cmpl<VL:mode>2"): Likewise.
6285 ("faligndata<VM64:mode>_vis"): Likewise.
6286 ("alignaddrsi_vis"): Likewise.
6287 ("alignaddrdi_vis"): Likweise.
6288 ("alignaddrlsi_vis"): Likewise.
6289 ("alignaddrldi_vis"): Likewise.
6290 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6291 ("bmaskdi_vis"): Likewise.
6292 ("bmasksi_vis"): Likewise.
6293 ("bshuffle<VM64:mode>_vis"): Likewise.
6294 ("cmask8<P:mode>_vis"): Likewise.
6295 ("cmask16<P:mode>_vis"): Likewise.
6296 ("cmask32<P:mode>_vis"): Likewise.
6297 ("pdistn<P:mode>_vis"): Likewise.
6298 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6300 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6302 * config/sparc/sparc.md ("subtype"): New insn attribute.
6303 ("*wrgsr_sp64"): Set insn subtype.
6304 ("*rdgsr_sp64"): Likewise.
6305 ("alignaddrsi_vis"): Likewise.
6306 ("alignaddrdi_vis"): Likewise.
6307 ("alignaddrlsi_vis"): Likewise.
6308 ("alignaddrldi_vis"): Likewise.
6309 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6310 ("fexpand_vis"): Likewise.
6311 ("fpmerge_vis"): Likewise.
6312 ("faligndata<VM64:mode>_vis"): Likewise.
6313 ("bshuffle<VM64:mode>_vis"): Likewise.
6314 ("cmask8<P:mode>_vis"): Likewise.
6315 ("cmask16<P:mode>_vis"): Likewise.
6316 ("cmask32<P:mode>_vis"): Likewise.
6317 ("fchksm16_vis"): Likewise.
6318 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
6319 ("fmean16_vis"): Likewise.
6320 ("fp<plusminus_insn>64_vis"): Likewise.
6321 ("<plusminus_insn>v8qi3"): Likewise.
6322 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6323 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
6324 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
6325 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
6326 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
6327 ("*movqi_insn"): Likewise.
6328 ("*movhi_insn"): Likewise.
6329 ("*movsi_insn"): Likewise.
6330 ("movsi_pic_gotdata_op"): Likewise.
6331 ("*movdi_insn_sp32"): Likewise.
6332 ("*movdi_insn_sp64"): Likewise.
6333 ("movdi_pic_gotdata_op"): Likewise.
6334 ("*movsf_insn"): Likewise.
6335 ("*movdf_insn_sp32"): Likewise.
6336 ("*movdf_insn_sp64"): Likewise.
6337 ("*zero_extendhisi2_insn"): Likewise.
6338 ("*zero_extendqihi2_insn"): Likewise.
6339 ("*zero_extendqisi2_insn"): Likewise.
6340 ("*zero_extendqidi2_insn"): Likewise.
6341 ("*zero_extendhidi2_insn"): Likewise.
6342 ("*zero_extendsidi2_insn_sp64"): Likewise.
6343 ("ldfsr"): Likewise.
6344 ("prefetch_64"): Likewise.
6345 ("prefetch_32"): Likewise.
6346 ("tie_ld32"): Likewise.
6347 ("tie_ld64"): Likewise.
6348 ("*tldo_ldub_sp32"): Likewise.
6349 ("*tldo_ldub1_sp32"): Likewise.
6350 ("*tldo_ldub2_sp32"): Likewise.
6351 ("*tldo_ldub_sp64"): Likewise.
6352 ("*tldo_ldub1_sp64"): Likewise.
6353 ("*tldo_ldub2_sp64"): Likewise.
6354 ("*tldo_ldub3_sp64"): Likewise.
6355 ("*tldo_lduh_sp32"): Likewise.
6356 ("*tldo_lduh1_sp32"): Likewise.
6357 ("*tldo_lduh_sp64"): Likewise.
6358 ("*tldo_lduh1_sp64"): Likewise.
6359 ("*tldo_lduh2_sp64"): Likewise.
6360 ("*tldo_lduw_sp32"): Likewise.
6361 ("*tldo_lduw_sp64"): Likewise.
6362 ("*tldo_lduw1_sp64"): Likewise.
6363 ("*tldo_ldx_sp64"): Likewise.
6364 ("*mov<VM32:mode>_insn"): Likewise.
6365 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6366 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6368 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6370 * config/sparc/sparc.md ("type"): New insn type viscmp.
6371 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
6373 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
6374 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
6375 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6376 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
6378 ("n7_vis_logical_11cycle"): Likewise.
6379 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
6380 * config/sparc/niagara2.md ("niag3_vis": Likewise.
6381 * config/sparc/niagara.md ("niag_vis"): Likewise.
6382 * config/sparc/ultra3.md ("us3_fga"): Likewise.
6383 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
6385 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6387 * config/sparc/sparc.md: New instruction type `bmask'.
6388 (bmaskdi_vis): Use the `bmask' type.
6389 (bmasksi_vis): Likewise.
6390 * config/sparc/ultra3.md (us3_array): Likewise.
6391 * config/sparc/niagara7.md (n7_array): Likewise.
6392 * config/sparc/niagara4.md (n4_array): Likewise.
6393 * config/sparc/niagara2.md (niag2_vis): Likewise.
6394 (niag3_vis): Likewise.
6395 * config/sparc/niagara.md (niag_vis): Likewise.
6397 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6399 * ipa-comdats.c: Remove optimize check from gate.
6400 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
6401 for functions not optimized.
6402 (ipa_fn_summary_read): Skip optimize check.
6403 (ipa_fn_summary_write): Likewise.
6404 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
6406 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
6408 (can_inline_edge_p): Check flag_pcc_struct_return for match.
6409 (check_callers): Give up on caller which is not optimized.
6410 (inline_small_functions): Likewise.
6411 (ipa_inline): Do not give up when not optimizing.
6412 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
6413 away unoptimizes cdtors.
6414 (whole_program_function_and_variable_visibility): Do
6415 ipa_discover_readonly_nonaddressable_vars in LTO mode.
6416 * ipa.c (process_references): Do not check optimize.
6417 (symbol_table::remove_unreachable_nodes): Update optimize check.
6418 (set_writeonly_bit): Update optimize check.
6419 (pass_ipa_cdtor_merge::gate): Do not check optimize.
6420 (pass_ipa_single_use::gate): Remove.
6422 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
6424 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
6425 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
6426 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
6427 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
6428 permute_load, permute_store, adjust_extract, adjust_splat,
6429 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
6430 replace_swap_with_copy, dump_swap_insn_table,
6431 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
6432 recombine_lvx_pattern, recombine_stvx_pattern,
6433 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
6434 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
6435 to file rs6000-p8swap.c.
6436 * config/rs6000/rs6000-p8swap.c: New file.
6437 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
6438 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
6439 and rs6000*-*-* targets.
6441 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6443 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
6445 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6447 * lto-wrapper.c (merge_and_complain): Do not merge
6448 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
6449 fsigned_zeros, ftrapping_math, fwrapv.
6450 (append_compiler_options): Do not track these options.
6451 (append_linker_options): Likewie
6453 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6455 * cgraphunit.c (cgraph_node::finalize_function): When
6456 !flag_toplevel_reorde set no_reorder flag.
6457 (varpool_node::finalize_decl): Likewise.
6458 (symbol_table::compile): Drop no toplevel reorder path.
6460 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6462 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
6463 edges; zero probability is not better than uninitialized.
6465 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6467 * asan.h (asan_sanitize_allocas_p): Declare.
6468 * asan.c (asan_sanitize_allocas_p): New function.
6469 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
6470 (handle_builtin_alloca): Likewise.
6471 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
6472 if !asan_sanitize_allocas_p.
6473 * params.def (asan-instrument-allocas): Add new option.
6474 * params.h (ASAN_PROTECT_ALLOCAS): Define.
6475 * opts.c (common_handle_option): Disable allocas sanitization for
6478 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6480 * asan.c: Include gimple-fold.h.
6481 (get_last_alloca_addr): New function.
6482 (handle_builtin_stackrestore): Likewise.
6483 (handle_builtin_alloca): Likewise.
6484 (asan_emit_allocas_unpoison): Likewise.
6485 (get_mem_refs_of_builtin_call): Add new parameter, remove const
6486 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
6487 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
6488 (instrument_builtin_call): Pass gimple iterator to
6489 get_mem_refs_of_builtin_call.
6490 (last_alloca_addr): New global.
6491 * asan.h (asan_emit_allocas_unpoison): Declare.
6492 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
6493 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
6494 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
6495 if function calls alloca.
6496 * gimple-fold.c (replace_call_with_value): Remove static keyword.
6497 * gimple-fold.h (replace_call_with_value): Declare.
6498 * internal-fn.c: Include asan.h.
6499 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
6500 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
6502 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6504 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
6505 (C_SELFTEST_FLAGS): New.
6506 (CPP_SELFTEST_FLAGS): New.
6507 (SELFTEST_DEPS): New, from deps of s-selftest.
6508 (C_SELFTEST_DEPS): New, from deps of s-selftest.
6509 (CPP_SELFTEST_DEPS): New.
6510 (selftest): Add dependency on s-selftest-c++.
6511 (s-selftest): Rename to...
6512 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
6513 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
6514 than SELFTEST_FLAGS.
6515 (selftest-gdb): Rename to...
6516 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
6518 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
6519 (selftest-valgrind): Rename to...
6520 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
6522 (selftest-valgrind): Reintroduce as an alias for
6523 selftest-c-valgrind.
6524 (s-selftest-c++): New.
6525 (selftest-c++-gdb): New.
6526 (selftest-c++-valgrind): New.
6528 2017-07-06 Olivier Hainque <hainque@adacore.com>
6530 * gcc.c (process_command): When deciding if undefined variables
6531 should be ignored when processing specs, accept "gcc -v" as well.
6533 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6535 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
6536 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
6538 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6540 * config/arm/arm-cpus.in (armv8-r): Add new entry.
6541 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
6542 * config/arm/arm-tables.opt: Regenerate.
6543 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
6545 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
6547 2017-07-06 Carl Love <cel@us.ibm.com>
6549 * ChangeLog: Clean up from mid air collision
6551 2017-07-06 Carl Love <cel@us.ibm.com>
6553 * config/rs6000/rs6000-c.c: Add support for built-in functions
6554 vector signed int vec_subc (vector signed int, vector signed int);
6555 vector signed __int128 vec_subc (vector signed __int128,
6556 vector signed __int128);
6557 vector unsigned __int128 vec_subc (vector unsigned __int128,
6558 vector unsigned __int128);
6559 vector signed int vec_sube (vector signed int, vector signed int,
6561 vector unsigned int vec_sube (vector unsigned int,
6562 vector unsigned int,
6563 vector unsigned int);
6564 vector signed __int128 vec_sube (vector signed __int128,
6565 vector signed __int128,
6566 vector signed__int128);
6567 vector unsigned __int128 vec_sube (vector unsigned __int128,
6568 vector unsigned __int128,
6569 vector unsigned __int128);
6570 vector signed int vec_subec (vector signed int, vector signed int,
6572 vector unsigned int vec_subec (vector unsigned int,
6573 vector unsigned int,
6574 vector unsigned int);
6575 vector signed __int128 vec_subec (vector signed __int128,
6576 vector signed __int128,
6577 vector signed__int128);
6578 vector unsigned __int128 vec_subec (vector unsigned __int128,
6579 vector unsigned __int128,
6580 vector unsigned __int128);
6581 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
6582 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
6583 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
6584 BU_ALTIVEC_OVERLOAD_X definitions.
6585 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
6586 * doc/extend.texi: Update the built-in documentation file for the new
6589 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6592 * diagnostic-show-locus.c (layout::layout): Use start and finish
6593 spelling location for the start and finish of each range.
6594 * genmatch.c (linemap_client_expand_location_to_spelling_point):
6595 Add unused aspect param.
6596 * input.c (expand_location_1): Add "aspect" param, and use it
6597 to access the correct part of the location.
6598 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
6600 (expand_location_to_spelling_point): Likewise.
6601 (linemap_client_expand_location_to_spelling_point): Add "aspect"
6602 param, and pass it to expand_location_1.
6604 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
6606 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
6607 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
6608 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
6609 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
6610 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
6611 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
6612 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
6613 _mm_maskz_getmant_ss): New intrinsics.
6614 (__builtin_ia32_getexpss128_mask): Changed to ...
6615 __builtin_ia32_getexpss128_round ... this.
6616 (__builtin_ia32_getexpsd128_mask): Changed to ...
6617 __builtin_ia32_getexpsd128_round ... this.
6618 * config/i386/i386-builtin-types.def
6619 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
6620 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
6621 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
6622 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
6623 __builtin_ia32_getmantss_mask_round): New builtins.
6624 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
6625 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
6626 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
6627 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
6628 * config/i386/sse.md
6629 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
6630 avx512f_sgetexp<mode><mask_scalar_name>
6631 <round_saeonly_scalar_name> ... this.
6632 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
6633 %0, %1, %2<round_saeonly_op3>}): Changed to ...
6634 vgetexp<ssescalarmodesuffix>
6635 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6636 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
6637 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
6638 avx512f_vgetmant<mode><mask_scalar_name>
6639 <round_saeonly_scalar_name> ... this.
6640 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
6641 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
6642 vgetmant<ssescalarmodesuffix>
6643 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
6644 %0<mask_scalar_operand4>, %1, %2
6645 <round_saeonly_scalar_mask_op4>, %3} ... this.
6646 * config/i386/subst.md (mask_scalar_operand4,
6647 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
6648 round_saeonly_scalar_nimm_predicate): New subst attributes.
6650 2017-07-06 Julia Koval <julia.koval@intel.com>
6652 * config/i386/i386.c (ix86_erase_embedded_rounding):
6653 Remove code for old rounding pattern.
6655 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
6657 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
6659 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
6661 * doc/sourcebuild.texi (Test Directives, Variants of
6662 dg-require-support): Add documentation for dg-require-stack-check.
6664 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
6666 * config/i386/subst.md (mask_scalar, round_scalar,
6667 round_saeonly_scalar): New meta-templates.
6668 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
6669 round_scalar_mask_operand3, round_scalar_mask_op3,
6670 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
6671 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
6672 round_saeonly_scalar_constraint,
6673 round_saeonly_scalar_prefix): New subst attribute.
6674 * config/i386/sse.md
6675 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
6676 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
6677 <round_scalar_name> ... this.
6678 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
6679 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
6680 <round_scalar_name> ... this.
6681 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
6682 <sse>_vm<code><mode>3<mask_scalar_name>
6683 <round_saeonly_scalar_name> ... this.
6684 (v<plusminus_mnemonic><ssescalarmodesuffix>
6685 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
6686 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
6687 v<plusminus_mnemonic><ssescalarmodesuffix>
6688 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6689 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
6690 (v<multdiv_mnemonic><ssescalarmodesuffix>
6691 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
6692 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
6693 v<multdiv_mnemonic><ssescalarmodesuffix>
6694 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6695 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
6696 (v<maxmin_float><ssescalarmodesuffix>
6697 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
6698 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
6699 v<maxmin_float><ssescalarmodesuffix>
6700 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6701 %0<mask_scalar_operand3>, %1, %<iptr>2
6702 <round_saeonly_scalar_mask_op3>} ... this.
6704 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
6706 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
6707 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
6709 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
6710 Alan Hayward <alan.hayward@arm.com>
6711 David Sherwood <david.sherwood@arm.com>
6713 * combine.c (simplify_if_then_else): Remove "enum" before
6715 * compare-elim.c (can_eliminate_compare): Likewise.
6716 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6718 (aarch64_lookup_simd_builtin_type): Likewise.
6719 (aarch64_simd_builtin_type): Likewise.
6720 (aarch64_init_simd_builtin_types): Likewise.
6721 (aarch64_simd_expand_args): Likewise.
6722 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
6724 (aarch64_reverse_mask): Likewise.
6725 (aarch64_simd_emit_reg_reg_move): Likewise.
6726 (aarch64_gen_adjusted_ldpstp): Likewise.
6727 (aarch64_ccmp_mode_to_code): Likewise.
6728 (aarch64_operands_ok_for_ldpstp): Likewise.
6729 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6730 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
6732 (aarch64_min_divisions_for_recip_mul): Likewise.
6733 (aarch64_reassociation_width): Likewise.
6734 (aarch64_get_condition_code_1): Likewise.
6735 (aarch64_simd_emit_reg_reg_move): Likewise.
6736 (aarch64_simd_attr_length_rglist): Likewise.
6737 (aarch64_reverse_mask): Likewise.
6738 (aarch64_operands_ok_for_ldpstp): Likewise.
6739 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6740 (aarch64_gen_adjusted_ldpstp): Likewise.
6741 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
6743 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
6744 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6745 (arm_lookup_simd_builtin_type): Likewise.
6746 (arm_simd_builtin_type): Likewise.
6747 (arm_init_simd_builtin_types): Likewise.
6748 (arm_expand_builtin_args): Likewise.
6749 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
6750 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
6751 (ft32_setup_incoming_varargs): Likewise.
6752 (ft32_function_arg): Likewise.
6753 (ft32_function_arg_advance): Likewise.
6754 (ft32_pass_by_reference): Likewise.
6755 (ft32_arg_partial_bytes): Likewise.
6756 (ft32_valid_pointer_mode): Likewise.
6757 (ft32_addr_space_pointer_mode): Likewise.
6758 (ft32_addr_space_legitimate_address_p): Likewise.
6759 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
6761 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
6762 (ix86_emit_outlined_ms2sysv_restore): Likewise.
6763 (iamcu_alignment): Likewise.
6764 (canonicalize_vector_int_perm): Likewise.
6765 (ix86_noce_conversion_profitable_p): Likewise.
6766 (ix86_mpx_bound_mode): Likewise.
6767 (ix86_operands_ok_for_move_multiple): Likewise.
6768 * config/microblaze/microblaze-protos.h
6769 (microblaze_expand_conditional_branch_reg): Likewise.
6770 * config/microblaze/microblaze.c
6771 (microblaze_expand_conditional_branch_reg): Likewise.
6772 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6774 (rs6000_reassociation_width): Likewise.
6775 (rs6000_invalid_binary_op): Likewise.
6776 (fusion_p9_p): Likewise.
6777 (emit_fusion_p9_load): Likewise.
6778 (emit_fusion_p9_store): Likewise.
6779 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
6781 (riscv_hard_regno_mode_ok_p): Likewise.
6782 (riscv_address_insns): Likewise.
6783 (riscv_split_symbol): Likewise.
6784 (riscv_legitimize_move): Likewise.
6785 (riscv_function_value): Likewise.
6786 (riscv_hard_regno_nregs): Likewise.
6787 (riscv_expand_builtin): Likewise.
6788 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
6789 (riscv_build_integer): Likewise.
6790 (riscv_split_integer): Likewise.
6791 (riscv_legitimate_constant_p): Likewise.
6792 (riscv_cannot_force_const_mem): Likewise.
6793 (riscv_regno_mode_ok_for_base_p): Likewise.
6794 (riscv_valid_base_register_p): Likewise.
6795 (riscv_valid_offset_p): Likewise.
6796 (riscv_valid_lo_sum_p): Likewise.
6797 (riscv_classify_address): Likewise.
6798 (riscv_legitimate_address_p): Likewise.
6799 (riscv_address_insns): Likewise.
6800 (riscv_load_store_insns): Likewise.
6801 (riscv_force_binary): Likewise.
6802 (riscv_split_symbol): Likewise.
6803 (riscv_force_address): Likewise.
6804 (riscv_legitimize_address): Likewise.
6805 (riscv_move_integer): Likewise.
6806 (riscv_legitimize_const_move): Likewise.
6807 (riscv_legitimize_move): Likewise.
6808 (riscv_address_cost): Likewise.
6809 (riscv_subword): Likewise.
6810 (riscv_output_move): Likewise.
6811 (riscv_canonicalize_int_order_test): Likewise.
6812 (riscv_emit_int_order_test): Likewise.
6813 (riscv_function_arg_boundary): Likewise.
6814 (riscv_pass_mode_in_fpr_p): Likewise.
6815 (riscv_pass_fpr_single): Likewise.
6816 (riscv_pass_fpr_pair): Likewise.
6817 (riscv_get_arg_info): Likewise.
6818 (riscv_function_arg): Likewise.
6819 (riscv_function_arg_advance): Likewise.
6820 (riscv_arg_partial_bytes): Likewise.
6821 (riscv_function_value): Likewise.
6822 (riscv_pass_by_reference): Likewise.
6823 (riscv_setup_incoming_varargs): Likewise.
6824 (riscv_print_operand): Likewise.
6825 (riscv_elf_select_rtx_section): Likewise.
6826 (riscv_save_restore_reg): Likewise.
6827 (riscv_for_each_saved_reg): Likewise.
6828 (riscv_register_move_cost): Likewise.
6829 (riscv_hard_regno_mode_ok_p): Likewise.
6830 (riscv_hard_regno_nregs): Likewise.
6831 (riscv_class_max_nregs): Likewise.
6832 (riscv_memory_move_cost): Likewise.
6833 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
6834 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
6835 (rl78_addr_space_address_mode): Likewise.
6836 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6838 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6839 (rs6000_reassociation_width): Likewise.
6840 (rs6000_invalid_binary_op): Likewise.
6841 (fusion_p9_p): Likewise.
6842 (emit_fusion_p9_load): Likewise.
6843 (emit_fusion_p9_store): Likewise.
6844 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
6845 (ok_for_simple_move_operands): Likewise.
6846 (ok_for_simple_move_strict_operands): Likewise.
6847 (ok_for_simple_arith_logic_operands): Likewise.
6848 (visium_legitimize_reload_address): Likewise.
6849 (visium_select_cc_mode): Likewise.
6850 (output_cbranch): Likewise.
6851 (visium_split_double_move): Likewise.
6852 (visium_expand_copysign): Likewise.
6853 (visium_expand_int_cstore): Likewise.
6854 (visium_expand_fp_cstore): Likewise.
6855 * config/visium/visium.c (visium_pass_by_reference): Likewise.
6856 (visium_function_arg): Likewise.
6857 (visium_function_arg_advance): Likewise.
6858 (visium_libcall_value): Likewise.
6859 (visium_setup_incoming_varargs): Likewise.
6860 (visium_legitimate_constant_p): Likewise.
6861 (visium_legitimate_address_p): Likewise.
6862 (visium_legitimize_address): Likewise.
6863 (visium_secondary_reload): Likewise.
6864 (visium_register_move_cost): Likewise.
6865 (visium_memory_move_cost): Likewise.
6866 (prepare_move_operands): Likewise.
6867 (ok_for_simple_move_operands): Likewise.
6868 (ok_for_simple_move_strict_operands): Likewise.
6869 (ok_for_simple_arith_logic_operands): Likewise.
6870 (visium_function_value_1): Likewise.
6871 (rtx_ok_for_offset_p): Likewise.
6872 (visium_legitimize_reload_address): Likewise.
6873 (visium_split_double_move): Likewise.
6874 (visium_expand_copysign): Likewise.
6875 (visium_expand_int_cstore): Likewise.
6876 (visium_expand_fp_cstore): Likewise.
6877 (visium_split_cstore): Likewise.
6878 (visium_select_cc_mode): Likewise.
6879 (visium_split_cbranch): Likewise.
6880 (output_cbranch): Likewise.
6881 (visium_print_operand_address): Likewise.
6882 * expmed.c (flip_storage_order): Likewise.
6883 * expmed.h (emit_cstore): Likewise.
6884 (flip_storage_order): Likewise.
6885 * genrecog.c (validate_pattern): Likewise.
6886 * hsa-gen.c (gen_hsa_addr): Likewise.
6887 * internal-fn.c (expand_arith_overflow): Likewise.
6888 * ira-color.c (allocno_copy_cost_saving): Likewise.
6889 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6890 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
6891 (process_invariant_for_inheritance): Likewise.
6892 * lra-eliminations.c (move_plus_up): Likewise.
6893 * omp-low.c (lower_oacc_reductions): Likewise.
6894 * simplify-rtx.c (simplify_subreg): Likewise.
6895 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
6896 (TARGET_CHKP_BOUND_MODE): Likewise..
6897 * targhooks.c (default_chkp_bound_mode): Likewise.
6898 (default_setup_incoming_vararg_bounds): Likewise.
6899 * targhooks.h (default_chkp_bound_mode): Likewise.
6900 (default_setup_incoming_vararg_bounds): Likewise.
6901 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
6902 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
6903 (have_whole_vector_shift): Likewise.
6904 * tree-vect-stmts.c (vectorizable_load): Likewise.
6905 * doc/tm.texi: Regenerate.
6907 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
6909 Graceful degrade if Binutils PR21472 is not available.
6912 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
6913 .rodata in flash test fails.
6914 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
6915 * confgure: Regenerate.
6916 * config.in: Regenerate.
6917 * config/avr/avr.c (avr_asm_named_section)
6918 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
6919 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
6920 (avr_asm_init_sections): Same.
6922 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
6924 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
6925 (fma<VH:mode>4_intrinsic): Likewise.
6926 (*fmsub<VCVTF:mode>4): Likewise.
6927 (*fmsub<VH:mode>4_intrinsic): Likewise.
6929 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
6932 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
6933 Don't depend on "optimize > 0".
6934 (out_movhi_r_mr, out_movqi_mr_r): Same.
6935 (out_movhi_mr_r, out_movqi_r_mr): Same.
6936 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
6937 io_address_operand on "optimize > 0".
6939 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6941 * tree-loop-distribution.c: Add general explanantion on the pass.
6942 (generate_loops_for_partition): Mark distributed loop.
6943 (pg_add_dependence_edges): New parameter. Handle alias data
6944 dependence specially and record it in the parameter if asked.
6945 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
6946 (init_partition_graph_vertices, add_partition_graph_edge): New.
6947 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
6948 (free_partition_graph_vdata, build_partition_graph): New.
6949 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
6950 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
6951 (data_ref_segment_size, latch_dominated_by_data_ref): New.
6952 (compute_alias_check_pairs, version_loop_by_alias_check): New.
6953 (version_for_distribution_p, finalize_partitions): New.
6954 (distribute_loop): Handle alias data dependence specially. Factor
6955 out loop fusion code as functions and call these functions.
6957 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6959 * tree-loop-distribution.c (classify_partition): New parameter and
6960 better handle reduction statement.
6961 (rdg_build_partitions): Revise comment.
6962 (distribute_loop): Compute statements in all partitions and pass it
6963 to classify_partition.
6965 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6967 * tree-loop-distribution.c (enum partition_type): New.
6968 (struct partition): New field type.
6969 (partition_merge_into): Add parameter. Update partition type.
6970 (data_dep_in_cycle_p, update_type_for_merge): New functions.
6971 (build_rdg_partition_for_vertex): Compute partition type.
6972 (rdg_build_partitions): Dump partition type.
6973 (distribute_loop): Update calls to partition_merge_into.
6975 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6977 * tree-loop-distribution.c (struct ddr_hasher): New.
6978 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
6980 (classify_partition): Call get_data_dependence.
6981 (pg_add_dependence_edges): Ditto.
6982 (distribute_loop): Release data dependence hash table.
6984 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6986 * tree-loop-distribution.c (ref_base_address): Delete.
6987 (similar_memory_accesses): Rename ...
6988 (share_memory_accesses): ... to this. Check if partitions access
6989 the same memory reference.
6990 (distribute_loop): Call share_memory_accesses.
6992 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6994 * tree-loop-distribution.c (struct partition): New field recording
6996 (partition_alloc, partition_free): Init and release data refs.
6997 (partition_merge_into): Merge data refs.
6998 (build_rdg_partition_for_vertex): Collect data refs for partition.
6999 (pg_add_dependence_edges): Change parameters from vector to bitmap.
7001 (distribute_loop): Remve data refs from vertice data of partition
7004 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7006 * tree-loop-distribution.c (params.h): Include header file.
7007 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
7008 (datarefs_vec): New global var.
7009 (create_rdg_vertices): Use datarefs_vec directly.
7010 (free_rdg): Don't free data references.
7011 (build_rdg): Update use. Don't free data references.
7012 (distribute_loop): Compute global variable for data references.
7013 Bail out if there are too many data references.
7015 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7017 * tree-loop-distribution.c (loop_nest): New global var.
7018 (build_rdg): Use loop directly, rather than loop nest.
7019 (pg_add_dependence_edges): Remove loop nest parameter. Use global
7021 (distribute_loop): Compute global variable loop nest. Update use.
7023 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7025 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
7026 (partition_merge_into): New parameter. Dump reason for fusion.
7027 (distribute_loop): Update use of partition_merge_into.
7029 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7031 * tree-loop-distribution.c (bb_top_order_index): New.
7032 (bb_top_order_index_size, bb_top_order_cmp): New.
7033 (stmts_from_loop): Use topological order.
7034 (pass_loop_distribution::execute): Compute and release topological
7035 order for basic blocks.
7037 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7039 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
7042 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7044 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
7045 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
7046 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
7047 * internal-fn.def (LOOP_DIST_ALIAS): New.
7048 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
7049 (fold_loop_internal_call): ... this.
7050 (vect_loop_dist_alias_call): New function.
7051 (set_uid_loop_bbs): Call fold_loop_internal_call.
7052 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
7055 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7058 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
7059 Require dead FLAGS_REG at the beginning of a peephole.
7061 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7064 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
7065 arguments in the call to __builtin_ia32_sbb_u32.
7066 (_subborrow_u64): Swap _X and _Y arguments in the call to
7067 __builtin_ia32_sbb_u64.
7069 2017-07-04 Jakub Jelinek <jakub@redhat.com>
7072 * tree-vrp.c (compare_assert_loc): Turn into a function template
7073 with stable template parameter. Only test if a->e is NULL,
7074 !a->e == !b->e has been verified already. Use e == NULL or
7075 e != NULL instead of e or ! e tests. If stable is true, don't use
7076 iterative_hash_expr, on the other side allow a or b or both NULL
7077 and sort the NULLs last.
7078 (process_assert_insertions): Sort using compare_assert_loc<false>
7079 instead of compare_assert_loc, later sort using
7080 compare_assert_loc<true> before calling process_assert_insertions_for
7081 in a loop. Use break instead of continue once seen NULL pointer.
7083 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
7085 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7086 Cortex-R7 and Cortex-R8 processors.
7088 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7090 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
7091 uninitialized while src is not.
7093 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
7095 * common/config/arm/arm-common.c: Adjust include path for
7097 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
7098 (arm-cpu.h): Create in build directory. Adjust dependency rules.
7099 (arm-cpu-data.h): Likewise.
7100 (arm-cpu-cdata.h): Likewise.
7101 * config/arm/arm-cpu.h: Delete.
7102 * config/arm/arm-cpu-cdata.h: Delete.
7103 * config/arm/arm-cpu-data.h: Delete.
7105 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
7107 * config/arm/arm-cpus.in (cortex-a55): New.
7108 (cortex-a75): Likewise.
7109 (cortex-a75.cortex-a55): Likewise.
7110 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
7112 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
7113 * config/arm/arm-cpu-cdata.h: Regenerate.
7114 * config/arm/arm-cpu-data.h: Regenerate.
7115 * config/arm/arm-cpu.h: Regenerate.
7116 * config/arm/arm-tables.opt: Regenerate.
7117 * config/arm/arm-tune.md: Regenerate.
7119 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7121 * haifa-sched.c (sched_create_recovery_edges): Update profile.
7123 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7125 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
7128 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
7130 PR tree-optimization/81292
7131 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
7132 full_string_p, also call adjust_related_strinfos if the adjustment
7133 is simple, otherwise invalidate related strinfos.
7135 2017-07-04 Martin Liska <mliska@suse.cz>
7138 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
7139 newly created variable as DECL_IGNORED_P.
7141 2017-07-04 Martin Liska <mliska@suse.cz>
7144 * ipa-inline.c (inline_small_functions):
7145 Use xstrdup_for_dump.
7147 2017-07-04 Tom de Vries <tom@codesourcery.com>
7149 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
7151 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
7154 * config/darwin.c (darwin_function_switched_text_sections):
7157 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
7159 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
7161 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
7163 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
7165 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7167 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
7168 min_profitable_iters, and th as inclusive lower bounds.
7169 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
7170 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
7171 for min_profitable_iters and min_profitable_estimate.
7172 (vect_transform_loop): Treat th as an inclusive lower bound.
7173 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
7175 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
7178 * config/darwin.c (darwin_function_switched_text_sections):
7179 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
7180 in two pieces, and suppress the use of buf.
7182 2017-07-03 Nathan Sidwell <nathan@acm.org>
7184 * hash-table.h (hash_table_mod1): Fix indentation.
7186 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7189 * predict.c (force_edge_cold): Be more careful about propagation
7191 * profile-count.h (profile_probability::guessed,
7192 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
7194 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
7196 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
7198 * doc/invoke.texi (rcpc architecture extension): Document it.
7200 2017-07-03 Richard Biener <rguenther@suse.de>
7202 PR tree-optimization/60510
7203 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
7204 the scalar reduction PHI and use it.
7205 (vectorizable_reduction): Properly guard the single_defuse_cycle
7206 path for non-SLP reduction chains where we cannot use it.
7207 Rework reduc_def/index and vector type deduction. Rework
7208 vector operand gathering during reduction op code-gen.
7209 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
7210 chains dissolve the chain and leave it to non-SLP reduction
7213 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7215 * tree-data-ref.h (dr_alignment): Declare.
7216 * tree-data-ref.c (dr_alignment): New function.
7217 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
7218 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
7220 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
7222 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7224 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
7225 and base_misalignment fields.
7226 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
7227 * tree-data-ref.c: Include builtins.h.
7228 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
7229 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
7230 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
7231 * tree-vect-data-refs.c: Include tree-cfg.h.
7232 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
7233 fields instead of calculating an alignment here.
7234 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
7235 innermost_loop_behavior fields.
7237 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7239 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
7241 (DR_STEP_ALIGNMENT): New macro.
7242 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
7243 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
7244 (create_data_ref): Print it.
7245 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
7246 to tell whether the step preserves vector (mis)alignment.
7247 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7248 Move the check for an integer step and generalise to all INTEGER_CST.
7249 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
7250 Print the outer step alignment.
7252 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7254 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
7255 with offset_alignment.
7256 (DR_ALIGNED_TO): Delete.
7257 (DR_OFFSET_ALIGNMENT): New macro.
7258 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
7259 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
7260 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
7261 (create_data_ref): Likewise.
7262 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7263 (vect_analyze_data_refs): Likewise.
7264 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
7265 creating dummy innermost behavior.
7267 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7269 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
7270 with a "innermost_loop_behavior *" and refeence tree.
7271 * tree-data-ref.c (dr_analyze_innermost): Likewise.
7272 (create_data_ref): Update call accordingly.
7273 * tree-predcom.c (find_looparound_phi): Likewise.
7275 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7277 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
7278 fields with dr_wrt_vec_loop.
7279 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
7280 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
7281 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
7282 (vect_dr_behavior): New function.
7283 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7284 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
7285 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
7286 track whether the step preserves the misalignment.
7287 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7288 Use vect_dr_behavior.
7289 (vect_setup_realignment): Update call accordingly.
7290 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
7291 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
7292 call to vect_create_addr_base_for_vector_ref.
7293 (vect_create_cond_for_align_checks): Likewise.
7294 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
7295 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
7296 (vect_recog_mask_conversion_pattern): Likewise.
7297 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
7298 (new_stmt_vec_info): Remove redundant zeroing.
7300 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
7302 * common/config/arm/arm-common.c (arm_be8_option): New function.
7303 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
7304 (ISA_ARMv6): Add isa_bit_be8.
7305 * config/arm/arm.h (arm_be8_option): Add prototype.
7306 (BE8_SPEC_FUNCTION): New define.
7307 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
7308 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
7309 (mlittle-endian): Similarly.
7310 (mbe8, mbe32): New options.
7311 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
7312 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
7314 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7316 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
7318 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7320 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
7321 (cleanup_tree_cfg_bb): Use it.
7322 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
7324 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
7326 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7329 * loop-doloop.c (add_test): Update profile.
7331 2017-07-03 Martin Liska <mliska@suse.cz>
7334 * sanopt.c (rewrite_usage_of_param): New function.
7335 (sanitize_rewrite_addressable_params): Likewise.
7336 (pass_sanopt::execute): Call rewrite_usage_of_param.
7338 2017-07-03 Richard Biener <rguenther@suse.de>
7340 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
7341 back to using VIEW_CONVERT_EXPR.
7343 2017-07-03 Martin Liska <mliska@suse.cz>
7346 * doc/extend.texi: Document when a resolver function is
7347 generated for target_clones.
7349 2017-07-03 Martin Liska <mliska@suse.cz>
7351 * asan.c (asan_emit_stack_protection): Unpoison just red zones
7352 and shadow memory of auto variables which are subject of
7353 use-after-scope sanitization.
7354 (asan_expand_mark_ifn): Add do set only when is_poison.
7356 2016-07-03 Richard Biener <rguenther@suse.de>
7358 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
7360 (vect_force_simple_reduction): Record reduction def -> phi mapping.
7361 (vectorizable_reduction): Perform reduction PHI creation when
7362 visiting a reduction PHI and adjust and simplify code generation
7363 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
7364 (vect_transform_loop): Visit reduction PHIs.
7365 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
7366 defs into the SLP tree.
7367 (vect_build_slp_tree): Reduction defs terminate the recursion.
7368 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
7370 (vect_get_vec_defs_for_stmt_copy): Export.
7371 (vect_get_vec_defs): Likewise.
7372 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
7374 (vect_get_vec_defs_for_stmt_copy): Declare.
7375 (vect_get_vec_defs): Likewise.
7377 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7379 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
7380 parameter with a "loop" parameter and use it instead of the
7381 loop containing DR_STMT. Don't check simple_iv when doing
7382 BB analysis. Describe the two analysis modes in the comment.
7384 2017-07-03 Tom de Vries <tom@codesourcery.com>
7386 PR tree-optimization/69468
7387 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
7388 (find_same_succ_bb): Handle ignore_edge_flags.
7390 2017-07-03 Tom de Vries <tom@codesourcery.com>
7392 PR tree-optimization/81192
7393 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
7395 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
7397 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
7399 2017-07-03 Tom de Vries <tom@codesourcery.com>
7401 PR tree-optimization/81192
7402 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
7403 BB_SAME_SUCC (bb) == NULL.
7405 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7407 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
7410 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7412 * dumpfile.c: Include profile-count.h
7413 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
7415 (insert_cond_bb): Update profile.
7416 * tree-cfg.h (insert_cond_bb): Update prototype.
7417 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
7418 * tree-dump.c: Do not include tree-cfg.
7420 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7422 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
7424 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7426 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
7429 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7431 * tree-complex.c (expand_complex_div_wide): update profile.
7433 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7434 Alan Hayward <alan.hayward@arm.com>
7435 David Sherwood <david.sherwood@arm.com>
7437 * Makefile.in (MACHMODE_H): Remove insn-modes.h
7438 (CORETYPES_H): New define.
7439 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
7440 (insn-modes-inline.h, s-modes-inline-h): New rules.
7441 (generated_files): Add insn-modes-inline.h.
7442 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
7443 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
7444 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
7445 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
7446 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
7447 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
7448 (build/gencodes.o, build/genconditions.o): Likewise.
7449 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
7450 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
7451 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
7452 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
7453 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
7454 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
7455 * coretypes.h: Include everything up to real.h for generators.
7456 Include insn-modes.h first. Include wide-int-print.h after
7457 wide-int.h. Include insn-modes-inline.h and then machmode.h.
7458 * machmode.h: Don't include insn-modes.h here.
7459 * function-tests.c: Remove includes of signop.h, machmode.h,
7460 double-int.h and wide-int.h.
7462 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
7464 * optc-save-gen.awk: Likewise.
7465 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
7466 * godump.c: Remove include of wide-int-print.h.
7467 * pretty-print.h: Likewise.
7468 * wide-int-print.cc: Likewise.
7469 * wide-int.cc: Likewise.
7470 * hash-map-tests.c: Remove include of signop.h.
7471 * hash-set-tests.c: Likewise.
7472 * rtl-tests.c: Likewise.
7473 * mkconfig.sh: Remove include of machmode.h.
7474 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
7476 (emit_insn_modes_inline_h): ...this new function. Emit the code
7477 into an insn-modes-inline.h header file, adding appropriate
7478 include guards and end comments.
7479 (emit_insn_modes_c_header): Remove include of machmode.h.
7480 (emit_min_insn_modes_c_header): Include coretypes.h rather than
7482 (main): Handle -i flag and call emit_insn_modes_inline_h when
7485 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7487 * tree-ssa-strlen.c (strinfo): Rename the length field to
7488 nonzero_chars. Add a full_string_p field.
7489 (compare_nonzero_chars, zero_length_string_p): New functions.
7490 (get_addr_stridx): Add an offset_out parameter.
7491 Use compare_nonzero_chars.
7492 (get_stridx): Update accordingly. Use compare_nonzero_chars.
7493 (new_strinfo): Update after above changes to strinfo.
7494 (set_endptr_and_length): Set full_string_p.
7495 (get_string_length): Update after above changes to strinfo.
7496 (unshare_strinfo): Update call to new_strinfo.
7497 (maybe_invalidate): Likewise.
7498 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
7499 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
7500 as a uhwi instead of an shwi. Update after above changes to
7501 strinfo and new_strinfo.
7502 (zero_length_string): Assert that chainsi contains full strings.
7503 Use zero_length_string_p. Update call to new_strinfo.
7504 (adjust_related_strinfos): Update after above changes to strinfo.
7505 Copy full_string_p from origsi.
7506 (adjust_last_stmt): Use zero_length_string_p.
7507 (handle_builtin_strlen): Update after above changes to strinfo and
7508 new_strinfo. Install the lhs as the string length if the previous
7509 entry didn't describe a full string.
7510 (handle_builtin_strchr): Update after above changes to strinfo
7512 (handle_builtin_strcpy): Likewise.
7513 (handle_builtin_strcat): Likewise.
7514 (handle_builtin_malloc): Likewise.
7515 (handle_pointer_plus): Likewise.
7516 (handle_builtin_memcpy): Likewise. Track nonzero characters
7517 that aren't necessarily followed by a nul terminator.
7518 (handle_char_store): Likewise.
7520 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7522 PR tree-optimization/80769
7523 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
7524 for malloc and calloc. Document the new invariant that all related
7525 strinfos have delayed lengths or none do.
7526 (verify_related_strinfos): Move earlier in file.
7527 (set_endptr_and_length): New function, split out from...
7528 (get_string_length): ...here. Also set the lengths of related
7530 (zero_length_string): Assert that chainsi has known (rather than
7532 (adjust_related_strinfos): Likewise.
7534 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7536 PR tree-optimization/81136
7537 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
7538 assert that two references with the same misalignment have the same
7539 compile-time misalignment if those compile-time misalignments
7542 2017-07-01 Andi Kleen <ak@linux.intel.com>
7544 * print-tree.c (print_node): Print all attributes.
7546 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7548 * cfg.c (scale_bbs_frequencies): New function.
7549 * cfg.h (scale_bbs_frequencies): Declare it.
7550 * cfgloopanal.c (single_likely_exit): Cleanup.
7551 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
7553 (scale_loop_profile): Likewise.
7554 (loop_version): Likewise.
7555 (create_empty_loop_on_edge): Update.
7556 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
7557 scale_loop_frequencies, scale_loop_profile, loopify,
7558 loop_version): Update prototypes.
7559 * modulo-sched.c (sms_schedule): Update.
7560 * predict.c (unlikely_executed_edge_p): Also check probability.
7561 (probably_never_executed_edge_p): Fix typo.
7562 * tree-if-conv.c (version_loop_for_if_conversion): Update.
7563 * tree-parloops.c (gen_parallel_loop): Update.
7564 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
7565 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7566 * tree-ssa-loop-split.c (split_loop): Update.
7567 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
7568 * tree-vect-loop-manip.c (vect_do_peeling): Update.
7569 (vect_loop_versioning): Update.
7570 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7572 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7574 * trans-mem.c (split_bb_make_tm_edge): Update profile.
7576 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7578 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
7579 to keep profile consistent.
7581 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7583 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
7584 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
7585 * profile-count.h (max_safe_multiplier): Make unsigned.
7586 (profile_count::guessed_zero): New.
7588 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7590 * bb-reorder.c (fix_up_crossing_landing_pad,
7591 fix_crossing_conditional_branches): Use make_single_succ_edge
7592 to keep profile consistent.
7594 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7596 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
7599 2017-07-01 Jakub Jelinek <jakub@redhat.com>
7602 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
7603 the right scopes, make sure cond_jump isn't preserved between multiple
7604 iterations. Search for fallthru edge whenever there are 3+ edges and
7605 use find_fallthru_edge for it.
7607 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7609 Patch by Alexander Monakov <amonakov@ispras.ru>
7610 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
7611 probabilities consistently.
7613 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7615 * pa.c (pa_expand_compare_and_swap_loop): Update call of
7616 emit_cmp_and_jump_insns.
7618 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7621 * tree-inline.c (expand_call_inline): Combine profile statuses.
7623 2017-06-30 Andrew Pinski <apinski@cavium.com>
7625 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
7626 fold_stmt returned true.
7628 2017-06-30 Nathan Sidwell <nathan@acm.org>
7630 * ggc.h (empty_string): Delete.
7631 * cfgexpand.c (expand_asm_stmt): Use plain "".
7632 * optabs.c (expand_asm_memory_barrier): Likewise.
7633 * stringpool.c (empty_string): Delete.
7634 (digit_vector, digit_string): Delete.
7635 (ggc_alloc_string): Use plain "", don't optimize single digit
7636 strings. Use ggc_alloc_atomic.
7638 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
7640 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
7641 comparison set and one other set, use the cost of the non-comparison
7644 2017-06-30 Nathan Sidwell <nathan@acm.org>
7646 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
7649 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
7651 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
7652 loops. Remove now unneeded calls to gimple_switch_set_label() that
7653 just set removed labels to NULL_TREE.
7655 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
7657 * tree-ssanames.c (set_range_info_raw): Abstract from ...
7658 (set_range_info): ...here. Only call set_range_info_raw if domain
7660 (set_nonzero_bits): Call set_range_info_raw.
7661 * tree-ssanames.h (set_range_info_raw): New.
7663 2017-06-30 Jakub Jelinek <jakub@redhat.com>
7666 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
7667 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
7668 of nonimmediate_operand and <store_mask_constraint> instead of m
7669 for the input operand. For V8FI iterator, always split if input
7670 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
7671 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
7672 <store_mask_predicate> instead of register_operand and
7673 <store_mask_constraint> instead of v for the input operand. Make
7674 sure both operands aren't MEMs for if not <mask_applied>.
7676 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
7678 * lto-wrapper.c (copy_file) Close both file descriptors before
7681 2017-06-30 Martin Liska <mliska@suse.cz>
7684 * multiple_target.c (create_dispatcher_calls): Make ifunc
7685 also for function that don't have calls or are not referenced.
7687 2017-06-30 Richard Biener <rguenther@suse.de>
7689 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
7690 analyze the first scalar stmt. Move vector type computation
7691 for the BB case here from ...
7692 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
7693 live operation processing in the SLP case properly.
7695 2017-06-30 Richard Biener <rguenther@suse.de>
7697 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
7699 2017-06-30 Martin Liska <mliska@suse.cz>
7702 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
7703 before BUILT_IN_UNWIND_RESUME when ASAN is used.
7705 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
7707 * doc/invoke.texi (AArch64): Add missing options and remove redundant
7710 2017-06-30 Richard Biener <rguenther@suse.de>
7712 PR tree-optimization/81249
7713 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
7714 condition reduction result to original scalar type.
7716 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7718 * profile-count.h (enum profile_quality): Fix typos and whitespace
7721 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7723 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
7724 type for branch probabilities.
7726 2017-06-29 Julian Brown <julian@codesourcery.com>
7727 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7729 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
7730 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
7731 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
7732 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
7734 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7736 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
7737 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
7738 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
7739 CC usage from generic code to here.
7740 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
7741 CC usage into the target macros.
7743 2017-06-29 Maya Rashish <coypu@sdf.org>
7745 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
7748 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7750 * arm/arm-builtins.c: Include profile-count.h
7751 * except.c (sjlj_emit_function_enter): Use
7752 profile_probability::unlikely.
7754 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7756 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
7757 and tocrel_offset be pointer args rather than implicitly using
7759 (legitimate_constant_pool_address_p, rs6000_emit_move,
7760 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
7761 tocrel_offset and use in toc_relative_expr_p call.
7762 (print_operand, print_operand_address): Use static tocrel_base_oac
7763 and tocrel_offset_oac.
7764 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
7767 2017-06-29 Maya Rashish <coypu@sdf.org>
7769 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
7771 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
7773 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
7774 objects, take into account only the alignment of 'op0' and 'mode1' if
7777 2017-06-29 Steve Ellcey <sellcey@cavium.com>
7779 * ccmp.c (ccmp_tree_comparison_p): New function.
7780 (ccmp_candidate_p): Update to use above function.
7781 (get_compare_parts): New function.
7782 (expand_ccmp_next): Update to use new functions.
7783 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
7785 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
7786 take mode as argument.
7787 * ccmp.h (expand_ccmp_expr): Add mode as argument.
7788 * expr.c (expand_expr_real_1): Pass mode as argument.
7790 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
7792 * combine.c (combine_instructions): Print insns to dump_file, together
7795 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7797 * asan.c (asan_emit_stack_protection): Update.
7798 (create_cond_insert_point): Update.
7799 * auto-profile.c (afdo_propagate_circuit): Update.
7800 * basic-block.h (struct edge_def): Turn probability to
7801 profile_probability.
7802 (EDGE_FREQUENCY): Update.
7803 * bb-reorder.c (find_traces_1_round): Update.
7804 (better_edge_p): Update.
7805 (sanitize_hot_paths): Update.
7806 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
7807 (make_single_succ_edge): Update.
7808 (check_bb_profile): Update.
7809 (dump_edge_info): Update.
7810 (update_bb_profile_for_threading): Update.
7811 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
7812 probabilitycount to 0.
7813 * cfgbuild.c (compute_outgoing_frequencies): Update.
7814 * cfgcleanup.c (try_forward_edges): Update.
7815 (outgoing_edges_match): Update.
7816 (try_crossjump_to_edge): Update.
7817 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
7818 (expand_gimple_tailcall): Update.
7819 (construct_init_block): Use make_single_succ_edge.
7820 (construct_exit_block): Use make_single_succ_edge.
7821 * cfghooks.c (verify_flow_info): Update.
7822 (redirect_edge_succ_nodup): Update.
7823 (split_edge): Update.
7824 (account_profile_record): Update.
7825 * cfgloopanal.c (single_likely_exit): Update.
7826 * cfgloopmanip.c (scale_loop_profile): Update.
7827 (set_zero_probability): Remove.
7828 (duplicate_loop_to_header_edge): Update.
7829 * cfgloopmanip.h (loop_version): Update prototype.
7830 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7831 (force_nonfallthru_and_redirect): Update.
7832 (update_br_prob_note): Update.
7833 (rtl_verify_edges): Update.
7834 (purge_dead_edges): Update.
7835 (rtl_lv_add_condition_to_bb): Update.
7836 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
7837 * cgraphunit.c (init_lowered_empty_function): Update.
7838 (cgraph_node::expand_thunk): Update.
7839 * cilk-common.c: Include profile-count.h
7840 * dojump.c (inv): Remove.
7841 (jumpifnot): Update.
7842 (jumpifnot_1): Update.
7843 (do_jump_1): Update.
7845 (do_jump_by_parts_greater_rtx): Update.
7846 (do_compare_rtx_and_jump): Update.
7847 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
7848 do_jump_1. do_compare_rtx_and_jump): Update prototype.
7849 * dwarf2cfi.c: Include profile-count.h
7850 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
7851 (sjlj_emit_dispatch_table): Likewise.
7852 * explow.c: Include profile-count.h
7853 * expmed.c (emit_store_flag_force): Update.
7854 (do_cmp_and_jump): Update.
7855 * expr.c (compare_by_pieces_d::generate): Update.
7856 (compare_by_pieces_d::finish_mode): Update.
7857 (emit_block_move_via_loop): Update.
7858 (store_expr_with_bounds): Update.
7859 (store_constructor): Update.
7860 (expand_expr_real_2): Update.
7861 (expand_expr_real_1): Update.
7862 * expr.h (try_casesi, try_tablejump): Update prototypes.
7863 * gimple-pretty-print.c (dump_probability): Update.
7864 (dump_profile): New.
7865 (dump_gimple_label): Update.
7866 (dump_gimple_bb_header): Update.
7867 * graph.c (draw_cfg_node_succ_edges): Update.
7868 * hsa-gen.c (convert_switch_statements): Update.
7869 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
7870 (find_if_case_1): Update.
7871 (find_if_case_2): Update.
7872 * internal-fn.c (expand_arith_overflow_result_store): Update.
7873 (expand_addsub_overflow): Update.
7874 (expand_neg_overflow): Update.
7875 (expand_mul_overflow): Update.
7876 (expand_vector_ubsan_overflow): Update.
7877 * ipa-cp.c (good_cloning_opportunity_p): Update.
7878 * ipa-split.c (split_function): Use make_single_succ_edge.
7879 * ipa-utils.c (ipa_merge_profiles): Update.
7880 * loop-doloop.c (add_test): Update.
7881 (doloop_modify): Update.
7882 * loop-unroll.c (compare_and_jump_seq): Update.
7883 (unroll_loop_runtime_iterations): Update.
7884 * lra-constraints.c (lra_inheritance): Update.
7885 * lto-streamer-in.c (input_cfg): Update.
7886 * lto-streamer-out.c (output_cfg): Update.
7887 * mcf.c (adjust_cfg_counts): Update.
7888 * modulo-sched.c (sms_schedule): Update.
7889 * omp-expand.c (expand_omp_for_init_counts): Update.
7890 (extract_omp_for_update_vars): Update.
7891 (expand_omp_ordered_sink): Update.
7892 (expand_omp_for_ordered_loops): Update.
7893 (expand_omp_for_generic): Update.
7894 (expand_omp_for_static_nochunk): Update.
7895 (expand_omp_for_static_chunk): Update.
7896 (expand_cilk_for): Update.
7897 (expand_omp_simd): Update.
7898 (expand_omp_taskloop_for_outer): Update.
7899 (expand_omp_taskloop_for_inner): Update.
7900 * omp-simd-clone.c (simd_clone_adjust): Update.
7901 * optabs.c (expand_doubleword_shift): Update.
7902 (expand_abs): Update.
7903 (emit_cmp_and_jump_insn_1): Update.
7904 (expand_compare_and_swap_loop): Update.
7905 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
7906 * predict.c (predictable_edge_p): Update.
7907 (edge_probability_reliable_p): Update.
7908 (set_even_probabilities): Update.
7909 (combine_predictions_for_insn): Update.
7910 (combine_predictions_for_bb): Update.
7911 (propagate_freq): Update.
7912 (estimate_bb_frequencies): Update.
7913 (force_edge_cold): Update.
7914 * profile-count.c (profile_count::dump): Add missing space into dump.
7915 (profile_count::debug): Add newline.
7916 (profile_count::differs_from_p): Explicitly convert to unsigned.
7917 (profile_count::stream_in): Update.
7918 (profile_probability::dump): New member function.
7919 (profile_probability::debug): New member function.
7920 (profile_probability::differs_from_p): New member function.
7921 (profile_probability::differs_lot_from_p): New member function.
7922 (profile_probability::stream_in): New member function.
7923 (profile_probability::stream_out): New member function.
7924 * profile-count.h (profile_count_quality): Rename to ...
7925 (profile_quality): ... this one.
7926 (profile_probability): New.
7927 (profile_count): Update.
7928 * profile.c (compute_branch_probabilities): Update.
7929 * recog.c (peep2_attempt): Update.
7930 * sched-ebb.c (schedule_ebbs): Update.
7931 * sched-rgn.c (find_single_block_region): Update.
7932 (compute_dom_prob_ps): Update.
7933 (schedule_region): Update.
7934 * sel-sched-ir.c (compute_succs_info): Update.
7935 * stmt.c (struct case_node): Update.
7936 (do_jump_if_equal): Update.
7937 (get_outgoing_edge_probs): Update.
7938 (conditional_probability): Update.
7939 (emit_case_dispatch_table): Update.
7940 (expand_case): Update.
7941 (expand_sjlj_dispatch_table): Update.
7942 (emit_case_nodes): Update.
7943 * targhooks.c: Update.
7944 * tracer.c (better_p): Update.
7945 (find_best_successor): Update.
7946 * trans-mem.c (expand_transaction): Update.
7947 * tree-call-cdce.c: Update.
7948 * tree-cfg.c (gimple_split_edge): Upate.
7949 (move_sese_region_to_fn): Upate.
7950 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
7951 * tree-eh.c (lower_resx): Upate.
7952 (cleanup_empty_eh_move_lp): Upate.
7953 * tree-if-conv.c (version_loop_for_if_conversion): Update.
7954 * tree-inline.c (copy_edges_for_bb): Update.
7955 (copy_cfg_body): Update.
7956 * tree-parloops.c (gen_parallel_loop): Update.
7957 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
7958 (gimple_gen_time_profiler): Update.
7959 * tree-ssa-dce.c (remove_dead_stmt): Update.
7960 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7961 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
7962 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
7963 (unloop_loops): Update.
7964 (try_peel_loop): Update.
7965 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7966 * tree-ssa-loop-split.c (connect_loops): Update.
7967 (split_loop): Update.
7968 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
7969 (hoist_guard): Update.
7970 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
7971 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
7972 (value_replacement): Update.
7973 * tree-ssa-reassoc.c (branch_fixup): Update.
7974 * tree-ssa-tail-merge.c (replace_block_by): Update.
7975 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
7976 (create_edge_and_update_destination_phis): Update.
7977 (compute_path_counts): Update.
7978 (recompute_probabilities): Update.
7979 (update_joiner_offpath_counts): Update.
7980 (freqs_to_counts_path): Update.
7981 (duplicate_thread_path): Update.
7982 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
7983 (struct switch_conv_info): Update.
7984 (gen_inbound_check): Update.
7985 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
7986 (vect_do_peeling): Update.
7987 (vect_loop_versioning): Update.
7988 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7989 (optimize_mask_stores): Update.
7990 * ubsan.c (ubsan_expand_null_ifn): Update.
7991 * value-prof.c (gimple_divmod_fixed_value): Update.
7992 (gimple_divmod_fixed_value_transform): Update.
7993 (gimple_mod_pow2): Update.
7994 (gimple_mod_pow2_value_transform): Update.
7995 (gimple_mod_subtract): Update.
7996 (gimple_mod_subtract_transform): Update.
7997 (gimple_ic): Update.
7998 (gimple_stringop_fixed_value): Update.
7999 (gimple_stringops_transform): Update.
8000 * value-prof.h: Update.
8002 2017-06-29 Carl Love <cel@us.ibm.com>
8004 * config/rs6000/rs6000-c.c: Add support for built-in functions
8005 vector signed int vec_signed (vector float);
8006 vector signed long long vec_signed (vector double);
8007 vector signed int vec_signed2 (vector double, vector double);
8008 vector signed int vec_signede (vector double);
8009 vector signed int vec_signedo (vector double);
8010 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
8011 instruction generator.
8012 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8013 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
8014 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
8016 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
8017 vunsignede_v2df): Add define_expands.
8018 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
8019 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
8020 VEC_UNSIGNEDO): Add definitions.
8021 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8022 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
8023 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
8024 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
8025 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
8026 * config/rs6000/altivec.h (vec_signed, vec_signed2,
8027 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
8028 vec_unsignede, vec_unsignedo): Add builtin defines.
8029 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
8031 * doc/extend.texi: Update the built-in documentation file for the
8032 new built-in functions.
8034 2017-06-29 Richard Biener <rguenther@suse.de>
8036 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
8037 reduction chains to LOOP_VINFO_REDUCTIONS.
8038 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
8039 SLP reductions after processing reduction chains.
8041 2017-06-29 Nathan Sidwell <nathan@acm.org>
8043 * builtins.c (fold_builtin_FUNCTION): Use
8044 lang_hooks.decl_printable_name.
8046 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
8049 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
8050 with only one label.
8051 * stmt.c (expand_case): Assert NCASES is greater than one.
8053 2017-06-29 Richard Biener <rguenther@suse.de>
8055 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
8057 (group_case_labels): Likewise.
8058 (find_taken_edge): Push sanity checking on val to workers...
8059 (find_taken_edge_cond_expr): ... here
8060 (find_taken_edge_switch_expr): ... and here, handle cases
8061 with just a default label.
8062 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
8063 (group_case_labels): Likewise.
8064 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
8065 group_case_labels does anything cleanup the CFG again.
8067 2017-06-29 Bin Cheng <bin.cheng@arm.com>
8069 PR tree-optimization/81196
8070 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
8071 exit condition comparing two IVs.
8073 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
8075 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
8076 profile to the dummy entry at the end of the list of architectures.
8077 * config/arm/arm-cpu-cdata.h: Regenerated.
8079 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8080 Michael Collison <michael.collison@arm.com>
8083 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
8085 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
8086 (*aarch64_reg_<mode>3_minus_mask): New pattern.
8087 (*aarch64_<optab>_reg_di3_mask2): New pattern.
8088 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
8089 of shift when the shift amount is masked with constant equal to
8090 the size of the mode.
8091 * config/aarch64/predicates.md (subreg_lowpart_operator): New
8094 2017-06-29 Martin Liska <mliska@suse.cz>
8096 * config/i386/i386.opt: Change range from [1,5] to [0,5].
8098 2017-06-29 Yury Gribov <tetra2005@gmail.com>
8101 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
8104 (edge_growth_cache_entry::edge_growth_cache_entry): New
8106 (reset_edge_growth_cache): Update to use constructor.
8108 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8110 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
8111 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
8112 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
8114 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
8116 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
8117 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
8119 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
8121 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
8122 (*-linux-uclibc*): Add t-uclibc tmake_file.
8123 * config/t-musl: New.
8124 * config/t-uclibc: New.
8126 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
8128 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
8130 (gen_comm_data): Emit architectural setting of arch_prof.
8131 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
8133 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
8134 (armv8-m.base, armv8-m.main): Likewise.
8135 * arm-protos.h (arm_build_target): Add profile field.
8136 (arch_option): Likewise.
8137 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
8139 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
8140 arm_active_target.profile.
8142 2017-06-28 Richard Biener <rguenther@suse.de>
8145 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
8146 TYPE_OVERFLOW_WRAPS.
8147 * match.pd (negate_expr_p): Likewise.
8148 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
8149 fold_build2, not fold_binary.
8151 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
8153 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
8154 Convert memory address to Pmode.
8155 (aarch64_print_operand): Assert MEM operands are always Pmode.
8157 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
8160 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
8161 Remove redundant if.
8162 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
8163 * config/arm/aarch-common-protos.h
8164 (aarch_forward_to_shift_is_not_shifted_re): Remove.
8165 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
8167 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
8170 * multiple_target.c (create_dispatcher_calls): Set the default
8171 clone to be static, not public.
8173 2017-06-28 Richard Biener <rguenther@suse.de>
8175 * tree-vect-loop.c (vectorizable_reduction): Move special
8176 cond reduction IV var creation ...
8177 (vect_create_epilog_for_reduction): ... here. Remove induction_index
8178 parameter. Use STMT_VINFO_VECTYPE.
8179 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
8182 2017-06-28 Martin Liska <mliska@suse.cz>
8185 * ipa-visibility.c (non_local_p): Handle visibility.
8187 2017-06-28 Martin Liska <mliska@suse.cz>
8190 * common.opt: Add IntegerRange to various options.
8191 * opt-functions.awk (integer_range_info): New function.
8192 * optc-gen.awk: Add integer_range_info to cl_options struct.
8193 * opts-common.c (decode_cmdline_option): Handle
8194 CL_ERR_INT_RANGE_ARG.
8195 (cmdline_handle_error): Likewise.
8196 * opts.c (print_filtered_help): Show valid interval in
8197 when --help is provided.
8198 * opts.h (struct cl_option): Add range_min and range_max fields.
8199 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
8201 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
8203 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
8204 (x * C EQ/NE y * C): New transformation.
8206 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
8208 * genmultilib (combination_space): Accept '+' in option names.
8210 2017-06-28 Martin Liska <mliska@suse.cz>
8213 * asan.c (instrument_derefs): Bail out inner references
8214 that are hard register variables.
8216 2017-06-28 Jakub Jelinek <jakub@redhat.com>
8219 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
8220 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
8222 2017-06-28 Richard Biener <rguenther@suse.de>
8224 * tree-vectorizer.h (vect_get_vec_defs): Remove.
8225 (vect_get_slp_defs): Adjust.
8226 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
8228 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
8230 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
8231 get_initial_defs_for_reduction instead of vect_get_vec_defs.
8232 (vectorizable_reduction): Adjust.
8233 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
8235 (vect_get_slp_defs): Likewise.
8236 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
8237 (vectorizable_bswap): Adjust.
8238 (vectorizable_call): Likewise.
8239 (vectorizable_conversion): Likewise.
8240 (vectorizable_assignment): Likewise.
8241 (vectorizable_shift): Likewise.
8242 (vectorizable_operation): Likewise.
8243 (vectorizable_store): Likewise.
8244 (vectorizable_condition): Likewise.
8245 (vectorizable_comparison): Likewise.
8247 2017-06-28 Michael Collison <michael.collison@arm.com>
8250 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
8252 (arm_gen_movmemqi): Removed unused variable 'i'.
8253 Convert 'for' loop into 'while' loop.
8254 (arm_expand_prologue): Remove last unnecessary set of insn.
8255 (thumb_pop): Remove unused variable 'pushed_words'.
8256 (thumb_exit): Remove last unnecessary set of regs_to_pop.
8258 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8260 * config/s390/predicates.md: Use s390_rel_address_ok_p.
8261 * config/s390/s390-protos.h: Add prototype of
8262 s390_rel_address_ok_p.
8263 * config/s390/s390.c (s390_got_symbol): New function.
8264 (s390_rel_address_ok_p): New function.
8265 (legitimize_pic_address): Use s390_rel_address_ok_p.
8266 (s390_load_got): Use s390_got_symbol.
8267 (s390_option_override): Issue error if
8268 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
8269 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
8271 * config/s390/s390.opt: New option mpic-data-is-text-relative.
8273 2017-06-27 Andrew Pinski <apinski@cavium.com>
8275 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
8276 (X * copysign (1.0, X)): New pattern.
8277 (X * copysign (1.0, -X)): New pattern.
8278 (copysign (-1.0, CST)): New pattern.
8280 2017-06-27 Joseph Myers <joseph@codesourcery.com>
8282 * genmultilib (combination_space): Remove variable.
8283 Validate reuse rules against regular expression for any sequence
8284 of multilib options in any order.
8286 2017-06-27 Michael Collison <michael.collison@arm.com>
8288 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
8289 call aarch64_split_simd_combine.
8290 * (aarch64_combine_internal<mode>): Delete pattern.
8291 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
8292 Allow register and subreg operands.
8294 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8296 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
8297 specific need, just fallback on defaults.
8298 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
8300 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8301 Olivier Hainque <hainque@adacore.com>
8303 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
8305 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
8306 targets. Pick a default if no particular attempt applied.
8307 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
8310 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8312 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
8313 (x86_64-wrs-vxworks7): Likewise.
8315 2017-06-27 Marek Polacek <polacek@redhat.com>
8318 * ubsan.c (instrument_null): Check get_base_address's result for null.
8320 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8322 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
8324 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8326 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
8327 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
8328 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
8329 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
8331 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
8332 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
8333 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
8334 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
8335 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
8336 BUILT_IN_FEUPDATEENV): New builtins.
8337 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
8338 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
8339 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
8340 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
8342 (builtin_structptr_types): Adjust size.
8343 * tree.c (builtin_structptr_types): Add four entries.
8345 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8346 Olivier Hainque <hainque@adacore.com>
8348 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
8349 (TLS_SYM): New local macro, forcing reference to __tls__ on
8350 link command lines for VxWorks 7 RTPs, triggering initialization
8352 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
8353 OS features TLS support, true for RTPs on VxWorks 7.
8354 * config/vxworks.c (vxworks_override_options): Setup emutls
8357 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8359 * predict.c (test_prediction_value_range): Use -1U instead of -1
8360 to avoid narrowing conversion warning.
8361 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
8362 to avoid narrowing conversion warning.
8363 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
8365 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
8367 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8369 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
8370 64bit configurations.
8371 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
8372 (SIZE_TYPE): Likewise.
8373 * config/vxworks.c (vxworks_emutls_var_fields): Use
8374 long_unsigned_type_node instead of unsigned_type_node as the offset
8375 field type, which is "pointer" mode in emutls.c.
8377 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8380 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
8383 * gimple-fold.c (replace_call_with_call_and_fold): Handle
8384 gimple_vuse copying separately from gimple_vdef copying.
8386 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8388 * value-prof.c (free_hist): Remove call to memset and the enclosing if
8391 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8392 Olivier Hainque <hainque@adacore.com>
8394 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
8395 for all vxworks7 targets.
8396 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
8397 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
8398 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
8399 variations for VX6/VX7 and 32/64bits later on in ...
8400 (VXWORKS_LIB_SPEC): Leverage new macros.
8401 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
8402 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
8404 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8406 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
8407 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
8409 2017-06-26 Carl Love <cel@us.ibm.com>
8411 * config/rs6000/rs6000-c.c: Add support for built-in functions
8412 vector bool char vec_reve (vector bool char);
8413 vector signed char vec_reve (vector signed char);
8414 vector unsigned char vec_reve (vector unsigned char);
8415 vector bool int vec_reve (vector bool int);
8416 vector signed int vec_reve (vector signed int);
8417 vector unsigned int vec_reve (vector unsigned int);
8418 vector bool long long vec_reve (vector bool long long);
8419 vector signed long long vec_reve (vector signed long long);
8420 vector unsigned long long vec_reve (vector unsigned long long);
8421 vector bool short vec_reve (vector bool short);
8422 vector signed short vec_reve (vector signed short);
8423 vector double vec_reve (vector double);
8424 vector float vec_reve (vector float);
8425 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
8426 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
8427 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
8428 (altivec_vreve): New pattern.
8429 * config/rs6000/altivec.h (vec_reve): New define.
8430 * doc/extend.texi (vec_rev): Update the built-in documentation file
8431 for the new built-in functions.
8433 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8435 PR tree-optimization/71815
8436 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
8438 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
8440 (slsr_process_phi): Likewise.
8441 (replace_uncond_cands_and_profitable_phis): Don't replace a
8442 multiply candidate with a stride of 1 (copy or cast).
8443 (phi_incr_cost): Call uses_consumed_by_stmt rather than
8445 (lowest_cost_path): Likewise.
8446 (total_savings): Likewise.
8448 2017-06-26 Richard Biener <rguenther@suse.de>
8451 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
8452 Use def_builtin_pure for all gather builtins.
8454 2017-06-26 Richard Biener <rguenther@suse.de>
8456 PR tree-optimization/81203
8457 * tree-tailcall.c (find_tail_calls): Do not move stmts into
8460 2017-06-26 Marek Polacek <polacek@redhat.com>
8463 * doc/invoke.texi: Document -Wmultistatement-macros.
8465 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
8467 * doc/sourcebuild.texi (ARM-specific attributes): Document new
8468 arm_neon_ok_no_float_abi effective target.
8470 2017-06-26 Richard Biener <rguenther@suse.de>
8472 PR tree-optimization/80928
8473 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
8474 (copy_bbs): Set BB_DUPLICATED flag early.
8475 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
8477 (execute_on_shrinking_pred): Likewise.
8478 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
8479 BB_DUPLICATED blocks.
8480 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
8481 iterate over all PHIs considering removal of *gsi.
8483 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
8485 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
8488 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8490 * config/rs6000/rs6000-string.c: (expand_block_clear,
8491 do_load_for_compare, select_block_compare_mode,
8492 compute_current_alignment, expand_block_compare,
8493 expand_strncmp_align_check, expand_strn_compare,
8494 expand_block_move, rs6000_output_load_multiple)
8495 Move functions related to string/block move/compare
8497 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
8498 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
8499 for this function which is now used in two files.
8500 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
8501 * config.gcc: Add rs6000-string.o to extra_objs for
8502 targets powerpc*-*-* and rs6000*-*-*.
8504 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
8507 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
8508 32-bit, since indexed is not valid for DImode.
8509 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
8510 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
8511 (define_peephole2 for Altivec d-form load): Add 32-bit support.
8512 (define_peephole2 for Altivec d-form store): Likewise.
8515 * multiple_target.c (create_dispatcher_calls): Only create the
8516 dispatcher call if the function is the default clone of a
8519 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
8522 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
8523 a call, force the call to not be a tail call.
8525 2017-06-23 Jeff Law <law@redhat.com>
8527 * doc/contrib.texi: Add entry for Steven Pemberton's work on
8530 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
8532 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
8533 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
8534 handling for early expansion of vector shifts (sl,sr,sra,rl).
8535 (builtin_function_type): Add vector shift right instructions
8536 to the unsigned argument list.
8538 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
8540 rtl-optimizatoin/79286
8541 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
8542 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
8543 trap. PIC register plus a const unspec without offset can never trap.
8545 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
8547 * tree.h (builtin_structptr_type): New type.
8548 (builtin_structptr_types): Declare new array.
8549 * tree.c (builtin_structptr_types): New array.
8550 (free_lang_data, build_common_tree_nodes): Use it.
8552 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
8555 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
8558 2017-06-22 Matt Turner <mattst88@gmail.com>
8560 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
8561 Lake models to skylake case. Assume skylake for unknown
8562 models with clflushopt.
8564 2017-06-22 Jeff Law <law@redhat.com>
8566 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
8567 frame sizes that do not satisfy aarch64_uimm12_shift.
8569 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
8571 * profile-count.h (apply_probability,
8572 apply_scale, probability_in): Fix checks for zero.
8574 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8576 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
8577 * doc/cppdiropts.texi (-I @var{dir}): Document it.
8579 2016-06-22 Richard Biener <rguenther@suse.de>
8581 * tree-vect-loop.c (vect_model_reduction_cost): Handle
8582 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
8583 REDUC_MAX_EXPR support.
8584 (vectorizable_reduction): Likewise.
8585 (vect_create_epilog_for_reduction): Likewise.
8587 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
8589 * match.pd (A / (1 << B) -> A >> B): New.
8590 * generic-match-head.c: Include optabs-tree.h.
8591 * gimple-match-head.c: Likewise.
8592 * optabs-tree.h (target_supports_op_p): New.
8593 * optabs-tree.c (target_supports_op_p): New.
8595 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8597 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
8598 $gcc_cv_ld --help output.
8599 (gcc_cv_ld_demangle): Likewise.
8600 (gcc_cv_ld_eh_frame_hdr): Likewise.
8601 (gcc_cv_ld_pie): Likewise.
8602 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
8603 (gcc_cv_ld_buildid): Likewise.
8604 (gcc_cv_ld_sysroot): Likewise.
8605 (ld_bndplt_support): Likewise.
8606 (ld_pushpopstate_support): Likewise.
8607 * configure: Regenerate.
8608 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
8610 2017-06-21 Jakub Jelinek <jakub@redhat.com>
8613 * config/i386/sse.md (round<mode>2): Renumber match_dup and
8614 operands indexes to avoid gap between operands and match_dups.
8616 2017-06-21 Andrew Pinski <apinski@cavium.com>
8618 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
8619 Increment Arith_shift and Arith_shift_reg by 1.
8620 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
8622 * config/aarch64/aarch64.c (thunderx_tunings): Enable
8623 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
8624 (aarch64_strip_extend): Add new argument and test for it.
8625 (aarch64_cheap_mult_shift_p): New function.
8626 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
8627 add a cost if it is true.
8628 Update calls to aarch64_strip_extend.
8629 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
8631 2017-06-21 Andrew Pinski <apinski@cavium.com>
8633 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
8635 (thunderxt88): Likewise.
8636 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
8637 (thunderx_prefetch_tune): New variable.
8638 (thunderx2t99_prefetch_tune): Update for the correct values.
8639 (thunderxt88_tunings): New variable.
8640 (thunderx_tunings): Use thunderx_prefetch_tune instead of
8641 generic_prefetch_tune.
8642 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
8644 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8646 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
8647 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
8648 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
8649 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
8650 (aarch64_atomic_cas<mode>, GPI): Likewise.
8652 2017-06-21 Martin Liska <mliska@suse.cz>
8654 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
8655 statements on cold and hot labels.
8656 * predict.c (tree_estimate_probability_bb): Remove the
8657 prediction from this place.
8659 2017-06-21 Martin Liska <mliska@suse.cz>
8661 PR tree-optimization/79489
8662 * gimplify.c (maybe_add_early_return_predict_stmt): New
8664 (gimplify_return_expr): Call the function.
8665 * predict.c (tree_estimate_probability_bb): Remove handling
8667 * predict.def: Update comment about early return predictor.
8668 * gimple-predict.h (is_gimple_predict): New function.
8669 * predict.def: Change default value of early return to 66.
8670 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
8672 * passes.def: Put pass_strip_predict_hints to the beginning of
8675 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
8677 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
8678 FUNCTION_DECL declarations.
8679 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
8681 (dwaf2out_decl): Likewise.
8682 * godump.c (go_early_global_decl): Skip call to the real debug hook
8683 for FUNCTION_DECL declarations.
8684 * passes.c (rest_of_decl_compilation): Skip call to the
8685 early_global_decl debug hook for FUNCTION_DECL declarations, unless
8686 -fdump-go-spec is passed.
8688 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
8690 * config/i386/i386.c (struct builtin_isa): New field pure_p.
8691 Reorder for compactness.
8692 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
8693 (def_builtin_pure, def_builtin_pure2): New functions.
8694 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
8696 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
8698 * match.pd (nop_convert): New predicate.
8699 ((A +- CST1) +- CST2): Allow some NOP conversions.
8701 2017-06-21 Jakub Jelinek <jakub@redhat.com>
8704 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
8705 with ctors/dtors if GOVD_SHARED is set.
8707 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8709 * config/aarch64/aarch64.md (movti_aarch64):
8710 Emit mov rather than orr.
8711 (movtf_aarch64): Likewise.
8712 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
8713 Emit mov rather than orr.
8715 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8717 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
8718 Swap alternatives, make integer dup more expensive.
8720 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8722 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8723 Return true for non-tls symbols.
8725 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
8727 * config/aarch64/aarch64-cores.def (cortex-a55): New.
8728 (cortex-a75): Likewise.
8729 (cortex-a75.cortex-a55): Likewise.
8730 * config/aarch64/aarch64-tune.md: Regenerate.
8731 * doc/invoke.texi (-mtune): Document new values for -mtune.
8733 2017-06-21 Tom de Vries <tom@codesourcery.com>
8735 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
8737 (Effective-Target Keywords, Other attributes): Suggest using
8738 dg-add-options stack_size feature to get stack limit in stack_size
8739 effective target documentation.
8741 2017-06-21 Julian Brown <julian@codesourcery.com>
8742 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8744 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
8745 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
8746 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
8748 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
8749 attribute type list for neon_multiply.
8750 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
8751 attribute type list for neon_multiply.
8752 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
8753 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
8754 attribute type list for neon_multiply.
8755 * config/arm/types.md (crypto_pmull): Add.
8756 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
8757 attribute type list.
8759 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
8761 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
8764 2017-06-20 Jakub Jelinek <jakub@redhat.com>
8766 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
8767 to make sure not to dereference a NULL cost_classes_ptr pointer.
8769 2017-06-20 Carl Love <cel@us.ibm.com>
8771 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8772 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8773 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8774 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8775 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8776 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8777 VMULOSW): New enum "unspec" values.
8778 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8779 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8780 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8781 altivec_vmulosw): New patterns.
8782 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8783 VMULOSW): Add definitions.
8785 2017-06-20 Julia Koval <julia.koval@intel.com>
8787 * config/i386/i386.c: Fix rounding expand for new pattern.
8788 * config/i386/subst.md: Fix pattern (parallel -> unspec).
8790 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8792 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
8793 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
8795 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8797 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
8800 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8802 * config/aarch64/aarch64-cores.def: Rearrange to sort by
8803 architecture, then by implementer ID.
8804 * config/aarch64/aarch64-tune.md: Regenerate.
8806 2017-06-20 Richard Biener <rguenther@suse.de>
8809 * fold-const.c (split_tree): Fold to type before negating.
8811 2017-06-20 David Malcolm <dmalcolm@redhat.com>
8813 * diagnostic-show-locus.c
8814 (selftest::test_fixit_deletion_affecting_newline): New function.
8815 (selftest::diagnostic_show_locus_c_tests): Call it.
8817 2017-06-20 Andreas Schwab <schwab@suse.de>
8820 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
8823 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8825 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
8826 __ARM_FEATURE_COPROC according to support.
8828 2017-06-20 Jakub Jelinek <jakub@redhat.com>
8830 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
8831 Rewritten to avoid overflow for > 32-bit pointers.
8834 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
8835 by removing enum keyword.
8836 (ubsan_type_descriptor): Likewise. Formatting fix.
8839 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
8840 splitter): Require TARGET_SSE2 in the condition.
8842 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
8845 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
8846 for doing vector set of SFmode on ISA 3.0.
8847 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
8848 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
8850 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
8851 SFmode value into a V4SF variable that was extracted from another
8852 V4SF variable without converting the element to double precision
8853 and back to single precision vector format.
8854 (vsx_insert_extract_v4sf_p9_2): Likewise.
8856 2017-06-19 Jakub Jelinek <jakub@redhat.com>
8858 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
8859 in UWHI to avoid undefined overflow.
8862 * ubsan.h (enum ubsan_encode_value_phase): New.
8863 (ubsan_encode_value): Change second argument to
8864 enum ubsan_encode_value_phase with default value of
8865 UBSAN_ENCODE_VALUE_GENERIC.
8866 * ubsan.c (ubsan_encode_value): Change second argument to
8867 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
8868 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
8869 create_tmp_var_raw instead of create_tmp_var and use a
8871 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
8872 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
8873 ubsan_encode_value callers.
8876 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
8877 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
8878 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
8880 2017-06-19 Richard Biener <rguenther@suse.de>
8883 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
8884 estimates if we changed anything.
8886 2017-06-19 Richard Biener <rguenther@suse.de>
8888 PR tree-optimization/80887
8889 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
8890 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
8891 simplified lookups, then reset mprts_hook.
8892 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
8894 (try_to_simplify): Likewise.
8896 2017-06-19 Martin Liska <mliska@suse.cz>
8899 * gimplify.c (gimplify_switch_expr):
8900 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
8902 2017-06-19 Martin Liska <mliska@suse.cz>
8904 * doc/install.texi: Document that PGO runs in 4 stages.
8906 2017-06-19 Martin Liska <mliska@suse.cz>
8909 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
8910 to dispatcher function name.
8911 * multiple_target.c (replace_function_decl): New function.
8912 (create_dispatcher_calls): Redirect both edges and references.
8914 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
8916 * profile-count.c (profile_count::dump): Dump quality.
8917 (profile_count::differs_from_p): Update for unsigned val.
8918 * profile-count.h (profile_count_quality): New enum.
8919 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
8921 2017-06-19 Richard Biener <rguenther@suse.de>
8923 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
8924 struct function as arg.
8925 (estimate_numbers_of_iterations): Export overload with loop arg.
8926 (free_numbers_of_iterations_estimates_loop): Use an overload of
8927 free_numbers_of_iterations_estimates instead.
8928 * tree-cfg.c (remove_bb): Adjust.
8929 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
8930 * tree-parloops.c (gen_parallel_loop): Likewise.
8931 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
8933 (tree_unroll_loops_completely): Likewise.
8934 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
8935 Use an overload instead and export.
8936 (estimated_loop_iterations): Adjust.
8937 (max_loop_iterations): Likewise.
8938 (likely_max_loop_iterations): Likewise.
8939 (estimate_numbers_of_iterations): Take struct function as arg
8941 (loop_exits_before_overflow): Adjust.
8942 (free_numbers_of_iterations_estimates_loop): Use an overload.
8943 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
8944 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
8946 2017-06-19 Richard Biener <rguenther@suse.de>
8949 * ipa-prop.c (find_constructor_constant_at_offset): Handle
8950 RANGE_EXPR conservatively.
8952 2017-06-16 Carl Love <cel@us.ibm.com>
8954 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8955 definitions for vec_float, vec_float2, vec_floato,
8956 vec_floate built-ins.
8957 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
8958 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
8960 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
8961 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
8962 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
8963 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
8964 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
8965 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
8966 vec_floato): Add builtin defines.
8967 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
8968 Update the built-in documentation file for the new built-in
8971 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8973 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
8974 (mthumb): Mark as the negative of -marm.
8976 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8978 * doc/invoke.texi (ARM Options, -mcpu): Document supported
8980 (ARM Options, -mtune): Document that this accepts the same
8981 extension options as -mcpu.
8982 (ARM Options, -mfpu): Document addition of -mfpu=auto.
8984 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8986 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
8987 permitted extensions.
8989 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8991 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
8992 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
8993 (armv8-m.main): Add option +nodsp.
8994 * config/arm/arm-cpu-cdata.h: Regenerated.
8996 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8998 * config/arm/t-fuchsia: New file.
8999 * config.gcc (arm*-*-fuchsia*): Use it.
9001 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9003 * config/arm/t-symbian: Rewrite for new option infrastructure.
9005 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9007 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
9008 (MULTILIB_REQUIRED): Likewise.
9010 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9012 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
9013 (MULTILIB_RESUE): Likewise.
9014 (MULTILIB_MATCHES): Likewise.
9015 (MULTLIB_REQUIRED): Likewise.
9017 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9019 * config/arm/t-rtems: Rewrite for new option framework.
9021 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9023 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
9024 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
9025 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
9026 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
9027 * config/arm/t-multilib: ... here.
9028 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
9029 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
9030 armv7-a and armv8*-a when A-profile libraries have not been built.
9031 * config/arm/t-rmprofile: Rewrite.
9033 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9035 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
9036 with a backslash. Remove the backslash after substituting unescaped
9038 * doc/fragments.texi (MULTILIB_REUSE): Document it.
9040 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9042 * config.gcc: (arm*-*-*): When building a-profile libraries, force
9043 the driver to pass through the default setting of -mfloat-abi.
9044 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
9046 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
9047 (all_feat_combs): New rule.
9048 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
9050 * config/arm/t-aprofile: Rewrite.
9052 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9054 * config/arm/arm.h (FPUTYPE_AUTO): Define.
9055 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
9056 fpu is not specified by the user/command-line.
9057 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
9058 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
9059 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
9060 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
9061 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
9062 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
9064 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9066 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
9067 * config/arm/t-arm-elf: Rewritten.
9069 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9071 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
9072 have some floating-point instructions.
9073 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
9074 (TARGET_MAYBE_HARD_FLOAT): New macro.
9075 * config/arm/arm-builtins.c (arm_init_builtins): Use
9076 TARGET_MAYBE_HARD_FLOAT.
9077 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
9079 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9081 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
9082 (configargs.h): Include it.
9083 (arm_print_hint_for_fpu_option): New function.
9084 (arm_parse_fpu_option): New function.
9085 (candidate_extension): New class.
9086 (arm_canon_for_multilib): New function.
9087 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
9088 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
9089 (ARCH_CANONICAL_SPECS): New macro.
9090 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
9092 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9094 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
9095 are set after handling multilib fragments. Set target_cpu_default2
9098 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9100 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
9102 (arm*-*-*): Set target_cpu_default2 to a quoted string.
9103 * config/arm/parsecpu.awk (check_cpu): Validate any extension
9105 (check_arch): Likewise.
9106 * config/arm/arm.c (arm_configure_build_target): Handle
9107 TARGET_CPU_DEFAULT being a string constant. Scan any feature
9108 options in the default.
9110 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9112 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
9113 when an option is an alias of another.
9114 * config/arm/parsecpu.awk (optalias): New parser token.
9115 (gen_comm_data): Mark non-alias options as such. Emit entries
9116 for extension aliases.
9117 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
9118 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
9119 (armv6kz, armv6zk, armv6t2): Likewise.
9120 (armv7): Make vfpv3-d16 an alias.
9121 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
9123 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
9124 Sort in canonical order.
9125 (armv8-a): Sort in canonical order.
9126 (armv8.1-a, armv8.2-a): Likewise.
9127 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
9129 (cortex-a9): Sort in canonical order.
9130 * config/arm/arm.c (selftests.h): Include it.
9131 (arm_test_cpu_arch_data): New function.
9132 (arm_run_self_tests): New function.
9133 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
9134 (targetm): Move declaration to the end of the file.
9135 * arm-cpu-cdata.h: Regenerated.
9137 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9139 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
9140 call to target_mode_check describing the type of option passed.
9141 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
9142 (arm_target_thumb_only): Use arm_parse_arch_option_name or
9143 arm_parse_cpu_option_name to match parameters against list of
9145 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
9146 arm_arch_core_flags data structure.
9147 * config/arm/arm-cpu_cdata.h: Regenerated.
9149 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9151 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
9153 (arm_print_hint_for_cpu_option): Likewise.
9154 (arm_print_hint_for_arch_option): Likewise.
9155 (arm_parse_cpu_option_name): Likewise.
9156 (arm_parse_arch_option_name): Likewise.
9157 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
9158 of entries in the all_fpus list.
9159 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
9160 (arm_parse_cpu_option_name): Declare.
9161 (arm_parse_arch_option_name): Declare.
9162 (arm_parse_option_features): Declare.
9163 (arm_intialize_isa): Declare.
9164 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
9166 (gen_comm_data): ... here. Make definitions non-static.
9167 * config/arm/arm-cpu-data.h: Regenerated.
9168 * config/arm/arm-cpu-cdata.h: Regenerated.
9170 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9172 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
9173 (cpu_arch_extension): New structure.
9174 (cpu_arch_option, arch_option, cpu_option): New structures.
9175 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
9177 (gen_data): Generate new format data tables.
9178 * config/arm/arm.c (cpu_tune): New structure.
9179 (cpu_option, processors): Delete.
9180 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
9181 (arm_print_hint_for_cpu_option): ... this and ...
9182 (arm_print_hint_for_arch_option): ... this.
9183 (arm_parse_arch_cpu_name): Delete. Replace with ...
9184 (arm_parse_cpu_option_name): ... this and ...
9185 (arm_parse_arch_option_name): ... this.
9186 (arm_unrecognized_feature): Change type of target parameter to
9188 (arm_parse_arch_cpu_features): Delete. Replace with ...
9189 (arm_parse_option_features): ... this.
9190 (arm_configure_build_target): Rework to use new configuration data
9192 (arm_print_tune_info): Rework for new configuration data tables.
9193 * config/arm/arm-cpu-data.h: Regenerated.
9194 * config/arm/arm-cpu.h: Regenerated.
9196 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9198 * Makefile.in (OBJS): Move sbitmap.o from here ...
9199 (OBJS-libcommon): ... to here.
9201 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9203 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
9204 (ISA_ALL_CRYPTO): New macro.
9205 (ISA_ALL_SIMD): New macro
9206 (ISA_ALL_FP): New macro.
9207 * config/arm/arm.c (fpu_bitlist): Update initializer.
9208 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
9210 (arm9e): Add fpu. Add option for nofp
9211 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
9212 (arm926ej-s, arm1026ej-s): Likewise.
9213 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
9214 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
9215 neon-fp16, neon-vfpv4, nofp and nosimd.
9216 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
9217 (cortex-a8): Add fpu. Add option for nofp.
9218 (cortex-a9): Add fpu. Add options for nosimd and nofp.
9219 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
9220 (cortex-r4f): Add fpu.
9221 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
9222 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
9224 (cortex-r8): Likewise.
9225 (cortex-m4): Add fpu. Add option for nofp.
9226 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
9227 (cortex-a17.cortex-a7): Likewise.
9228 (cortex-a32): Add fpu. Add options for crypto and nofp.
9229 (cortex-a35, cortex-a53): Likewise.
9230 (cortex-a57): Add fpu. Add option for crypto.
9231 (cortex-a72, cortex-a73): Likewise.
9232 (exynos-m1): Likewise.
9233 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
9234 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
9235 (cortex-m33): Add fpu. Add option for nofp.
9236 * config/arm/arm-cpu-cdata.h: Regenerated
9237 * config/arm/arm-cpu-data.h: Regenerated.
9239 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9241 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
9242 (armv5te, armv5tej): Likewise.
9243 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
9244 (armv7): Add options fp and vfpv3-d16.
9245 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
9246 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
9248 (armv7ve): Likewise.
9249 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
9250 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
9251 (armv8-a): Add nocrypto option.
9252 (armv8.1-a, armv8.2-a): Likewise.
9253 (armv8-m.main): add options fp, fp.dp and nofp.
9255 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9257 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
9259 (armv8-a+crc): Delete.
9260 (armv8.1-a): Add options simd, crypto and nofp.
9261 (armv8.2-a): Add options fp16, simd, crypto and nofp.
9262 (armv8.2-a+fp16): Delete.
9263 (armv8-m.main): Add option dsp.
9264 (armv8-m.main+dsp): Delete.
9265 (cortex-a8): Add fpu. Add nofp option.
9266 (cortex-a9): Add fpu. Add nofp and nosimd options.
9267 * config/arm/parsecpu.awk (gen_data): Generate option tables and
9268 link to main cpu and architecture data structures.
9269 (gen_comm_data): Only put isa attributes from the main architecture
9271 (option): New statement for architecture and CPU entries.
9272 * arm.c (struct cpu_option): New structure.
9273 (struct processors): Add entry for options.
9274 (arm_unrecognized_feature): New function.
9275 (arm_parse_arch_cpu_name): Ignore any characters after the first
9277 (arm_parse_arch_cpu_feature): New function.
9278 (arm_configure_build_target): Separate out any CPU and architecture
9279 features and parse separately. Don't error out if -mfpu=auto is
9280 used with only an architecture string.
9281 (arm_print_asm_arch_directives): New function.
9282 (arm_file_start): Call it.
9283 * config/arm/arm-cpu-cdata.h: Regenerated.
9284 * config/arm/arm-cpu-data.h: Likewise.
9285 * config/arm/arm-tables.opt: Likewise.
9287 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9289 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
9290 assembler when it is not -mfpu=auto.
9292 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9294 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
9295 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
9296 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
9297 (ASM_CPU_SPEC): Rewrite.
9298 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
9299 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
9300 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
9301 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
9302 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
9303 copied string is NUL-terminated. Also strip any characters prefixed
9305 (arm_rewrite_selected_arch): New function.
9306 (arm_rewrite_march): New function.
9308 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9310 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
9311 (x_arm_cpu_string, x_arm_tune_string): Likewise.
9312 (march, mcpu, mtune): Convert to string-based options.
9313 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
9314 (arm_parse_arch_cpu_name): New function.
9315 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
9316 identify selected architecture or CPU.
9317 (arm_option_save): New function.
9318 (TARGET_OPTION_SAVE): Redefine.
9319 (arm_option_restore): Restore string options.
9320 (arm_option_print): Print string options.
9322 2017-06-16 Martin Sebor <msebor@redhat.com>
9324 PR tree-optimization/80933
9325 PR tree-optimization/80934
9326 * builtins.c (fold_builtin_3): Do not handle bcmp here.
9327 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
9328 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
9329 (gimple_fold_builtin): Call them.
9331 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9333 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
9334 as unlikely; update profile.
9336 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9338 * predict.c (force_edge_cold): Handle declaring edges impossible
9341 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9343 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
9345 (try_unroll_loop_completely): Fix reporting.
9347 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9349 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
9351 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
9354 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
9355 if given a non-constant argument for an intrinsic which requires a
9358 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9360 * profile.c (compare_freqs): New function.
9361 (branch_prob): Sort edge list.
9362 (find_spanning_tree): Assume that the list is priority sorted.
9364 2017-06-16 Richard Biener <rguenther@suse.de>
9366 PR tree-optimization/81090
9367 * passes.def (pass_record_bounds): Remove.
9368 * tree-pass.h (make_pass_record_bounds): Likewise.
9369 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
9370 make_pass_record_bounds): Likewise.
9371 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
9372 not free niter estimates at the beginning but at the end.
9373 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
9375 2017-06-16 Richard Biener <rguenther@suse.de>
9377 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
9378 initializer to workaround ICE in host GCC 4.8.
9380 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9382 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
9384 (clone_inlined_nodes): Update.
9386 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9388 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
9389 prefetch settings, and enable prefetching by default at -O3.
9391 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9393 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9394 Set flag_prefetch_loop_arrays according to tuning data.
9396 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9398 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
9400 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
9401 [Unrelated to main purpose of the patch] Place the pointer field last
9402 to enable type checking errors when tune structure are wrongly merged.
9403 * config/aarch64/aarch64.c (generic_prefetch_tune,)
9404 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
9405 (thunderx2t99_prefetch_tune): New tune constants.
9406 (tune_params *_tunings): Update all tunings (no functional change).
9407 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
9408 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
9409 from tunings structures.
9411 2017-06-16 Jakub Jelinek <jakub@redhat.com>
9414 * ubsan.c (instrument_null): Add T argument, use it instead
9415 of computing it based on IS_LHS.
9416 (instrument_object_size): Likewise.
9417 (pass_ubsan::execute): Adjust instrument_null and
9418 instrument_object_size callers to pass gimple_get_lhs or
9419 gimple_assign_rhs1 result to it. Use instrument_null instead of
9420 calling get_base_address and instrument_mem_ref. Handle
9421 aggregate call arguments for object-size sanitization.
9423 2017-06-16 Yury Gribov <tetra2005@gmail.com>
9425 PR tree-optimization/81089
9426 * tree-vrp.c (is_masked_range_test): Validate operands of
9429 2017-06-15 Martin Sebor <msebor@redhat.com>
9432 * dumpfile.c (dump_register): Avoid calling memset to initialize
9433 a class with a default ctor.
9434 * gcc.c (struct compiler): Remove const qualification.
9435 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
9436 * hash-table.h: Ditto.
9437 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
9439 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
9440 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
9442 * params.h (struct param_info): Make struct members non-const.
9443 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
9444 with default initialization.
9445 * vec.h (vec_copy_construct, vec_default_construct): New helper
9447 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
9448 with vec_copy_construct.
9449 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
9450 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
9451 * doc/invoke.texi (-Wclass-memaccess): Document.
9453 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9455 * emit-rtl.h (is_leaf): Update comment about local
9458 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9461 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
9462 for a variable to have a section before checking if the section has a
9464 Set section to.persistent if persistent attribute is set.
9465 Warn if .persistent attribute is used on an automatic variable.
9467 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
9469 PR rtl-optimization/80474
9470 * reorg.c (update_block): Do not ignore instructions in a delay slot.
9472 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
9474 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
9477 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
9479 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
9480 (casesi): Emit bounds checking as RTL.
9481 (casesi_internal_mips16_<mode>): Remove bounds checking.
9483 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
9485 * config/xtensa/xtensa.c (xtensa_option_override): Append
9486 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
9487 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
9488 xtensa_doloop_hooks): Define unconditionally.
9489 (xtensa_reorg_loops): Only call reorg_loops in the presence of
9491 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
9492 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
9493 for it in xtensa_option_override.
9494 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
9495 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
9497 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
9499 * doc/cppopts.texi: Document '-' special value to -MF.
9501 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
9503 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
9504 (cortex_a53_fconst): Likewise.
9505 (cortex_a53_fpmul): Likewise.
9506 (cortex_a53_f_load_64): Likewise.
9507 (cortex_a53_f_load_many): Likewise.
9508 (cortex_a53_advsimd_alu): Likewise.
9509 (cortex_a53_advsimd_alu_q): Likewise.
9510 (cortex_a53_advsimd_mul): Likewise.
9511 (cortex_a53_advsimd_mul_q): Likewise.
9512 (fpmac bypass): Add new bypass for fpmac-fpmac case.
9513 Add missing fmul, r2f_cvt and fconst cases.
9515 2017-06-14 Richard Biener <rguenther@suse.de>
9518 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
9520 (fold_binary_loc): When associating do not treat pre-existing
9521 TREE_OVERFLOW on literal constants as a reason to allow
9522 TREE_OVERFLOW on associated literal constants.
9524 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9526 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
9527 (MASK_FEATURES): New macro.
9528 * config/sparc/sparc.c (sparc_option_override): Remove the special
9529 handling of -mfpu and generalize it to all MASK_FEATURES switches.
9531 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9533 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
9534 a division of 0 if non-call exceptions are enabled.
9536 2017-06-14 Andrew Pinski <apinski@cavium.com>
9537 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9540 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
9541 Improve vector initialization code gen for only variable case.
9543 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9545 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
9547 2017-06-14 Richard Biener <rguenther@suse.de>
9549 PR tree-optimization/81083
9550 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
9553 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9555 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
9556 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
9557 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
9558 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
9559 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
9560 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
9562 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9564 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
9565 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
9567 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9569 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
9571 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9573 * config/rs6000/t-rtems: Don't handle SPE.
9575 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9577 * config/rs6000/t-linux: Don't handle SPE.
9579 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9581 * config/rs6000/eabispe.h: Delete file.
9583 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9585 * config/rs6000/t-spe: Delete file.
9587 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9589 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
9590 (rs6000_legitimate_offset_address_p): Return false for anything in
9591 V2SImode or V2SFmode.
9593 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9595 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
9596 except V2SF and V2SI. Rearrange the vector modes, and add comments.
9597 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
9599 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
9600 (rs6000_legitimate_offset_address_p): Ditto.
9601 (rs6000_emit_move): Ditto.
9602 (rs6000_init_builtins): Remove V4HI_type_node.
9604 2017-06-13 Martin Liska <mliska@suse.cz>
9607 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
9608 (gate_asan): Likewise.
9609 * asan.h (asan_no_sanitize_address_p): Remove the function.
9610 (sanitize_flags_p): New function.
9611 * builtins.def: Fix coding style.
9612 * common.opt: Use renamed enum value.
9613 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
9614 * doc/extend.texi: Document no_sanitize attribute.
9615 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
9616 to SANITIZE_UNDEFINED_NONDEFAULT.
9617 * gcc.c (sanitize_spec_function): Use the renamed enum value.
9618 * gimple-fold.c (optimize_atomic_compare_exchange_p):
9619 Use sanitize_flags_p.
9620 * gimplify.c (gimplify_function_tree): Likewise.
9621 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
9622 * opts.c (parse_no_sanitize_attribute): New function.
9623 (common_handle_option): Use renamed enum value.
9624 * opts.h (parse_no_sanitize_attribute): Declare.
9625 * tree.c (sanitize_flags_p): New function.
9626 * tree.h: Declared here.
9627 * tsan.c: Use sanitize_flags_p.
9628 * ubsan.c (ubsan_expand_null_ifn): Likewise.
9629 (instrument_mem_ref): Likewise.
9630 (instrument_bool_enum_load): Likewise.
9631 (do_ubsan_in_current_function): Remove the function.
9632 (pass_ubsan::execute): Use sanitize_flags_p.
9633 * ubsan.h: Remove do_ubsan_in_current_function
9634 * tree-cfg.c (print_no_sanitize_attr_value): New function.
9635 (dump_function_to_file): Use it here.
9637 2017-06-13 Martin Jambor <mjambor@suse.cz>
9639 PR tree-optimization/80803
9640 PR tree-optimization/81063
9641 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
9642 (propagate_subaccesses_across_link): Enqueue subtree whenever
9643 necessary instead of relying on the caller.
9645 2017-06-13 Martin Jambor <mjambor@suse.cz>
9647 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
9648 that have a first_link.
9649 (sort_and_splice_var_accesses): Do not check first_link before
9651 (subtree_mark_written_and_enqueue): Likewise.
9652 (propagate_all_subaccesses): Likewise and do not stop at first
9653 parent with a first_link.
9655 2017-06-13 Martin Jambor <mjambor@suse.cz>
9657 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
9660 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9662 * match.pd: New pattern.
9664 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9666 * tree-vrp.c (is_masked_range_test): New function.
9667 (register_edge_assert_for): Determine ranges for
9670 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9672 PR tree-optimization/67328
9673 * fold-const.c (maskable_range_p): New function.
9674 (build_range_check): Generate bittests if possible.
9676 2017-06-13 Martin Liska <mliska@suse.cz>
9678 * gimple-pretty-print.c (dump_probability): Add new argument.
9679 (dump_edge_probability): Dump both probability and count.
9680 (dump_gimple_label): Likewise.
9681 (dump_gimple_bb_header): Likewise.
9683 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
9686 * config/avr/avr-devices.c: Fix indentation.
9687 * config/avr/gen-avr-mmcu-specs.c: Dito.
9689 2017-06-13 Richard Biener <rguenther@suse.de>
9691 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
9692 instead get vector type from stmt_info.
9693 (vectorizable_reduction): Adjust. Remove dead code.
9695 2017-06-13 Richard Biener <rguenther@suse.de>
9698 * fold-const.c (extract_muldiv_1): Remove bogus distribution
9699 case of C * (x * C2 + C3).
9700 (fold_addr_of_array_ref_difference): Properly fold index difference.
9702 2017-06-12 David S. Miller <davem@davemloft.net>
9705 * config/sparc/sparc.md (return expander): Emit frame blockage if
9706 function uses alloca.
9708 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9710 * combine.c (make_field_assignment): Check len rather than the mode
9711 precision when calling force_to_mode.
9713 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
9715 Support multilibs and devices that see flash in RAM address range.
9718 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
9719 (avr_mcu_t) <flash_pm_offset>: New field.
9720 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
9721 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
9722 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
9723 (AVR_TINY_PM_OFFSET): Remove macro.
9724 * config/avr/avr.opt (-mshort-calls): New option.
9725 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9726 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
9727 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
9728 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
9729 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
9730 instead of avr_arch->have_jmp_call.
9731 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
9732 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
9733 avr_arch->flash_pm_offset to define.
9734 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
9735 new field flash_pm_offset. Add entry for avrxmega3.
9736 (avr_texinfo): Add entry for avrxmega3.
9737 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
9738 attiny212, attiny214,
9739 attiny412, attiny414, attiny416, attiny417,
9740 attiny814, attiny816, attiny817,
9741 attiny1614, attiny1616, attiny1617,
9742 attiny3214, attiny3216, attiny3217.
9743 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
9744 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
9745 (avr_print_operand_address) [AVR_TINY]: Same.
9746 (avr_asm_init_sections) <readonly_data_section>: Only patch
9747 callback if avr_arch->flash_pm_offset = 0.
9748 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
9749 for rodata if avr_arch->flash_pm_offset != 0.
9750 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
9751 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
9752 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
9753 (m_options): Append opt_rcall.
9754 (m_dirnames): Append dir_rcall.
9755 * config/avr/t-multilib: Regenerate.
9757 * configure.ac [target=avr]: Check whether avrxmega3 default
9758 linker description file works as needed.
9759 * configure: Regenerate.
9760 * doc/avr-mmcu.texi: Regenerate.
9761 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
9762 <__AVR_ARCH__>: Document avrxmega3 and 103.
9763 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
9764 <__AVR_SHORT_CALLS__>: Document it.
9765 <__AVR_PM_BASE_ADDRESS__>: Document it.
9766 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
9767 (AVR Variable Attributes) <progmem>: Document this is
9768 not needed for avrxmega3.
9769 (AVR Named Address Spaces) <__flash>: Dito.
9771 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
9773 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
9775 2017-06-12 Doug Rupp <rupp@adacore.com>
9777 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
9778 Append vxworks-stdint.h to the tm_file list.
9779 * config/vxworks-stdint.h: New file.
9781 2017-06-12 Martin Liska <mliska@suse.cz>
9783 PR tree-optimization/81041
9784 * tree-profile.c (gimple_gen_ic_func_profiler):
9785 Create an extra BB in profile-generate
9786 (gimple_gen_time_profiler): Likewise.
9788 2017-06-12 Jakub Jelinek <jakub@redhat.com>
9790 PR tree-optimization/81003
9791 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
9792 (update_range_test): Use it instead of force_gimple_operand_gsi.
9794 2017-06-12 Richard Biener <rguenther@suse.de>
9796 PR tree-optimization/81053
9797 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
9798 with backedge value not defined in loop. Simplify def stmt
9801 2017-06-11 Tom de Vries <tom@codesourcery.com>
9804 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
9806 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
9807 nvptx_cannot_force_const_mem.
9809 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9811 * opts.c (finish_options): Move test for flag_split_stack after
9812 it has been initialized.
9814 2017-06-11 Jason Merrill <jason@redhat.com>
9816 * tree.h (id_equal): New.
9817 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
9818 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
9819 instead of strcmp of IDENTIFIER_POINTER.
9821 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9823 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
9824 (mark_all_inlined_calls_cdtor): Fix formating.
9825 (inline_transform): Rescale profile before inlining.
9827 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9829 * cgraph.h (cgraph_edge::clone): Update prototype.
9830 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
9831 (cgraph_node::create_clone): Update.
9832 (cgraph_node::create_version_clone): Update.
9833 * tree-inline.c (copy_bb): Update.
9834 (expand_call_inline): Update.
9836 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
9838 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
9839 factored out from ...
9840 (rs6000_emit_prologue): ... here.
9842 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
9844 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
9845 factored out from ...
9846 (rs6000_emit_prologue): ... here.
9848 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9850 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
9852 (handle_missing_profiles): Fix computation of tp_first_run.
9853 (counts_to_freqs): Do not touch freqs when count is 0.
9855 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9857 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
9860 2017-06-10 Tom de Vries <tom@codesourcery.com>
9862 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
9863 attributes): Document signal effective target.
9865 2017-06-10 Tom de Vries <tom@codesourcery.com>
9867 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
9868 Document effective target stack_size.
9870 2017-06-09 David Malcolm <dmalcolm@redhat.com>
9872 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
9873 to the edit_context if they can be auto-applied.
9875 2017-06-9 Ian Lance Taylor <iant@golang.org>
9877 * opts.c (finish_options): If -fsplit-stack, disable implicit
9878 -forder-blocks-and-partition.
9879 * doc/invoke.texi (Optimize Options): Document that when using
9880 -fsplit-stack -forder-blocks-and-partition is not implicitly
9883 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9885 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
9886 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
9887 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
9888 * builtins.def (abort, trap, unreachable): Declare cold.
9889 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
9890 * tree-core.h (ECF_COLD): New.
9891 * tree.c (set_call_expr_flags): Handle ECF_COLD.
9892 (build_common_builtin_nodes): Mark unreachable and abort as cold.
9894 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9896 * predict.c (unlikely_executed_stmt_p): Cleanup.
9898 2017-06-09 Richard Biener <rguenther@suse.de>
9900 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
9901 model if the ref is always written to.
9903 2017-06-09 Tamar Christina <tamar.christina@arm.com>
9905 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
9907 2017-06-09 Tamar Christina <tamar.christina@arm.com>
9909 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
9912 2017-06-09 Tom de Vries <tom@codesourcery.com>
9915 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
9916 "target cannot support label values" when encountering LABEL_REF.
9918 2017-06-09 Martin Liska <mliska@suse.cz>
9920 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
9921 (gimple_gen_ic_func_profiler): Emit direct comparison
9922 of __gcov_indirect_call_callee with NULL.
9923 (gimple_gen_time_profiler): Change probability from
9924 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
9926 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9928 * profile.c (edge_gcov_counts): Turn to pointer.
9929 (compute_branch_probabilities, compute_branch_probabilities): Update.
9930 (branch_prob): Do not clear edge_gcov_count.
9931 * profile.h (edge_gcov_counts): Turn to pointer.
9932 (edge_gcov_count): Update.
9934 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9936 * gimple.h (gimple_check_failed): Mark cold.
9938 2017-06-09 Richard Biener <rguenther@suse.de>
9940 PR tree-optimization/66623
9941 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
9942 refactor check_reduction into two parts, properly computing
9943 whether we have to check reduction validity for outer loop
9946 2017-06-09 Richard Biener <rguenther@suse.de>
9948 PR tree-optimization/79483
9949 * graphite-scop-detection.c (order): New global.
9950 (get_order): Compute bb to order mapping that satisfies code
9951 generation constraints.
9952 (cmp_pbbs): New helper.
9953 (build_scops): Start domwalk at entry block, sort generated
9956 2017-06-09 Richard Biener <rguenther@suse.de>
9959 * ipa-polymorphic-call.c
9960 (ipa_polymorphic_call_context::restrict_to_inner_class):
9961 Skip FIELD_DECLs with error_mark_node type.
9962 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
9965 2017-06-09 Martin Liska <mliska@suse.cz>
9967 * predict.c (struct branch_predictor): New struct.
9968 (test_prediction_value_range): New test.
9969 (predict_c_tests): New function.
9970 * selftest-run-tests.c (selftest::run_tests): Run the function.
9971 * selftest.h: Declare new tests.
9973 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
9976 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
9977 gen_add3_insn did not fail.
9978 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
9979 r0, construct that number in a temporary reg and add that reg to r0.
9980 If asked to put the result in r0 as well, fail.
9982 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
9984 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
9985 for early expansion of vec_eqv.
9987 2017-06-08 Jakub Jelinek <jakub@redhat.com>
9990 * ubsan.c (instrument_null): Avoid pointless code temporary.
9991 (pass_ubsan::execute): Instrument aggregate arguments of calls.
9993 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
9997 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
10000 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
10003 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10005 * predict.c (unlikely_executed_edge_p): Move ahead.
10006 (probably_never_executed_edge_p): Use it.
10008 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10010 PR middle-end/79988
10011 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
10012 gimple_call_builtin_p call.
10014 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10016 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
10017 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
10018 rtl_check_failed_type2, rtl_check_failed_code1,
10019 rtl_check_failed_code2, rtl_check_failed_code_mode,
10020 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
10021 rtvec_check_failed_bounds, rtl_check_failed_flag,
10022 _fatal_insn_not_found, _fatal_insn): Likewise.
10023 * tree.h (tree_contains_struct_check_failed,
10024 tree_check_failed, tree_not_check_failed,
10025 tree_class_check_failed, tree_range_check_failed,
10026 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
10027 tree_vec_elt_check_failed, phi_node_elt_check_failed,
10028 tree_operand_check_failed, omp_clause_check_failed,
10029 omp_clause_operand_check_failed, omp_clause_range_check_failed):
10032 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10034 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
10035 flag_branch_probabilities.
10036 * ipa-inline.c (edge_badness): Likewise.
10037 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
10038 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
10039 * predict.c (maybe_hot_frequency_p): Likewise.
10040 (probably_never_executed): Likewise.
10041 * sched-ebb.c (schedule_ebbs): Likewise.
10042 * sched-rgn.c (find_single_block_region): Likewise.
10043 * tracer.c (tail_duplicate): Likewise.
10045 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10047 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
10048 longer requires x_flag_profile_use.
10050 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10052 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
10053 instead of flag_reorder_blocks_and_partition.
10054 * dbxout.c (dbxout_function_end): Likewise.
10055 * dwarf2out.c (gen_subprogram_die): Likewise.
10056 * haifa-sched.c (sched_create_recovery_edges): Likewise.
10057 * hw-doloop.c (reorg_loops): Likewise.
10058 * varasm.c (assemble_start_function,
10059 assemble_end_function): Likewise.
10060 (decide_function_section): Do not check for
10061 flag_reorder_blocks_and_partition.
10063 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10065 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10067 (chkp_get_hard_register_fake_addr_expr): Ditto.
10068 (chkp_build_addr_expr): Add check for hard reg case.
10069 (chkp_parse_array_and_component_ref): Ditto.
10070 (chkp_find_bounds_1): Ditto.
10071 (chkp_process_stmt): Don't generate bounds store for
10074 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10076 * predict.c (maybe_hot_bb_p): Do not check profile status.
10077 (maybe_hot_edge_p): Likewise.
10078 (probably_never_executed): Check for zero counts even if profile
10080 (unlikely_executed_edge_p): New function.
10081 (unlikely_executed_stmt_p): New function.
10082 (unlikely_executed_bb_p): New function.
10083 (set_even_probabilities): Use unlikely predicates.
10084 (combine_predictions_for_bb): Likewise.
10085 (predict_paths_for_bb): Likewise.
10086 (predict_paths_leading_to_edge): Likewise.
10087 (determine_unlikely_bbs): New function.
10088 (estimate_bb_frequencies): Use it.
10089 (compute_function_frequency): Use zero counts even if profile is
10091 * profile-count.h: Fix typo.
10093 2017-08-08 Julia Koval <julia.koval@intel.com>
10095 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
10096 _mm512_mask_cvtsepi16_storeu_epi8,
10097 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
10098 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
10099 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
10100 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
10101 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
10102 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
10103 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
10104 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
10105 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
10106 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
10107 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
10108 __builtin_ia32_pmovuswb256mem_mask,
10109 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
10110 __builtin_ia32_pmovwb512mem_mask): New builtins.
10112 2017-08-08 Julia Koval <julia.koval@intel.com>
10114 PR target/73350,80862
10115 * config/i386/subst.md (round): Fix round pattern.
10116 * config/i386/i386.c (ix86_erase_embedded_rounding):
10117 Fix erasing rounding for the fixed pattern.
10119 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10121 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
10123 2017-06-08 Martin Liska <mliska@suse.cz>
10125 PR gcov-profile/80911
10126 * gcov.c (block_info::block_info): New constructor.
10128 2017-06-07 Carl Love <cel@us.ibm.com>
10130 * config/rs6000/rs6000-c: The return type of the following
10131 built-in functions was implemented as int not long long. Fix sign
10132 of return value for the unsigned version of vec_mulo and vec_mule.
10133 vector unsigned long long vec_bperm (vector unsigned long long,
10134 vector unsigned char)
10135 vector signed long long vec_mule (vector signed int,
10137 vector unsigned long long vec_mule (vector unsigned int,
10138 vector unsigned int)
10139 vector signed long long vec_mulo (vector signed int,
10141 vector unsigned long long vec_mulo (vector unsigned int,
10142 vector unsigned int)
10143 * doc/extend.texi: Fix the documentation for the built-in
10146 2017-06-07 Carl Love <cel@us.ibm.com>
10149 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
10152 2017-06-07 Carl Love <cel@us.ibm.com>
10154 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
10155 support, Generate doublehv for signed int/float for BE case only.
10157 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
10159 * doc/invoke.texi (mcx16): Rewrite.
10161 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10163 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
10164 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
10165 *mov<mode>_softfloat, and an anonymous splitter): Use
10166 nonimmediate_operand instead of rs6000_nonimmediate_operand.
10168 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10170 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
10172 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
10173 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
10174 (rs6000_debug_reg_global): Adjust.
10175 (rs6000_init_hard_regno_mode_ok): Adjust.
10176 (rs6000_dbx_register_number): Adjust.
10177 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
10178 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
10179 Remove SPE_ACC and SPEFSCR.
10180 (REG_ALLOC_ORDER): Ditto.
10181 (FRAME_POINTER_REGNUM): Change to 111.
10182 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
10183 (REG_CLASS_NAMES): Ditto.
10184 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
10185 (REGISTER_NAMES): Ditto.
10186 (ADDITIONAL_REG_NAMES): Ditto.
10187 (rs6000_reg_names): Ditto.
10188 * config/rs6000/rs6000.md: Renumber some register number
10191 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10193 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
10195 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
10196 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
10198 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
10199 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
10200 Delete the SPE high registers.
10201 (REG_ALLOC_ORDER): Ditto.
10202 (enum reg_class): Remove SPE_HIGH_REGS.
10203 (REG_CLASS_NAMES): Ditto.
10204 (REG_CLASS_CONTENTS): Delete the SPE high registers.
10205 (REGISTER_NAMES): Ditto.
10206 (rs6000_reg_names): Ditto.
10207 * doc/tm.texi.in: Remove SPE as example.
10208 * doc/tm.texi: Regenerate.
10210 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10212 * config/rs6000/8540.md (ppc8540_brinc): Delete.
10213 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
10214 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
10215 * config/rs6000/rs6000.md (type): Remove "brinc".
10217 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10219 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
10220 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
10221 * config/rs6000/linuxspe.h: Delete file.
10222 * config/rs6000/rs6000.md: Don't include spe.md.
10223 * config/rs6000/spe.h: Delete file.
10224 * config/rs6000/spe.md: Delete file.
10225 * config/rs6000/t-rs6000: Remove spe.md.
10227 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10229 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
10230 (reg_or_none500mem_operand): Delete.
10231 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
10232 instead of reg_or_none500mem_operand.
10234 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10236 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
10237 handling of SPE flags.
10238 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
10240 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10242 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
10244 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
10245 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
10246 paired_divv2sf3): Similar.
10247 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
10248 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
10249 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
10251 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
10252 Rename the paired_* instruction patterns.
10253 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
10255 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
10256 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
10257 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
10258 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
10259 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
10260 PAIRED_VECTOR_MODE.
10261 (struct machine_function): Delete field spe_insn_chain_scanned_p.
10262 (spe_func_has_64bit_regs_p): Delete.
10263 (spe_expand_predicate_builtin): Delete.
10264 (spe_expand_evsel_builtin): Delete.
10265 (TARGET_DWARF_REGISTER_SPAN): Do not define.
10266 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
10267 (invalid_e500_subreg): Delete.
10268 (rs6000_legitimize_address): Always force_reg op2 as well, for
10269 paired single memory accesses.
10270 (rs6000_member_type_forces_blk): Delete.
10271 (rs6000_spe_function_arg): Delete.
10272 (rs6000_expand_unop_builtin): Delete SPE handling.
10273 (rs6000_expand_binop_builtin): Ditto.
10274 (spe_expand_stv_builtin): Delete.
10275 (bdesc_2arg_spe): Delete.
10276 (spe_expand_builtin): Delete.
10277 (spe_expand_predicate_builtin): Delete.
10278 (spe_expand_evsel_builtin): Delete.
10279 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
10280 (spe_init_builtins): Delete.
10281 (spe_func_has_64bit_regs_p): Delete.
10282 (savres_routine_name): Delete "info" parameter. Adjust callers.
10283 (rs6000_emit_stack_reset): Ditto.
10284 (rs6000_dwarf_register_span): Delete.
10285 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
10286 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
10287 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
10289 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
10291 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
10292 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
10293 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
10294 mulv2sf3, divv2sf3): Delete expanders.
10296 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10298 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
10300 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10302 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
10303 * config/rs6000/rs6000.c: Ditto.
10305 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10307 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
10308 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
10309 comparison_operator.
10311 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10313 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
10314 * config/rs6000/rs6000.opt: Ditto.
10315 * config/rs6000/t-rtems: Ditto.
10317 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10319 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
10320 TARGET_E500_SINGLE by 0, simplify.
10321 * config/rs6000/rs6000.c: Ditto.
10322 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
10323 (spe_build_register_parallel): Delete.
10324 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
10325 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
10326 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
10327 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
10328 (E500_CONVERT): Delete.
10329 * config/rs6000/spe.md: Remove many patterns and all define_constants.
10331 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10333 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
10334 * config/rs6000/dfp.md: Ditto.
10335 (negdd2, *negdd2_fpr): Merge.
10336 (absdd2, *absdd2_fpr): Merge.
10337 (negtd2, *negtd2_fpr): Merge.
10338 (abstd2, *abstd2_fpr): Merge.
10339 * config/rs6000/e500.h: Delete file.
10340 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
10341 TARGET_FPRS by 1 and simplify.
10342 * config/rs6000/rs6000-c.c: Ditto.
10343 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
10344 TARGET_DF_SPE by 0.
10345 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
10347 * config/rs6000/rs6000.md: Ditto.
10348 (floatdidf2, *floatdidf2_fpr): Merge.
10349 (move_from_CR_gt_bit): Delete.
10350 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
10351 (E500_CR_IOR_COMPARE): Delete.
10352 (All patterns that require !TARGET_FPRS): Delete.
10353 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
10355 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10357 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
10359 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10361 * graphds.c (add_edge): Intitialize edge's attached data.
10362 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
10363 pointer parameter. Call pointed function on each edge during
10364 graph traversing. Skip traversing the edge when the function
10366 (graphds_dfs, graphds_scc): Ditto.
10367 (for_each_edge): New parameter. Pass the new parameter to callback
10369 * graphds.h (skip_edge_callback): New function pointer type.
10370 (graphds_dfs, graphds_scc): New function pointer parameter.
10371 (graphds_edge_callback, for_each_edge): New parameter.
10373 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10375 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
10376 out code checking if runtime alias check is possible to below ...
10377 Call the new function.
10378 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
10379 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
10381 2017-06-07 Marek Polacek <polacek@redhat.com>
10384 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
10385 TYPE_OVERFLOW_WRAPS check.
10387 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10389 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
10390 if versioning is required.
10391 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
10392 peeling with the check for versioning.
10394 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10396 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
10397 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
10398 Set true to new parameter if new ssa variable is defined.
10399 (vect_gen_vector_loop_niters): Refactor. Set range information
10400 for the new vector loop bound variable.
10401 (vect_do_peeling): Ditto.
10403 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10405 * tree-affine.c (ssa.h): Include header file.
10406 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
10407 has wrapping overflow behavior.
10409 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10411 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
10413 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10415 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
10416 (tree_to_aff_combination): ... here.
10418 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10420 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
10421 reg_pressure model function.
10422 (ivopts_global_cost_for_size): Delete.
10423 (determine_set_costs, iv_ca_recount_cost): Call new model function
10424 ivopts_estimate_reg_pressure.
10426 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10428 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
10429 expensive than udiv. Remove floating point cases from mod.
10431 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10433 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
10434 Increase idiv cost.
10436 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10438 * config/aarch64/aarch64.md
10439 (copysignsf3): Fix mask generation.
10441 2017-06-07 Jakub Jelinek <jakub@redhat.com>
10443 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
10445 (class dump_manager): Add register_dumps method.
10446 * dumpfile.c: Include langhooks.h.
10447 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
10448 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
10449 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
10450 (dump_manager::dump_register): Start with 512 entries instead of 32.
10451 (dump_manager::register_dumps): New method.
10452 * toplev.c (general_init): Instead of invoking register_dumps
10453 langhook, invoke register_dumps method on the dump manager.
10454 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
10457 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
10459 * doc/md.texi: Clarify the restrictions on a define_insn condition.
10460 Say that # requires an associated define_split to exist, and that
10461 the define_split must be suitable for use after register allocation.
10463 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10465 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
10466 (compute_outgoing_frequencies): Also initialize zero counts.
10467 (find_many_sub_basic_blocks): Do not produce uninitialized profile
10468 around loops; preserve more of profile when nothing changes.
10470 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
10472 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
10474 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
10475 * config/arm/arm-cpu-cdata.h: Regenerate.
10476 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
10477 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
10478 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
10479 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
10480 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
10482 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
10484 * config/arm/t-rmprofile: Likewise.
10485 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
10487 2017-06-06 David S. Miller <davem@davemloft.net>
10490 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
10491 blockage if function uses alloca.
10493 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10495 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
10496 New "uid" fields to hold pretty-print IDs of group and ref.
10497 Memory references are now identified as <group_id>:<ref_id>
10498 instead of using [random] addresses.
10499 (dump_mem_details): Simplify, no functional change.
10500 (dump_mem_ref): Simplify and make output more concise.
10501 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
10502 (find_or_create_group): Initialize group uid.
10503 (record_ref): Initialize ref uid. Improve debug output.
10504 (prune_group_by_reuse, should_issue_prefetch_p,)
10505 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
10506 (mark_nontemporal_store, determine_loop_nest_reuse):
10507 Improve debug output.
10509 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10511 * dbgcnt.def (prefetch): New debug counter.
10512 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
10513 (schedule_prefetches): Stop issueing prefetches if debug counter
10516 2017-06-06 Tom de Vries <tom@codesourcery.com>
10518 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
10519 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
10521 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10523 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
10524 Use aarch64_reg_or_zero predicate for operand 4.
10525 (aarch64_compare_and_swap<mode> define_insn_and_split):
10526 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
10527 (aarch64_store_exclusive<mode>): Likewise for operand 2.
10529 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10531 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
10532 (arm_compute_save_core_reg_mask): This.
10533 (thumb1_compute_save_reg_mask): Rename into ...
10534 (thumb1_compute_save_core_reg_mask): This.
10535 (arm_compute_save_reg0_reg12_mask): Adapt comment.
10536 (arm_compute_frame_layout): Likewise.
10538 2017-06-06 Richard Biener <rguenther@suse.de>
10540 PR tree-optimization/80974
10541 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
10542 keep or clear leaders SSA info.
10544 2017-06-06 Tom de Vries <tom@codesourcery.com>
10546 * config/nvptx/nvptx.c (split_mode_p): New function.
10547 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
10549 2017-06-06 Tom de Vries <tom@codesourcery.com>
10551 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
10553 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10556 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
10559 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10561 * shrink-wrap.c (handle_simple_exit): Update profile.
10562 (try_shrink_wrapping): Upate profile.
10564 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10566 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
10567 (tree_guess_outgoing_edge_probabilities): New.
10568 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
10569 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
10571 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10573 * ipa-split.c (split_function): Initialize return bb profile.
10575 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10577 * profile.c (compute_branch_probabilities): Also initialize
10578 EXIT_BLOCK profile.
10580 2017-06-06 Richard Biener <rguenther@suse.de>
10582 PR tree-optimization/80928
10583 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
10584 (vect_analyze_loop_operations): Properly guard analysis for
10586 (vect_transform_loop): Likewise.
10587 (vect_analyze_loop_2): Also reset SLP type on PHIs.
10588 (vect_model_induction_cost): Do not cost for pure SLP.
10589 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
10590 of induction in inner loop vectorization.
10591 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
10592 (vect_get_and_check_slp_defs): Handle vect_induction_def.
10593 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
10595 (vect_analyze_slp_cost_1): Cost induction.
10596 (vect_detect_hybrid_slp_stmts): Handle PHIs.
10597 (vect_get_slp_vect_defs): Likewise.
10598 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
10599 (vect_transform_stmt): Handle SLP reductions.
10600 * tree-vectorizer.h (vectorizable_induction): Adjust.
10602 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
10604 * config/rs6000/rs6000.c (make_resolver_func): Update
10605 init_lowered_empty_function call.
10607 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
10609 * doc/invoke.texi: Document the -fprofile-abs-path option.
10610 * common.opt (fprofile-abs-path): New option.
10611 * gcov-io.h (gcov_write_filename): Declare.
10612 * gcov-io.c (gcov_write_filename): New function.
10613 * coverage.c (coverage_begin_function): Use gcov_write_filename.
10614 * profile.c (output_location): Likewise.
10616 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10618 * shring-wrap.c: Revert accidental commit.
10620 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
10622 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
10624 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10626 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
10628 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
10629 profile in callgraph edge.
10630 * profile-count.h (apply_probability): If THIS is 0, then result is 0
10631 (apply_scale): Likewise.
10632 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
10633 Also scale profile when inlining function with zero profile.
10634 (initialize_cfun): Update exit block profile even when it is zero.
10635 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
10636 when profile is read.
10638 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
10640 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
10641 (CLONE_*): New constants to define the processors we can generate
10642 code for with the target_clone attribute.
10643 (rs6000_clone_map): New array to identify which clone processors
10644 the current program is running on.
10645 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
10646 target_clone attribute.
10647 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
10648 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
10649 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
10650 (cpu_expand_builtin): Add support for target_clone attribute.
10651 (rs6000_valid_attribute_p): Allow "default" attribute.
10652 (get_decl_name): New debug function to simplify printing the
10653 current function name in debugging statements.
10654 (rs6000_clone_priority): New functions to support the target_clone
10655 attribute, and be able to generate code to switch between ISA 2.05
10656 through ISA 3.0 (power6 through power9).
10657 (rs6000_compare_version_priority): Likewise.
10658 (rs6000_get_function_versions_dispatcher): Likewise.
10659 (make_resolver_func): Likewise.
10660 (add_condition_to_bb): Likewise.
10661 (dispatch_function_versions): Likewise.
10662 (rs6000_generate_version_dispatcher_body): Likewise.
10663 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
10664 (fusion_gpr_load_p): Fix a spacing issue.
10665 * doc/extend.texi (Common Function Attributes): Document that the
10666 PowerPC supports the target_clone attribute.
10668 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
10670 * config/arm/arm.h: explain F symbol found in description of ARM
10671 register allocation in its legend.
10673 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10675 * config/mips/frame-header-opt.c: Include profile-count.h.
10676 * config/riscv/riscv.c: Include profile-count.h
10678 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10680 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
10682 (sm_set_flag_if_changed): Add bbs field.
10683 (execute_sm_if_changed_flag_set): Pass BBS.
10684 (execute_sm): Update.
10686 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10688 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
10691 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10693 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
10694 (peephole2): New peephole2 to emit the above.
10695 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
10697 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10699 * config/aarch64/aarch64.c (define_peephole2 above
10700 *sub_<shift>_<mode>): New peephole.
10702 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
10704 * config/i386/i386.c (make_resolver_func): Update.
10705 * Makefile.in: Add profile-count.h and profile-count.o
10706 * auto-profile.c (afdo_indirect_call): Update to new API.
10707 (afdo_set_bb_count): Update.
10708 (afdo_propagate_edge): Update.
10709 (afdo_propagate_circuit): Update.
10710 (afdo_calculate_branch_prob): Update.
10711 (afdo_annotate_cfg): Update.
10712 * basic-block.h: Include profile-count.h
10713 (struct edge_def): Turn count to profile_count.
10714 (struct basic_block_def): Likewie.
10715 (REG_BR_PROB_BASE): Move to profile-count.h
10716 (RDIV): Move to profile-count.h
10717 * bb-reorder.c (max_entry_count): Turn to profile_count.
10718 (find_traces): Update.
10719 (rotate_loop):Update.
10720 (connect_traces):Update.
10721 (sanitize_hot_paths):Update.
10722 * bt-load.c (migrate_btr_defs): Update.
10723 * cfg.c (RDIV): Remove.
10724 (init_flow): Use alloc_block.
10725 (alloc_block): Uninitialize count.
10726 (unchecked_make_edge): Uninitialize count.
10727 (check_bb_profile): Update.
10728 (dump_edge_info): Update.
10729 (dump_bb_info): Update.
10730 (update_bb_profile_for_threading): Update.
10731 (scale_bbs_frequencies_int): Update.
10732 (scale_bbs_frequencies_gcov_type): Update.
10733 (scale_bbs_frequencies_profile_count): New.
10734 * cfg.h (update_bb_profile_for_threading): Update.
10735 (scale_bbs_frequencies_profile_count): Declare.
10736 * cfgbuild.c (compute_outgoing_frequencies): Update.
10737 (find_many_sub_basic_blocks): Update.
10738 * cfgcleanup.c (try_forward_edges): Update.
10739 (try_crossjump_to_edge): Update.
10740 * cfgexpand.c (expand_gimple_tailcall): Update.
10741 (construct_exit_block): Update.
10742 * cfghooks.c (verify_flow_info): Update.
10743 (dump_bb_for_graph): Update.
10744 (split_edge): Update.
10745 (make_forwarder_block): Update.
10746 (duplicate_block): Update.
10747 (account_profile_record): Update.
10748 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10749 (get_estimated_loop_iterations): Update.
10750 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10751 (single_likely_exit): Update.
10752 * cfgloopmanip.c (scale_loop_profile): Update.
10754 (set_zero_probability): Update.
10755 (lv_adjust_loop_entry_edge): Update.
10756 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10757 (purge_dead_edges): Update.
10758 (rtl_account_profile_record): Update.
10759 * cgraph.c (cgraph_node::create): Uninitialize count.
10760 (symbol_table::create_edge): Uninitialize count.
10761 (cgraph_update_edges_for_call_stmt_node): Update.
10762 (cgraph_edge::dump_edge_flags): Update.
10763 (cgraph_node::dump): Update.
10764 (cgraph_edge::maybe_hot_p): Update.
10765 * cgraph.h: Include profile-count.h
10766 (create_clone), create_edge, create_indirect_edge): Update.
10767 (cgraph_node): Turn count to profile_count.
10768 (cgraph_edge0: Likewise.
10769 (make_speculative, clone): Update.
10770 (create_edge): Update.
10771 (init_lowered_empty_function): Update.
10772 * cgraphclones.c (cgraph_edge::clone): Update.
10773 (duplicate_thunk_for_node): Update.
10774 (cgraph_node::create_clone): Update.
10775 * cgraphunit.c (cgraph_node::analyze): Update.
10776 (cgraph_node::expand_thunk): Update.
10777 * final.c (dump_basic_block_info): Update.
10778 * gimple-streamer-in.c (input_bb): Update.
10779 * gimple-streamer-out.c (output_bb): Update.
10780 * graphite.c (print_global_statistics): Update.
10781 (print_graphite_scop_statistics): Update.
10782 * hsa-brig.c: Include basic-block.h.
10783 * hsa-dump.c: Include basic-block.h.
10784 * hsa-gen.c (T sum_slice): Update.
10785 (convert_switch_statements):Update.
10786 * hsa-regalloc.c: Include basic-block.h.
10787 * ipa-chkp.c (chkp_produce_thunks): Update.
10788 * ipa-cp.c (struct caller_statistics): Update.
10789 (init_caller_stats): Update.
10790 (gather_caller_stats): Update.
10791 (ipcp_cloning_candidate_p): Update.
10792 (good_cloning_opportunity_p): Update.
10793 (get_info_about_necessary_edges): Update.
10794 (dump_profile_updates): Update.
10795 (update_profiling_info): Update.
10796 (update_specialized_profile): Update.
10797 (perhaps_add_new_callers): Update.
10798 (decide_about_value): Update.
10799 (ipa_cp_c_finalize): Update.
10800 * ipa-devirt.c (struct odr_type_warn_count): Update.
10801 (struct decl_warn_count): Update.
10802 (struct final_warning_record): Update.
10803 (possible_polymorphic_call_targets): Update.
10804 (ipa_devirt): Update.
10805 * ipa-fnsummary.c (redirect_to_unreachable): Update.
10806 * ipa-icf.c (sem_function::merge): Update.
10807 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
10808 * ipa-inline.c (compute_uninlined_call_time): Update.
10809 (compute_inlined_call_time): Update.
10810 (want_inline_small_function_p): Update.
10811 (want_inline_self_recursive_call_p): Update.
10812 (edge_badness): Update.
10813 (lookup_recursive_calls): Update.
10814 (recursive_inlining): Update.
10815 (inline_small_functions): Update.
10816 (dump_overall_stats): Update.
10817 (dump_inline_stats): Update.
10818 * ipa-profile.c (ipa_profile_generate_summary): Update.
10819 (ipa_propagate_frequency): Update.
10820 (ipa_profile): Update.
10821 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
10822 * ipa-utils.c (ipa_merge_profiles): Update.
10823 * loop-doloop.c (doloop_modify): Update.
10824 * loop-unroll.c (report_unroll): Update.
10825 (unroll_loop_runtime_iterations): Update.
10826 * lto-cgraph.c (lto_output_edge): Update.
10827 (lto_output_node): Update.
10828 (input_node): Update.
10829 (input_edge): Update.
10830 (merge_profile_summaries): Update.
10831 * lto-streamer-in.c (input_cfg): Update.
10832 * lto-streamer-out.c (output_cfg): Update.
10833 * mcf.c (create_fixup_graph): Update.
10834 (adjust_cfg_counts): Update.
10835 (sum_edge_counts): Update.
10836 * modulo-sched.c (sms_schedule): Update.
10837 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10838 * predict.c (maybe_hot_count_p): Update.
10839 (probably_never_executed): Update.
10840 (dump_prediction): Update.
10841 (combine_predictions_for_bb): Update.
10842 (propagate_freq): Update.
10843 (handle_missing_profiles): Update.
10844 (counts_to_freqs): Update.
10845 (rebuild_frequencies): Update.
10846 (force_edge_cold): Update.
10847 * predict.h: Include profile-count.h
10848 (maybe_hot_count_p, counts_to_freqs): UPdate.
10849 * print-rtl-function.c: Do not include cfg.h
10850 * print-rtl.c: Include basic-block.h
10851 * profile-count.c: New file.
10852 * profile-count.h: New file.
10853 * profile.c (is_edge_inconsistent): Update.
10854 (correct_negative_edge_counts): Update.
10855 (is_inconsistent): Update.
10856 (set_bb_counts): Update.
10857 (read_profile_edge_counts): Update.
10858 (compute_frequency_overlap): Update.
10859 (compute_branch_probabilities): Update; Initialize and deinitialize
10861 (branch_prob): Update.
10862 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
10863 (edge_gcov_count): New.
10864 (bb_gcov_count): New.
10865 * shrink-wrap.c (try_shrink_wrapping): Update.
10866 * tracer.c (better_p): Update.
10867 * trans-mem.c (expand_transaction): Update.
10868 (ipa_tm_insert_irr_call): Update.
10869 (ipa_tm_insert_gettmclone_call): Update.
10870 * tree-call-cdce.c: Update.
10871 * tree-cfg.c (gimple_duplicate_sese_region): Update.
10872 (gimple_duplicate_sese_tail): Update.
10873 (gimple_account_profile_record): Update.
10874 (execute_fixup_cfg): Update.
10875 * tree-inline.c (copy_bb): Update.
10876 (copy_edges_for_bb): Update.
10877 (initialize_cfun): Update.
10878 (freqs_to_counts): Update.
10879 (copy_cfg_body): Update.
10880 (expand_call_inline): Update.
10881 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10882 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
10883 (try_unroll_loop_completely): Update.
10884 (try_peel_loop): Update.
10885 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10886 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
10887 * tree-ssa-loop-split.c (connect_loops): Update.
10888 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10889 * tree-ssa-reassoc.c (branch_fixup): Update.
10890 * tree-ssa-tail-merge.c (replace_block_by): Update.
10891 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
10892 (compute_path_counts): Update.
10893 (update_profile): Update.
10894 (recompute_probabilities): Update.
10895 (update_joiner_offpath_counts): Update.
10896 (estimated_freqs_path): Update.
10897 (freqs_to_counts_path): Update.
10898 (clear_counts_path): Update.
10899 (ssa_fix_duplicate_block_edges): Update.
10900 (duplicate_thread_path): Update.
10901 * tree-switch-conversion.c (case_bit_test_cmp): Update.
10902 (struct switch_conv_info): Update.
10903 * tree-tailcall.c (decrease_profile): Update.
10904 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10905 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10906 * value-prof.c (check_counter): Update.
10907 (gimple_divmod_fixed_value): Update.
10908 (gimple_mod_pow2): Update.
10909 (gimple_mod_subtract): Update.
10910 (gimple_ic_transform): Update.
10911 (gimple_stringop_fixed_value): Update.
10912 * value-prof.h (gimple_ic): Update.
10914 2017-06-02 Carl Love <cel@us.ibm.com>
10916 * config/rs6000/rs6000-c: Add support for built-in functions
10917 vector double vec_doublee (vector signed int);
10918 vector double vec_doublee (vector unsigned int);
10919 vector double vec_doublee (vector float);
10920 vector double vec_doubleh (vector signed int);
10921 vector double vec_doubleh (vector unsigned int);
10922 vector double vec_doubleh (vector float);
10923 vector double vec_doublel (vector signed int);
10924 vector double vec_doublel (vector unsigned int);
10925 vector double vec_doublel (vector float);
10926 vector double vec_doubleo (vector signed int);
10927 vector double vec_doubleo (vector unsigned int);
10928 vector double vec_doubleo (vector float);.
10929 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
10930 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
10932 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
10933 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
10934 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
10936 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
10937 vec_doublel, vec_doubleh.
10938 * doc/extend.texi: Update the built-in documentation file for the
10939 new built-in functions.
10941 2017-06-02 David Malcolm <dmalcolm@redhat.com>
10944 * ipa-inline-analysis.c (free_growth_caches): Set
10945 edge_removal_hook_holder to NULL after removing it.
10947 2017-06-02 Sudakshina Das <sudi.das@arm.com>
10949 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
10950 comparision with zero.
10952 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
10953 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10954 for early expansion of vec_min and vec_max builtins.
10955 (builtin_function_type): Add min/max unsigned variants to those
10956 identified as having unsigned arguments.
10958 2017-06-02 Olivier Hainque <hainque@adacore.com>
10960 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
10962 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10964 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
10965 Use VALL_F16 iterator rather than VALL.
10967 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10969 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
10970 Emit CBNZ inside loop when doing a strong exchange and comparing
10971 against zero. Generate the CC flags after the loop.
10973 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
10975 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
10976 (dl_section_ref): New.
10977 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
10978 On AIX, append an expression to subtract the size of the
10979 section length to dl_section_ref.
10981 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
10983 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10984 for early expansion of vector absolute builtins.
10986 2017-06-02 Richard Biener <rguenther@suse.de>
10988 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
10989 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
10991 2017-06-02 Richard Biener <rguenther@suse.de>
10993 PR tree-optimization/80948
10994 * tree-tailcall.c (find_tail_calls): Track stmts to move in
10995 stmt order as well.
10997 2017-06-02 Richard Biener <rguenther@suse.de>
10999 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
11001 * tree-vect-stmts.c (process_use): Do not mark backedge defs
11002 for inductions as relevant.
11004 2017-06-02 Richard Biener <rguenther@suse.de>
11006 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
11007 (vectorizable_induction): ... this. Remove dead code.
11009 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
11011 * builtins. (expand_builtin_alloca): Remove second parameter and
11012 infer its value from the first parameter instead.
11013 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
11015 2017-06-02 Jakub Jelinek <jakub@redhat.com>
11017 PR rtl-optimization/80903
11018 * loop-doloop.c (add_test): Unshare sequence.
11020 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
11022 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
11024 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
11026 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
11028 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
11029 xlogue_layout::get_instance, logue_layout::xlogue_layout,
11030 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
11031 (xlogue_layout::get_stub_rtx): Make static.
11032 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
11033 (xlogue_layout::compute_stub_managed_regs): Rename to...
11034 (xlogue_layout::count_stub_managed_regs): ...this.
11035 (xlogue_layout::is_stub_managed_reg): New function.
11036 (xlogue_layout::m_stub_names): Rename to...
11037 (xlogue_layout::s_stub_names): ...this, make static.
11038 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
11039 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
11040 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
11041 xlogue_layout::s_stub_names): Instantiate statics.
11042 (stub_managed_regs): Remove.
11043 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
11044 (disable_call_ms2sysv_xlogues): Rename to...
11045 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
11046 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
11048 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
11049 change after reload_completed.
11050 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
11052 (ix86_expand_prologue): Likewise.
11053 (ix86_expand_epilogue): Likewise.
11054 (ix86_expand_split_stack_prologue): Likewise.
11055 (ix86_compute_frame_layout): Remove frame parameter ...
11056 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
11057 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
11059 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
11060 (ix86_frame): Move from here ...
11061 * config/i386/i386.h (ix86_frame): ... to here.
11062 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
11063 complete ix86_frame data structure instead. Remove some_ld_name.
11065 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
11067 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
11068 symbols that hold a DECL_VALUE_EXPR.
11070 2017-06-01 Martin Jambor <mjambor@suse.cz>
11072 PR tree-optimization/80898
11073 * tree-sra.c (process_subtree_disqualification): Removed.
11074 (disqualify_candidate): Do not acll
11075 process_subtree_disqualification.
11076 (subtree_mark_written_and_enqueue): New function.
11077 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
11078 RHS has been disqualified and re-queue LHS if necessary. Apart
11079 from that, ignore disqualified RHS.
11081 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11083 * config/s390/s390.c (s390_emit_epilogue): Disable early return
11084 address fetch for z10 or later.
11086 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11088 * config/arc/arc.md (tst_movb): Add guard when splitting.
11090 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11092 * config/arc/arc.c (arc_can_eliminate): Test against
11093 arc_frame_pointer_needed.
11095 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11097 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
11098 to prevent store reordering.
11099 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
11100 (type): Add block type.
11101 (stack_tie): Define special instruction to be used in
11104 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11106 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
11107 constraint. It is not valid for the pattern.
11108 (noncommutative_binary_comparison): Likewise.
11110 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11112 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
11115 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11117 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
11118 be used by the reg-alloc.
11120 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11122 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
11123 reg-alloc when having mul64 or mul32x16 instructions.
11124 (mulsidi3): Likewise.
11125 (umulsidi3): Likewise.
11126 (mulsi32x16): New pattern.
11127 (mulsi64): Likewise.
11128 (mulsidi64): Likewise.
11129 (umulsidi64): Likewise.
11130 (MUL32x16_REG): Define.
11131 (mul64_600): Use MUL32x16_REG.
11132 (mac64_600): Likewise.
11133 (umul64_600): Likewise.
11134 (umac64_600): Likewise.
11136 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11138 * config/arc/arc.md (mulsi3_700): Make it commutative.
11140 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
11142 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
11144 (*sign_extendsidi2_insn): Likewise for movstosw.
11146 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
11148 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
11149 the type of the input discriminant value. Convert the
11150 discriminant value of signedness vary.
11152 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
11154 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
11155 Add to -Wall section.
11157 2017-06-01 Richard Biener <rguenther@suse.de>
11159 PR middle-end/66313
11160 * fold-const.c (fold_plusminus_mult_expr): If the factored
11161 factor may be zero use a wrapping type for the inner operation.
11162 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
11163 and handle moved defs.
11164 (process_assignment): Properly guard the unary op case. Return a
11165 tri-state indicating that moving the stmt before the call may allow
11166 to continue. Pass through to_move.
11167 (find_tail_calls): Handle moving unrelated defs before
11170 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
11173 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
11174 splitter result in the canonical way.
11176 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
11178 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
11179 also for 32bit target. Update insn attributes.
11180 (zero-extendsidi2 splitter): Allow all registers for operand 1.
11182 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
11184 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
11185 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
11186 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
11187 (_mm_maskz_min_ss): New intrinsics.
11189 2017-05-31 Martin Liska <mliska@suse.cz>
11191 * tree-vect-loop.c (vect_create_epilog_for_reduction):
11192 Change comment style to one we normally use.
11193 (vectorizable_reduction): Likewise.
11194 (vectorizable_induction): Likewise.
11195 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11196 (vectorizable_call): Likewise.
11197 (vectorizable_simd_clone_call): Likewise.
11198 (vectorizable_conversion): Likewise.
11199 (vectorizable_assignment): Likewise.
11200 (vectorizable_shift): Likewise.
11201 (vectorizable_operation): Likewise.
11202 (vectorizable_store): Likewise.
11203 (vectorizable_load): Likewise.
11204 * tree-vectorizer.h: Likewise.
11206 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
11208 * passes.c (emergency_dump_function): New.
11209 * tree-pass.h (emergency_dump_function): Declare.
11210 * plugin.c (plugins_internal_error_function): Remove.
11211 * plugin.h (plugins_internal_error_function): Remove declaration.
11212 * toplev.c (internal_error_function): New static function. Use it...
11213 (general_init): ...here.
11215 2017-05-31 Graham Markall <graham.markall@embecosm.com>
11217 * config/arc/arc.c (arc_print_operand): Handle constant operands.
11218 (arc_rtx_costs): Add costs for new patterns.
11219 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
11220 * config/arc/predicates.md: Add _1_2_3_operand predicate.
11222 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11224 * tree-ssa-strlen.c (get_next_strinfo): New function.
11225 (get_stridx_plus_constant): Use it.
11226 (zero_length_string): Likewise.
11227 (adjust_related_strinfos): Likewise.
11228 (adjust_last_stmt): Likewise.
11230 2017-05-31 Richard Biener <rguenther@suse.de>
11233 * config/i386/i386.c (ix86_expand_builtin): Remove assert
11234 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
11236 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11238 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
11239 loop_vinfo argument and use of dependence distance vectors.
11240 Check instead whether the two references differ only in their
11241 initial value and assume that they have the same alignment if the
11242 difference is a multiple of the vector alignment.
11243 (vect_analyze_data_refs_alignment): Update call accordingly.
11245 2017-05-31 Martin Liska <mliska@suse.cz>
11248 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
11250 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11252 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
11253 (create_intersect_range_checks): Move from ...
11254 * tree-data-ref.c (create_intersect_range_checks_index)
11255 (create_intersect_range_checks): ... to here.
11256 (create_runtime_alias_checks): New function factored from ...
11257 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
11258 here. Call above function.
11259 * tree-data-ref.h (create_runtime_alias_checks): New function.
11261 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11263 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
11264 segment length for dr_b and compute it in wide_int.
11266 2017-05-31 Richard Biener <rguenther@suse.de>
11268 PR tree-optimization/80906
11269 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
11270 and pass through iv_map.
11271 (copy_bb_and_scalar_dependences): Adjust.
11272 (translate_pending_phi_nodes): Likewise.
11273 (copy_loop_close_phi_args): Handle code-generating IVs instead
11276 2017-05-30 David Malcolm <dmalcolm@redhat.com>
11278 * diagnostic-color.c (color_dict): Add "type-diff".
11279 (parse_gcc_colors): Update comment.
11280 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
11281 -fdiagnostics-show-template-tree and -fno-elide-type.
11282 (GCC_COLORS): Add type-diff to example.
11284 (-fdiagnostics-show-template-tree): New.
11285 (-fno-elide-type): New.
11286 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
11287 the pp_format_decoder callback. Call any m_format_postprocessor's
11289 (pretty_printer::pretty_printer): Initialize
11290 m_format_postprocessor.
11291 (pretty_printer::~pretty_printer): Delete any
11292 m_format_postprocessor.
11293 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
11294 (class format_postprocessor): New class.
11295 (struct pretty_printer::format_decoder): Document the new parameters.
11296 (struct pretty_printer::m_format_postprocessor): New field.
11297 * tree-diagnostic.c (default_tree_printer): Update for new
11298 bool and const char ** params.
11299 * tree-diagnostic.h (default_tree_printer): Likewise.
11301 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
11303 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
11304 (lwa_operand): Delete rs6000_gen_cell_microcode test.
11305 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
11306 rs6000_gen_cell_microcode code.
11307 (rs6000_final_prescan_insn): Delete.
11308 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
11309 "warn-cell-microcode" entries.
11310 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
11311 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
11312 throughout. Change cc_reg_not_micro_cr0_operand to
11313 cc_reg_not_cr0_operand throughout.
11314 (*extendhi<mode>2_noload): Delete.
11315 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
11316 (mwarn-cell-microcode): Delete.
11317 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
11318 -mgen-cell-microcode and -mwarn-cell-microcode.
11320 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
11323 * config/i386/constraints.md (Yd): New constraint.
11325 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
11326 and (?Yd, r) alternatives. Update insn attributes.
11327 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
11328 and (?*Yd, r) alternatives. Update insn attributes.
11329 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
11331 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
11333 * gimplify.c (gimplify_modify_expr): Don't create a
11334 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
11337 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
11339 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
11341 2017-05-30 Richard Biener <rguenther@suse.de>
11343 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
11344 and reduc_def fields.
11345 (STMT_VINFO_REDUC_TYPE): New define.
11346 (STMT_VINFO_REDUC_DEF): Likewise.
11347 (vect_force_simple_reduction): Adjust prototype.
11348 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
11349 (vect_is_simple_reduction): Remove check_reduction argument.
11350 (vect_force_simple_reduction): Adjust and set
11351 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11352 (vectorizable_reduction): Do not re-do reduction analysis
11353 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11354 * tree-parloops.c (gather_scalar_reductions): Adjust.
11356 2017-05-30 Richard Biener <rguenther@suse.de>
11358 PR middle-end/80901
11359 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
11362 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
11364 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
11365 Introduce unknown_misalignment parameter and remove vf.
11366 (vect_peeling_hash_get_lowest_cost):
11367 Pass unknown_misalignment parameter.
11368 (vect_enhance_data_refs_alignment):
11369 Fix unsupportable data ref treatment.
11371 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11373 * tree-vect-data-refs.c (vect_get_data_access_cost):
11374 Workaround for SLP handling.
11375 (vect_enhance_data_refs_alignment):
11376 Compute costs for doing no peeling at all, compare to the best
11377 peeling costs so far and avoid peeling if cheaper.
11379 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11381 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
11382 Return peeling info and set costs to zero for unlimited cost
11384 (vect_enhance_data_refs_alignment): Also inspect all datarefs
11385 with unknown misalignment. Compute and costs for unknown
11386 misalignment, compare them to the costs for known misalignment
11387 and choose the cheapest for peeling.
11389 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11391 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
11392 (vect_get_peeling_costs_all_drs): Create function.
11393 (vect_peeling_hash_get_lowest_cost):
11394 Use vect_get_peeling_costs_all_drs.
11395 (vect_peeling_supportable): Create function.
11396 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
11398 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11400 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
11401 DR_HAS_NEGATIVE_STEP.
11402 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
11403 (vect_enhance_data_refs_alignment): Use.
11404 (vect_duplicate_ssa_name_ptr_info): Use.
11405 * tree-vectorizer.h (dr_misalignment): Use.
11406 (known_alignment_for_access_p): Use.
11408 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11411 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
11413 (has_section_name): New function.
11415 2017-05-30 Martin Liska <mliska@suse.cz>
11418 * auto-profile.c (get_function_decl_from_block): Fix
11421 2017-05-30 Richard Biener <rguenther@suse.de>
11423 PR middle-end/80876
11424 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
11426 2017-05-30 Martin Liska <mliska@suse.cz>
11428 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
11429 * dumpfile.h (struct dump_file_info): Remove ctors.
11431 2017-05-30 Martin Liska <mliska@suse.cz>
11433 * predict.def: Fix GNU coding style.
11435 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11437 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
11438 Mark 'to' argument with ATTRIBUTE_UNUSED.
11440 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11442 * config/xtensa/xtensa.c (xtensa_emit_call): Use
11443 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
11444 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
11447 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
11449 * doc/install.texi (Options specification): Restore entry of
11450 --enable-sjlj-exceptions.
11452 2017-05-27 Michael Eager <eager@eagercon.com>
11455 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
11457 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
11459 * config/microblaze/microblaze.h
11460 (FIXED_REGISTERS): Update in macro.
11461 (CALL_USED_REGISTERS): Update in macro.
11463 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
11465 * doc/install.texi: Add links to macOS binary distributions.
11467 2017-05-27 Jakub Jelinek <jakub@redhat.com>
11471 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11473 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
11475 2017-05-26 Martin Liska <mliska@suse.cz>
11477 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
11479 2017-05-26 Martin Liska <mliska@suse.cz>
11481 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
11482 always leading ';; '.
11483 (dump_bb_info): Likewise.
11484 (brief_dump_cfg): Likewise.
11485 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
11486 * dumpfile.c: Remove usage of TDF_VERBOSE.
11487 * dumpfile.h (enum dump_kind): Likewise.
11488 (dump_gimple_bb_header): Do not use TDF_COMMENT.
11489 * print-tree.c (debug_verbose): Remove.
11490 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
11491 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
11492 * tree-diagnostic.c (default_tree_printer): Replace
11493 TDF_DIAGNOSTIC with TDF_SLIM.
11495 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11497 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
11498 in parameter loop, rather than loop_vinfo.
11499 (create_intersect_range_checks): Ditto.
11500 (vect_create_cond_for_alias_checks): Update call to above functions.
11502 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11504 PR tree-optimization/80815
11505 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
11506 for merging runtime alias checks. Handle negative DR_STEPs.
11508 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11510 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
11512 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
11513 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
11514 out code pruning runtime alias checks.
11515 * tree-data-ref.c (prune_runtime_alias_test_list): New function
11516 factored out from above.
11517 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
11519 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
11521 (prune_runtime_alias_test_list): New decalaration.
11523 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11525 * tree-vect-data-refs.c (compare_tree): Rename and move ...
11526 * tree-data-ref.c (data_ref_compare_tree): ... to here.
11527 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
11528 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
11529 (operator==, comp_dr_with_seg_len_pair): Ditto.
11530 (vect_prune_runtime_alias_test_list): Ditto.
11532 2017-05-26 Martin Liska <mliska@suse.cz>
11535 * params.def: Bound partial-inlining-entry-probability param.
11537 2017-05-26 Marek Polacek <polacek@redhat.com>
11540 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
11543 2017-05-26 Richard Biener <rguenther@suse.de>
11545 PR tree-optimization/80842
11546 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
11549 2017-05-26 Richard Biener <rguenther@suse.de>
11551 PR tree-optimization/80844
11552 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
11554 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
11556 * doc/md.texi (Machine Constraints): Update x86 family
11557 machine constraints section to match 'config/i386/constraints.md'.
11559 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
11561 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
11563 2017-05-25 Nathan Sidwell <nathan@acm.org>
11565 * doc/invoke.texi (--enable-languages): Update documentation.
11567 2017-05-25 Martin Liska <mliska@suse.cz>
11569 * dumpfile.c: Add TDF_FOLDING.
11570 * dumpfile.h (enum dump_kind): Likewise.
11571 * genmatch.c (dt_simplify::gen_1): Use it.
11573 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11575 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
11577 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11579 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
11580 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
11582 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11584 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
11585 * match.pd (X == C): Rewrite it here.
11586 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
11587 with_certain_nonzero_bits2): New predicates.
11588 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
11590 2017-05-24 Nathan Sidwell <nathan@acm.org>
11592 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
11595 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
11598 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
11600 * config/powerpcspe: New port. Files are copied from the rs6000
11601 port, with "rs6000" in filenames replaced by "powerpcspe".
11603 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
11605 PR rtl-optimization/80754
11606 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
11608 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
11610 * config/sparc/sparc.md (length): Return the correct value for -mflat
11611 sibcalls to match output_sibcall.
11613 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
11617 * config/rs6000/rs6000.c (struct machine_function): Add new field
11619 (rs6000_get_separate_components): Init that field, use it.
11620 (rs6000_components_for_bb): Use the field.
11622 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
11624 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
11626 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
11628 PR middle-end/80823
11629 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
11631 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11634 * config/s390/s390.c (s390_check_qrst_address): Check incoming
11635 address against address_operand predicate.
11636 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
11638 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
11640 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
11641 parameters passed indirectly.
11643 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
11645 * config/i386/i386.md (*movdi_internal): Remove SSE4
11646 alternative 18 (?r, *v). Update insn attributes.
11647 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
11648 Update insn attributes.
11649 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
11650 Update insn attributes.
11651 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
11652 alternative 1 (r, v). Remove isa attribute.
11653 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
11654 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
11655 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
11657 2017-05-23 Tom de Vries <tom@codesourcery.com>
11659 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
11662 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11664 * cgraphunit.c (symbol_table::process_new_functions): Update.
11665 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
11666 (inline_generate_summary): Rename to ...
11667 (ipa_fn_summary_generate): ... this one.
11668 (inline_read_summary): Rename to ...
11669 (ipa_fn_summary_read): ... this one.
11670 (inline_write_summary): Rename to ...
11671 (ipa_fn_summary_write): ... this one.
11672 (inline_free_summary): Rename to ...
11673 (ipa_free_fn_summary): ... this one.
11674 (pass_data_local_fn_summary, pass_local_fn_summary,
11675 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
11676 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
11677 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
11678 make_pass_ipa_fn_summary): New.
11679 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
11680 inline_write_summary, inline_free_summary): Remove.
11681 (ipa_free_fn_summary) : New.
11682 * ipa-inline.c (ipa_inline): Update.
11683 (pass_ipa_inline): Do not generate summaries.
11684 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
11686 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
11687 and add pass_ipa_fn_summary.
11688 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
11690 (make_pass_inline_parameters): Remove.
11692 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
11694 * omp-low.c (struct omp_context): Remove "default_kind" member.
11697 * omp-offload.c (execute_oacc_device_lower): Remove the
11698 parallelism dimensions function attributes for unparallelized
11699 OpenACC kernels constructs.
11701 2017-05-23 Martin Liska <mliska@suse.cz>
11703 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
11705 (cgraph_edge::make_speculative): Likewise.
11706 (cgraph_edge::resolve_speculation): Likewise.
11707 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
11708 (cgraph_node::dump): Likewise.
11709 * cgraph.h: Likewise.
11710 * cgraphunit.c (analyze_functions): Likewise.
11711 (symbol_table::compile): Likewise.
11712 * ipa-cp.c (print_all_lattices): Likewise.
11713 (determine_versionability): Likewise.
11714 (initialize_node_lattices): Likewise.
11715 (ipcp_verify_propagated_values): Likewise.
11716 (estimate_local_effects): Likewise.
11717 (update_profiling_info): Likewise.
11718 (create_specialized_node): Likewise.
11719 (perhaps_add_new_callers): Likewise.
11720 (decide_about_value): Likewise.
11721 (decide_whether_version_node): Likewise.
11722 (identify_dead_nodes): Likewise.
11723 (ipcp_store_bits_results): Likewise.
11724 * ipa-devirt.c (dump_targets): Likewise.
11725 (ipa_devirt): Likewise.
11726 * ipa-icf.c (sem_item::dump): Likewise.
11727 (sem_function::equals): Likewise.
11728 (sem_variable::equals): Likewise.
11729 (sem_item_optimizer::read_section): Likewise.
11730 (sem_item_optimizer::execute): Likewise.
11731 (congruence_class::dump): Likewise.
11732 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
11733 (dump_inline_summary): Likewise.
11734 (estimate_node_size_and_time): Likewise.
11735 (inline_analyze_function): Likewise.
11736 * ipa-inline-transform.c (inline_call): Likewise.
11737 * ipa-inline.c (report_inline_failed_reason): Likewise.
11738 (want_early_inline_function_p): Likewise.
11739 (edge_badness): Likewise.
11740 (update_edge_key): Likewise.
11741 (inline_small_functions): Likewise.
11742 * ipa-profile.c (ipa_profile): Likewise.
11743 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
11744 (ipa_make_edge_direct_to_target): Likewise.
11745 (remove_described_reference): Likewise.
11746 (ipa_impossible_devirt_target): Likewise.
11747 (propagate_controlled_uses): Likewise.
11748 (ipa_print_node_params): Likewise.
11749 (ipcp_transform_function): Likewise.
11750 * ipa-pure-const.c (pure_const_read_summary): Likewise.
11751 (propagate_pure_const): Likewise.
11752 * ipa-reference.c (generate_summary): Likewise.
11753 (read_write_all_from_decl): Likewise.
11754 (propagate): Likewise.
11755 (ipa_reference_read_optimization_summary): Likewise.
11756 * ipa-utils.c (ipa_merge_profiles): Likewise.
11757 * ipa.c (walk_polymorphic_call_targets): Likewise.
11758 (symbol_table::remove_unreachable_nodes): Likewise.
11759 (ipa_single_use): Likewise.
11760 * passes.c (execute_todo): Likewise.
11761 * predict.c (drop_profile): Likewise.
11762 * symtab.c (symtab_node::get_dump_name): New function.
11763 (symtab_node::dump_name): Likewise.
11764 (symtab_node::dump_asm_name): Likewise.
11765 (symtab_node::dump_references): Likewise.
11766 (symtab_node::dump_referring): Likewise.
11767 (symtab_node::dump_base): Likewise.
11768 (symtab_node::debug_symtab): Likewise.
11769 * tree-sra.c (convert_callers_for_node): Likewise.
11770 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
11771 * value-prof.c (init_node_map): Likewise.
11773 2017-05-23 Martin Liska <mliska@suse.cz>
11775 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
11776 and symtab_node::debug_symtab to symbol_table::debug.
11777 * cgraphunit.c (analyze_functions): Use the renamed function.
11778 (symbol_table::compile): Likewise.
11779 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
11780 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
11781 * passes.c (execute_todo): Likewise.
11782 * symtab.c (symbol_table::dump): New function.
11783 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
11785 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11787 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
11788 that nonconst implies exec.
11790 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11792 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
11793 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
11794 (inline_edge_summary_vec): Turn into ...
11795 (ipa_call_summaries): ... this one.
11796 (redirect_to_unreachable, edge_set_predicate,
11797 evaluate_properties_for_edge, inline_summary_alloc,
11798 reset_ipa_call_summary, reset_inline_summary,
11799 inline_summary_t::duplicate): Update.
11800 (inline_edge_duplication_hook): Turn to ...
11801 (ipa_call_summary_t::duplicate): ... this one.
11802 (inline_edge_removal_hook): Turn to ...
11803 (ipa_call_summary_t::remove): ... this one.
11804 (dump_inline_edge_summary): Turn to ...
11805 (dump_ipa_call_summary): ... this one.
11806 (estimate_function_body_sizes): Update.
11807 (inline_update_callee_summaries): Update.
11808 (remap_edge_change_prob): Update.
11809 (remap_edge_summaries): Update.
11810 (inline_merge_summary): Update.
11811 (do_estimate_edge_time): Update.
11812 (inline_generate_summary): Update.
11813 (inline_read_section): Update.
11814 (inline_read_summary): Update.
11815 (inline_free_summary): Update.
11816 * ipa-inline.c (can_inline_edge_p): Update.
11817 (compute_inlined_call_time): Update.
11818 (want_inline_small_function_p): Update.
11819 (edge_badness): Update.
11820 (early_inliner): Update.
11821 * ipa-inline.h (inline_edge_summary): Turn to ...
11822 (ipa_call_summary): ... this one.
11823 (ipa_call_summary_t): New class.
11824 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
11825 (ipa_call_summaries): New.
11826 (inline_edge_summary): Remove.
11827 (estimate_edge_growth): Update.
11828 * ipa-profile.c (ipa_propagate_frequency_1): Update.
11829 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11830 * ipa-split.c (execute_split_functions): Update.
11831 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
11833 2017-05-23 Tom de Vries <tom@codesourcery.com>
11835 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
11836 attributes): Document rdrand effective target.
11838 2017-05-23 Tom de Vries <tom@codesourcery.com>
11840 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
11841 attributes): Sort alphabetically.
11843 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
11845 * config/avr/genmultilib.awk: Use gsub instead of gensub.
11847 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
11850 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
11851 V2DF/V2DI splat into two separate patterns, one that handles
11852 registers, and the other that only handles memory. Drop support
11853 for splatting from a GPR on ISA 2.07 and then splitting the
11854 splat into direct move and splat.
11855 (vsx_splat_<mode>_reg): Likewise.
11856 (vsx_splat_<mode>_mem): Likewise.
11858 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
11860 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
11862 2017-05-22 Jakub Jelinek <jakub@redhat.com>
11864 PR middle-end/80809
11865 * omp-low.c (finish_taskreg_remap): New function.
11866 (finish_taskreg_scan): If unit size of ctx->record_type
11867 is non-constant, unshare the size expression and replace
11868 decls in it with possible outer var refs.
11870 PR middle-end/80809
11871 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
11872 GOVD_SHARED rather than GOVD_PRIVATE with it.
11873 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
11874 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
11876 PR middle-end/80853
11877 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
11878 as last argument to build_outer_var_ref for pointer bases of array
11879 section reductions.
11881 2017-05-19 Martin Sebor <msebor@redhat.com>
11883 * print-tree.c (print_node): Print DECL_READ_P flag.
11885 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11887 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
11888 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
11889 * cgraph.c: Likewise.
11890 * cgraphunit.c: Likewise.
11891 * gengtype.c: Likewise.
11892 * ipa-cp.c: Likewise.
11893 * ipa-devirt.c: Likewise.
11894 * ipa-icf.c: Likewise.
11895 * ipa-predicate.c: Likewise.
11896 * ipa-profile.c: Likewise.
11897 * ipa-prop.c: Likewise.
11898 * ipa-split.c: Likewise.
11900 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
11901 edge_predicate_pool, dump_inline_hints,
11902 inline_summary::account_size_time, redirect_to_unreachable,
11903 edge_set_predicate, set_hint_predicate,
11904 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
11905 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
11906 inline_summary_t::remove, remap_hint_predicate_after_duplication,
11907 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
11908 ipa_call_summary_t::remove, initialize_growth_caches,
11909 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
11910 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
11911 mark_modified, unmodified_parm_1, unmodified_parm,
11912 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
11913 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
11914 compute_bb_predicates, will_be_nonconstant_expr_predicate,
11915 will_be_nonconstant_predicate, record_modified_bb_info,
11916 get_minimal_bb, record_modified, param_change_prob,
11917 phi_result_unknown_predicate, predicate_for_phi_result,
11918 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
11919 estimate_function_body_sizes, compute_inline_parameters,
11920 compute_inline_parameters_for_curren, pass_data_inline_parameters,
11921 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
11922 inline_update_callee_summaries, remap_edge_change_prob,
11923 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
11924 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
11925 inline_analyze_function, inline_summary_t::insert,
11926 inline_generate_summary, read_ipa_call_summary, inline_read_section,
11927 inline_read_summary, write_ipa_call_summary, inline_write_summary,
11928 inline_free_summary): Move to ipa-fnsummary.h
11929 (predicate_t): Remove.
11930 * ipa-fnsummary.c: New file.
11931 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
11932 (enum inline_hints_vals, inline_hints, agg_position_info,
11933 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
11934 inline_summaries, ipa_call_summary, ipa_call_summary_t,
11935 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
11936 dump_inline_summary, dump_inline_hints, inline_generate_summary,
11937 inline_read_summary, inline_write_summary, inline_free_summary,
11938 inline_analyze_function, initialize_inline_failed,
11939 inline_merge_summary, inline_update_overall_summary,
11940 compute_inline_parameters): Move to ipa-fnsummary.h
11941 * ipa-fnsummary.h: New file.
11942 * ipa-inline-transform.h: Include ipa-inline.h.
11943 * ipa-inline.c: LIkewise.
11945 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11947 * ipa-inline.c (edge_badness): Use inlined_time instead of
11948 inline_summaries->get.
11950 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11952 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
11954 2017-05-22 Nathan Sidwell <nathan@acm.org>
11956 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
11957 (fdump-lang): Document 'raw' option.
11958 * dumpfile.h (TDI_tu): Delete.
11959 * dumpfile.c (dump_files): Remove translation-unit.
11960 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
11962 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
11964 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
11965 command option from $(AWK) call.
11966 * config/avr/genmultilib.awk: Simplify and rewrite so that it
11967 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
11968 [FORMAT]: Remove handling of variable.
11969 * config/avr/t-multilib: Regenerate.
11971 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11973 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
11975 (dump_inline_summary): Do not print self_time.
11976 (estimate_function_body_sizes): Do not set self_time.
11977 (compute_inline_parameters): Likewise.
11978 (inline_read_section, inline_write_summary): Do not stream self_time.
11979 * ipa-inline.h (inline_summary): Drop self_time.
11981 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11983 * ipa-inline-analysis.c (account_size_time): Rename to ...
11984 (inline_summary::account_size_time): ... this one.
11985 (reset_ipa_call_summary): Turn to ...
11986 (ipa_call_summary::reset): ... this one.
11987 (reset_inline_summary): Turn to ...
11988 (inline_summary::reset): ... this one.
11989 (inline_summary_t::remove): Update.
11990 (inline_summary_t::duplicate): Update.
11991 (ipa_call_summary_t::remove): Update.
11992 (dump_inline_summary): Update.
11993 (estimate_function_body_sizes): Update.
11994 (compute_inline_parameters): Update.
11995 (estimate_node_size_and_time): Update.
11996 (inline_merge_summary): Update.
11997 (inline_update_overall_summary): Update.
11998 (inline_read_section): Update.
11999 (inline_write_summary): Update.
12000 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
12001 add account_size_time and reset member functions.
12002 (ipa_call_summary): Add reset function.
12003 * ipa-predicate.h (predicate::operator &): Constify.
12005 2017-05-22 Richard Biener <rguenther@suse.de>
12007 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
12009 2017-05-19 Jason Merrill <jason@redhat.com>
12011 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
12013 2017-05-19 Marek Polacek <polacek@redhat.com>
12016 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
12017 TYPE_OVERFLOW_WRAPS checks.
12019 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
12021 * tree-core.h (enum omp_clause_default_kind): Add
12022 "OMP_CLAUSE_DEFAULT_PRESENT".
12023 * tree-pretty-print.c (dump_omp_clause): Handle it.
12024 * gimplify.c (enum gimplify_omp_var_data): Add
12025 "GOVD_MAP_FORCE_PRESENT".
12026 (gimplify_adjust_omp_clauses_1): Map it to
12027 "GOMP_MAP_FORCE_PRESENT".
12028 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
12030 * gimplify.c (oacc_default_clause): Clarify.
12032 2017-05-19 Nathan Sidwell <nathan@acm.org>
12034 LANG_HOOK_REGISTER_DUMPS
12035 * toplev.c (general_init): Call register dump lang hook.
12036 * doc/invoke.texi: Document -fdump-lang option family.
12037 * dumpfile.c (dump_files): Remove class dump here.
12038 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
12039 * dumpfile.h (tree_dump_index): Remove TDI_class.
12040 * langhooks-def.h (lhd_register_dumps): Declare.
12041 (LANG_HOOKS_REGISTER_DUMPS): Define.
12042 (LANG_HOOKS_INITIALIZER): Add it.
12043 * langhooks.c (lhd_register_dumps): Define.
12044 * langhooks.h (struct lang_hooks): Add register_dumps.
12046 2017-05-19 Nathan Sidwell <nathan@acm.org>
12048 * context.h (context::set_passes): New.
12049 * context.c (context::context): Do not create pass manager.
12050 * toplev.c (general_init): Create pass manager here.
12052 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
12054 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
12055 use this splitter if two add or or instructions would also work for
12056 the constant we want to generate.
12058 2017-05-19 Richard Biener <rguenther@suse.de>
12061 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
12062 predicate evaluation.
12064 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12066 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
12068 * ipa-inline.c (want_inline_small_function_p): Do not cast to
12071 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12073 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
12074 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
12075 (inline_edge_summary_vec): Turn into ...
12076 (ipa_call_summaries): ... this one.
12077 (redirect_to_unreachable, edge_set_predicate,
12078 evaluate_properties_for_edge, inline_summary_alloc,
12079 reset_ipa_call_summary, reset_inline_summary,
12080 inline_summary_t::duplicate): Update.
12081 (inline_edge_duplication_hook): Turn to ...
12082 (ipa_call_summary_t::duplicate): ... this one.
12083 (inline_edge_removal_hook): Turn to ...
12084 (ipa_call_summary_t::remove): ... this one.
12085 (dump_inline_edge_summary): Turn to ...
12086 (dump_ipa_call_summary): ... this one.
12087 (estimate_function_body_sizes): Update.
12088 (inline_update_callee_summaries): Update.
12089 (remap_edge_change_prob): Update.
12090 (remap_edge_summaries): Update.
12091 (inline_merge_summary): Update.
12092 (do_estimate_edge_time): Update.
12093 (inline_generate_summary): Update.
12094 (inline_read_section): Update.
12095 (inline_read_summary): Update.
12096 (inline_free_summary): Update.
12097 * ipa-inline.c (can_inline_edge_p): Update.
12098 (compute_inlined_call_time): Update.
12099 (want_inline_small_function_p): Update.
12100 (edge_badness): Update.
12101 (early_inliner): Update.
12102 * ipa-inline.h (inline_edge_summary): Turn to ...
12103 (ipa_call_summary): ... this one.
12104 (ipa_call_summary_t): New class.
12105 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
12106 (ipa_call_summaries): New.
12107 (inline_edge_summary): Remove.
12108 (estimate_edge_growth): Update.
12109 * ipa-profile.c (ipa_propagate_frequency_1): Update.
12110 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12111 * ipa-split.c (execute_split_functions): Update.
12112 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
12114 2017-05-19 Richard Biener <rguenther@suse.de>
12116 PR middle-end/80764
12117 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
12119 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
12121 * config/rs6000/rs6000.c (struct machine_function): Add field
12122 fpr_is_wrapped_separately.
12123 (rs6000_get_separate_components): Use 64 components. Handle the
12124 new FPR components.
12125 (rs6000_components_for_bb): Handle the FPR components.
12126 (rs6000_emit_prologue_components): Handle the FPR components.
12127 (rs6000_emit_epilogue_components): Handle the FPR components.
12128 (rs6000_set_handled_components): Handle the FPR components.
12129 (rs6000_emit_prologue): Don't output prologue code for those FPRs
12130 that are already separately shrink-wrapped.
12131 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
12132 that are already separately shrink-wrapped.
12134 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
12137 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
12140 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
12141 (define_peephole2 for Altivec d-form load): Add peepholes to catch
12142 cases where the register allocator uses a move and an offsettable
12143 memory operation to/from a FPR register on ISA 2.06/2.07.
12144 (define_peephole2 for Altivec d-form store): Likewise.
12146 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
12149 * config/i386/mmx.md (*mov<mode>_internal): Enable
12150 alternatives 11, 12, 13 and 14 also for 32bit targets.
12151 Remove alternatives 15, 16, 17 and 18.
12152 * config/i386/sse.md (vec_concatv2di): Change
12153 alternative (!x, *y) to (x, ?!*Yn).
12155 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
12157 * dumpfile.h (enum dump_kind): Remove stray comma.
12159 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
12161 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
12162 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
12163 predicate::num_conditions
12164 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
12165 (CHANGED): turn into predicate::changed.
12166 (agg_position_info): Move to ipa-predicate.h
12167 (add_condition, predicate::add_clause, predicate::operator &=,
12168 predicate::or_with, predicate::evaluate, predicate::probability,
12169 dump_condition, dump_clause, predicate::dump,
12170 predicate::remap_after_duplication, predicate::remap_after_inlining,
12171 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
12172 (evaluate_conditions_for_known_args): Update.
12173 (set_cond_stmt_execution_predicate): Update.
12174 * ipa-inline.h: Include ipa-predicate.h
12175 (condition, inline_param_summary, conditions, agg_position_info,
12176 predicate): Move to ipa-predicate.h
12177 * ipa-predicate.c: New file.
12178 * ipa-predicate.h: New file.
12180 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
12182 * final.c (leaf_function_p): Check we are not in a sequence.
12184 2017-05-18 Martin Liska <mliska@suse.cz>
12186 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
12187 * dumpfile.c (dump_register): Use new enum dump_kind.
12188 (get_dump_file_name): Likewise.
12189 (dump_enable_all): Likewise.
12190 (dump_switch_p_1): Likewise.
12191 (enable_rtl_dump_file): Remove usage of TDF_RTL.
12192 * dumpfile.h (enum dump_kind): New enum type.
12193 (struct dump_file_info): Create constructor and
12194 format fields and comments.
12195 * passes.c (pass_manager::register_one_dump_file):
12197 * statistics.c (statistics_early_init): Likewise.
12198 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
12199 TDF_TREE with TDF_SLIM.
12200 (gather_memory_references_ref): Likewise.
12202 2017-05-18 Martin Liska <mliska@suse.cz>
12204 * vec.h (struct vnull): Use it.
12206 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
12208 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
12209 (true_predicate, false_predicate, true_predicate_p,
12210 false_predicate_p): Remove.
12211 (single_cond_predicate, not_inlined_predicate): Turn to member function
12213 (add_condition): Update.
12214 (add_clause): Turn to...
12215 (predicate::add_clause): ... this one; update; allow passing NULL
12217 (and_predicates): Turn to ...
12218 (predicate::operator &=): ... this one.
12219 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
12220 (or_predicates): Turn to ...
12221 (predicate::or_with): ... this one.
12222 (evaluate_predicate): Turn to ...
12223 (predicate::evaluate): ... this one.
12224 (predicate_probability): Turn to ...
12225 (predicate::probability): ... this one.
12226 (dump_condition): Update.
12227 (dump_predicate): Turn to ...
12228 (predicate::dump): ... this one.
12229 (account_size_time): Update.
12230 (edge_set_predicate): Update.
12231 (set_hint_predicate): UPdate.
12232 (evaluate_conditions_for_known_args): Update.
12233 (evaluate_properties_for_edge): Update.
12234 (remap_predicate_after_duplication): Turn to...
12235 (predicate::remap_after_duplication): ... this one.
12236 (remap_hint_predicate_after_duplication): Update.
12237 (inline_summary_t::duplicate): UPdate.
12238 (dump_inline_edge_summary): Update.
12239 (dump_inline_summary): Update.
12240 (set_cond_stmt_execution_predicate): Update.
12241 (set_switch_stmt_execution_predicate): Update.
12242 (compute_bb_predicates): Update.
12243 (will_be_nonconstant_expr_predicate): Update.
12244 (will_be_nonconstant_predicate): Update.
12245 (phi_result_unknown_predicate): Update.
12246 (predicate_for_phi_result): Update.
12247 (array_index_predicate): Update.
12248 (estimate_function_body_sizes): Update.
12249 (estimate_node_size_and_time): Update.
12250 (estimate_ipcp_clone_size_and_time): Update.
12251 (remap_predicate): Rename to ...
12252 (predicate::remap_after_inlining): ... this one.
12253 (remap_hint_predicate): Update.
12254 (inline_merge_summary): Update.
12255 (inline_update_overall_summary): Update.
12256 (estimate_size_after_inlining): Update.
12257 (read_predicate): Rename to ...
12258 (predicate::stream_in): ... this one.
12259 (read_inline_edge_summary): Update.
12260 (write_predicate): Rename to ...
12261 (predicate::stream_out): ... this one.
12262 (write_inline_edge_summary): Update.
12263 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
12264 (clause_t): Turn to uint32_t
12265 (predicate): Turn to class; implement constructor and operators
12267 (size_time_entry): Update.
12268 (inline_summary): Update.
12269 (inline_edge_summary): Update.
12271 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
12273 * fold-const.c (fold_binary_loc): Move transformation...
12274 * match.pd (C - X CMP X): ... here.
12276 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
12278 * config/sparc/sparc.c (sparc_option_override): Set function
12279 alignment for -mcpu=niagara7 to 64 to match the I$ line.
12280 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
12282 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
12284 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
12286 2017-05-18 Marek Polacek <polacek@redhat.com>
12289 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
12290 (pass_ubsan::execute): Call gimple_assign_single_p instead of
12291 gimple_assign_load_p.
12293 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
12295 PR middle-end/80692
12296 * real.c (do_compare): Give decimal_do_compare preference over
12297 comparing just the signs.
12299 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
12301 * doc/md.texi (Canonicalization of Instructions): Describe the
12302 canonical form of instructions that inherently set a condition
12305 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
12307 PR middle-end/80775
12308 * tree-cfg.c: Move deletion of unreachable case statements to after
12309 the merging of consecutive case labels.
12311 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
12313 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
12314 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
12315 restoring of callee-saved registers.
12317 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
12319 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
12320 * config/visium/visium.c (single_set_and_flags): Likewise.
12321 * config/visium/visium.md (Substitutions): Likewise.
12323 2017-05-17 Martin Liska <mliska@suse.cz>
12325 * cfg.c: Introduce dump_flags_t type and
12326 use it instead of int type.
12328 * cfghooks.c: Likewise.
12329 * cfghooks.h (struct cfg_hooks): Likewise.
12330 * cfgrtl.c: Likewise.
12331 * cfgrtl.h: Likewise.
12332 * cgraph.c (cgraph_node::get_body): Likewise.
12333 * coretypes.h: Likewise.
12334 * domwalk.c: Likewise.
12335 * domwalk.h: Likewise.
12336 * dumpfile.c (struct dump_option_value_info): Likewise.
12337 (dump_enable_all): Likewise.
12338 (dump_switch_p_1): Likewise.
12339 (opt_info_switch_p): Likewise.
12340 * dumpfile.h (enum tree_dump_index): Likewise.
12341 (struct dump_file_info): Likewise.
12342 * genemit.c: Likewise.
12343 * generic-match-head.c: Likewise.
12344 * gengtype.c (open_base_files): Likewise.
12345 * gimple-pretty-print.c: Likewise.
12346 * gimple-pretty-print.h: Likewise.
12347 * graph.c (print_graph_cfg): Likewise.
12348 * graphite-scop-detection.c (dot_all_sese): Likewise.
12349 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
12350 * loop-unroll.c (report_unroll): Likewise.
12351 * passes.c (pass_manager::register_one_dump_file): Likewise.
12352 * print-tree.c: Likewise.
12353 * statistics.c: Likewise.
12354 * tree-cfg.c: Likewise.
12355 * tree-cfg.h: Likewise.
12356 * tree-dfa.c: Likewise.
12357 * tree-dfa.h: Likewise.
12358 * tree-dump.c (dump_function): Likewise.
12359 * tree-dump.h (struct dump_info): Likewise.
12360 * tree-pretty-print.c: Likewise.
12361 * tree-pretty-print.h: Likewise.
12362 * tree-ssa-live.c: Likewise.
12363 * tree-ssa-live.h: Likewise.
12364 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
12365 * tree-vect-loop.c: Likewise.
12366 * tree-vect-slp.c: Likewise.
12368 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
12369 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12371 PR tree-optimization/80457
12372 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
12373 of all arguments to a statement as scalar_to_vec operations.
12374 (vectorizable_call): Adjust call to vect_model_simple_cost for
12376 (vectorizable_conversion): Likewise.
12377 (vectorizable_assignment): Likewise.
12378 (vectorizable_shift): Likewise.
12379 (vectorizable_operation): Likewise.
12380 (vectorizable_comparison): Likewise.
12381 (vect_is_simple_cond): Record the def types for operands.
12382 (vectorizable_condition): Likewise, call vect_model_simple_cost.
12383 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
12384 for statement argument count.
12386 2017-05-16 Carl Love <cel@us.ibm.com>
12388 * config/rs6000/rs6000-c: Add support for built-in functions
12389 vector unsigned long long vec_bperm (vector unsigned long long,
12390 vector unsigned char)
12391 vector signed long long vec_mule (vector signed int,
12393 vector unsigned long long vec_mule (vector unsigned int,
12394 vector unsigned int)
12395 vector signed long long vec_mulo (vector signed int,
12397 vector unsigned long long vec_mulo (vector unsigned int,
12398 vector unsigned int)
12399 vector signed char vec_sldw (vector signed char,
12400 vector signed char,
12402 vector unsigned char vec_sldw (vector unsigned char,
12403 vector unsigned char,
12405 vector signed short vec_sldw (vector signed short,
12406 vector signed short,
12408 vector unsigned short vec_sldw (vector unsigned short,
12409 vector unsigned short,
12411 vector signed int vec_sldw (vector signed int,
12414 vector unsigned int vec_sldw (vector unsigned int,
12415 vector unsigned int,
12417 vector signed long long vec_sldw (vector signed long long,
12418 vector signed long long,
12420 vector unsigned long long vec_sldw (vector unsigned long long,
12421 vector unsigned long long,
12423 * config/rs6000/rs6000-c: Add support for built-in functions
12424 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
12425 * config/rs6000/altivec.h: Add defintion for vec_sldw.
12426 * doc/extend.texi: Update the built-in documentation for the
12427 new built-in functions.
12429 2017-05-16 Marek Polacek <polacek@redhat.com>
12433 * tree.c (save_expr): Don't fold the expression.
12435 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
12437 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
12438 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
12439 and (?*y,m). Update insn attributes.
12441 2017-05-16 Martin Liska <mliska@suse.cz>
12443 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
12444 flags argument of print_gimple_stmt, print_gimple_expr,
12445 print_generic_stmt and print_generic_expr.
12446 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
12447 * coretypes.h: Likewise.
12448 * except.c (dump_eh_tree): Likewise.
12449 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12450 * gimple-pretty-print.h: Likewise.
12451 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
12452 (backprop::push_to_worklist): Likewise.
12453 (backprop::pop_from_worklist): Likewise.
12454 (backprop::process_use): Likewise.
12455 (backprop::intersect_uses): Likewise.
12456 (note_replacement): Likewise.
12457 * gimple-ssa-store-merging.c
12458 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
12459 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
12460 (pass_store_merging::execute): Likewise.
12461 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
12462 (ssa_base_cand_dump_callback): Likewise.
12463 (dump_incr_vec): Likewise.
12464 (replace_refs): Likewise.
12465 (replace_mult_candidate): Likewise.
12466 (create_add_on_incoming_edge): Likewise.
12467 (create_phi_basis): Likewise.
12468 (insert_initializers): Likewise.
12469 (all_phi_incrs_profitable): Likewise.
12470 (introduce_cast_before_cand): Likewise.
12471 (replace_one_candidate): Likewise.
12472 * gimplify.c (gimplify_expr): Likewise.
12473 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
12474 (set_rename): Likewise.
12475 (rename_uses): Likewise.
12476 (copy_loop_phi_nodes): Likewise.
12477 (add_close_phis_to_merge_points): Likewise.
12478 (copy_loop_close_phi_args): Likewise.
12479 (copy_cond_phi_args): Likewise.
12480 (graphite_copy_stmts_from_block): Likewise.
12481 (translate_pending_phi_nodes): Likewise.
12482 * graphite-poly.c (print_pdr): Likewise.
12483 (dump_gbb_cases): Likewise.
12484 (dump_gbb_conditions): Likewise.
12485 (print_scop_params): Likewise.
12486 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
12487 (build_cross_bb_scalars_use): Likewise.
12488 (gather_bbs::before_dom_children): Likewise.
12489 * hsa-dump.c (dump_hsa_immed): Likewise.
12490 * ipa-cp.c (print_ipcp_constant_value): Likewise.
12491 (get_replacement_map): Likewise.
12492 * ipa-inline-analysis.c (dump_condition): Likewise.
12493 (estimate_function_body_sizes): Likewise.
12494 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
12495 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
12496 * ipa-prop.c (ipa_dump_param): Likewise.
12497 (ipa_print_node_jump_functions_for_edge): Likewise.
12498 (ipa_modify_call_arguments): Likewise.
12499 (ipa_modify_expr): Likewise.
12500 (ipa_dump_param_adjustments): Likewise.
12501 (ipa_dump_agg_replacement_values): Likewise.
12502 (ipcp_modif_dom_walker::before_dom_children): Likewise.
12503 * ipa-pure-const.c (check_stmt): Likewise.
12504 (pass_nothrow::execute): Likewise.
12505 * ipa-split.c (execute_split_functions): Likewise.
12506 * omp-offload.c (dump_oacc_loop_part): Likewise.
12507 (dump_oacc_loop): Likewise.
12508 * trans-mem.c (tm_log_emit): Likewise.
12509 (tm_memopt_accumulate_memops): Likewise.
12510 (dump_tm_memopt_set): Likewise.
12511 (dump_tm_memopt_transform): Likewise.
12512 * tree-cfg.c (gimple_verify_flow_info): Likewise.
12513 (print_loop): Likewise.
12514 * tree-chkp-opt.c (chkp_print_addr): Likewise.
12515 (chkp_gather_checks_info): Likewise.
12516 (chkp_get_check_result): Likewise.
12517 (chkp_remove_check_if_pass): Likewise.
12518 (chkp_use_outer_bounds_if_possible): Likewise.
12519 (chkp_reduce_bounds_lifetime): Likewise.
12520 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
12521 (chkp_mark_completed_bounds): Likewise.
12522 (chkp_register_incomplete_bounds): Likewise.
12523 (chkp_mark_invalid_bounds): Likewise.
12524 (chkp_maybe_copy_and_register_bounds): Likewise.
12525 (chkp_build_returned_bound): Likewise.
12526 (chkp_get_bound_for_parm): Likewise.
12527 (chkp_build_bndldx): Likewise.
12528 (chkp_get_bounds_by_definition): Likewise.
12529 (chkp_generate_extern_var_bounds): Likewise.
12530 (chkp_get_bounds_for_decl_addr): Likewise.
12531 * tree-chrec.c (chrec_apply): Likewise.
12532 * tree-data-ref.c (dump_data_reference): Likewise.
12533 (dump_subscript): Likewise.
12534 (dump_data_dependence_relation): Likewise.
12535 (analyze_overlapping_iterations): Likewise.
12536 * tree-inline.c (expand_call_inline): Likewise.
12537 (tree_function_versioning): Likewise.
12538 * tree-into-ssa.c (dump_defs_stack): Likewise.
12539 (dump_currdefs): Likewise.
12540 (dump_names_replaced_by): Likewise.
12541 (dump_update_ssa): Likewise.
12542 (update_ssa): Likewise.
12543 * tree-object-size.c (pass_object_sizes::execute): Likewise.
12544 * tree-parloops.c (build_new_reduction): Likewise.
12545 (try_create_reduction_list): Likewise.
12546 (ref_conflicts_with_region): Likewise.
12547 (oacc_entry_exit_ok_1): Likewise.
12548 (oacc_entry_exit_single_gang): Likewise.
12549 * tree-pretty-print.h: Likewise.
12550 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
12551 (get_scalar_evolution): Likewise.
12552 (add_to_evolution): Likewise.
12553 (get_loop_exit_condition): Likewise.
12554 (analyze_evolution_in_loop): Likewise.
12555 (analyze_initial_condition): Likewise.
12556 (analyze_scalar_evolution): Likewise.
12557 (instantiate_scev): Likewise.
12558 (number_of_latch_executions): Likewise.
12559 (gather_chrec_stats): Likewise.
12560 (final_value_replacement_loop): Likewise.
12561 (scev_const_prop): Likewise.
12562 * tree-sra.c (dump_access): Likewise.
12563 (disqualify_candidate): Likewise.
12564 (create_access): Likewise.
12565 (reject): Likewise.
12566 (maybe_add_sra_candidate): Likewise.
12567 (create_access_replacement): Likewise.
12568 (analyze_access_subtree): Likewise.
12569 (analyze_all_variable_accesses): Likewise.
12570 (sra_modify_assign): Likewise.
12571 (initialize_constant_pool_replacements): Likewise.
12572 (find_param_candidates): Likewise.
12573 (decide_one_param_reduction): Likewise.
12574 (replace_removed_params_ssa_names): Likewise.
12575 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
12576 * tree-ssa-copy.c (dump_copy_of): Likewise.
12577 (copy_prop_visit_cond_stmt): Likewise.
12578 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
12579 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
12580 (record_equivalences_from_stmt): Likewise.
12581 * tree-ssa-dse.c (compute_trims): Likewise.
12582 (delete_dead_call): Likewise.
12583 (delete_dead_assignment): Likewise.
12584 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
12585 (forward_propagate_into_cond): Likewise.
12586 (pass_forwprop::execute): Likewise.
12587 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
12588 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
12590 (move_computations_worker): Likewise.
12591 (execute_sm): Likewise.
12592 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
12593 (remove_exits_and_undefined_stmts): Likewise.
12594 (remove_redundant_iv_tests): Likewise.
12595 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
12596 (adjust_iv_update_pos): Likewise.
12597 * tree-ssa-math-opts.c (bswap_replace): Likewise.
12598 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
12599 (value_replacement): Likewise.
12600 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
12601 * tree-ssa-pre.c (print_pre_expr): Likewise.
12602 (get_representative_for): Likewise.
12603 (create_expression_by_pieces): Likewise.
12604 (insert_into_preds_of_block): Likewise.
12605 (eliminate_insert): Likewise.
12606 (eliminate_dom_walker::before_dom_children): Likewise.
12607 (eliminate): Likewise.
12608 (remove_dead_inserted_code): Likewise.
12609 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
12610 * tree-ssa-reassoc.c (get_rank): Likewise.
12611 (eliminate_duplicate_pair): Likewise.
12612 (eliminate_plus_minus_pair): Likewise.
12613 (eliminate_not_pairs): Likewise.
12614 (undistribute_ops_list): Likewise.
12615 (eliminate_redundant_comparison): Likewise.
12616 (update_range_test): Likewise.
12617 (optimize_range_tests_var_bound): Likewise.
12618 (optimize_vec_cond_expr): Likewise.
12619 (rewrite_expr_tree): Likewise.
12620 (rewrite_expr_tree_parallel): Likewise.
12621 (linearize_expr): Likewise.
12622 (break_up_subtract): Likewise.
12623 (linearize_expr_tree): Likewise.
12624 (attempt_builtin_powi): Likewise.
12625 (attempt_builtin_copysign): Likewise.
12626 (transform_stmt_to_copy): Likewise.
12627 (transform_stmt_to_multiply): Likewise.
12628 (dump_ops_vector): Likewise.
12629 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
12630 (print_scc): Likewise.
12631 (set_ssa_val_to): Likewise.
12632 (visit_reference_op_store): Likewise.
12633 (visit_use): Likewise.
12634 (sccvn_dom_walker::before_dom_children): Likewise.
12635 (run_scc_vn): Likewise.
12636 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
12638 (expr_hash_elt::print): Likewise.
12639 (const_and_copies::pop_to_marker): Likewise.
12640 (const_and_copies::record_const_or_copy_raw): Likewise.
12641 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
12642 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
12643 (dump_predicates): Likewise.
12644 (find_uninit_use): Likewise.
12645 (warn_uninitialized_phi): Likewise.
12646 (pass_late_warn_uninitialized::execute): Likewise.
12647 * tree-ssa.c (verify_vssa): Likewise.
12648 (verify_ssa): Likewise.
12649 (maybe_optimize_var): Likewise.
12650 * tree-vrp.c (dump_value_range): Likewise.
12651 (dump_all_value_ranges): Likewise.
12652 (dump_asserts_for): Likewise.
12653 (register_edge_assert_for_2): Likewise.
12654 (vrp_visit_cond_stmt): Likewise.
12655 (vrp_visit_switch_stmt): Likewise.
12656 (vrp_visit_stmt): Likewise.
12657 (vrp_visit_phi_node): Likewise.
12658 (simplify_cond_using_ranges_1): Likewise.
12659 (fold_predicate_in): Likewise.
12660 (evrp_dom_walker::before_dom_children): Likewise.
12661 (evrp_dom_walker::push_value_range): Likewise.
12662 (evrp_dom_walker::pop_value_range): Likewise.
12663 (execute_early_vrp): Likewise.
12665 2017-05-16 Richard Biener <rguenther@suse.de>
12667 * dwarf2out.c (loc_list_from_tree_1): Do not create
12668 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
12670 2017-05-16 Richard Biener <rguenther@suse.de>
12672 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
12674 (note_variable_value_in_expr): If we resolved the decl ref
12675 do not push to the stack.
12677 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
12679 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
12680 operations in fast-math mode.
12681 (vaddq_f16): Likewise.
12682 (vmul_f16): Likewise.
12683 (vmulq_f16): Likewise.
12684 (vsub_f16): Likewise.
12685 (vsubq_f16): Likewise.
12686 * config/arm/neon.md (add<mode>3): New.
12688 (fma:<VH:mode>3): New. Also remove outdated comment.
12691 2017-05-16 Martin Liska <mliska@suse.cz>
12695 * ipa-devirt.c (warn_types_mismatch): Fix typo.
12696 (odr_types_equivalent_p): Likewise.
12698 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
12700 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
12702 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
12705 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
12706 non-interunit SSE move alternatives with '?'.
12707 (zero-extendsidi peephole2): New peephole to skip intermediate
12708 general register in SSE zero-extend sequence.
12710 2017-05-15 Jeff Law <law@redhat.com>
12712 * reorg.c (relax_delay_slots): Create a new variable to hold
12713 the temporary target rather than clobbering TARGET_LABEL.
12715 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
12716 missing argument to extract_bit_field call.
12717 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
12719 2017-05-15 Martin Liska <mliska@suse.cz>
12722 * gcc.c (process_command): Do not allow empty argument of -o option.
12724 2017-05-15 Renlin Li <renlin.li@arm.com>
12726 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
12727 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
12728 * config/aarch64/constraints.md (Usf): Add long call check.
12729 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
12730 (call_value): Likewise.
12731 (sibcall): Likewise.
12732 (sibcall_value): Likewise.
12734 (call_value_insn): New.
12735 (sibcall_insn): Update rtx pattern.
12736 (sibcall_value_insn): Likewise.
12737 (call_internal): Remove.
12738 (call_value_internal): Likewise.
12739 (sibcall_internal): Likewise.
12740 (sibcall_value_internal): Likewise.
12741 (call_reg): Likewise.
12742 (call_symbol): Likewise.
12743 (call_value_reg): Likewise.
12744 (call_value_symbol): Likewise.
12746 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
12749 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
12751 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
12753 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
12754 compatible with CCGOCmode and with CCZmode.
12756 2017-05-14 Martin Sebor <msebor@redhat.com>
12758 PR middle-end/77671
12759 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
12760 (gimple_fold_builtin_snprintf): Same.
12761 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
12762 (gimple_fold_builtin_snprintf): Same.
12763 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
12764 of character types.
12765 (is_call_safe): New function.
12766 (try_substitute_return_value): Call it.
12767 (try_simplify_call): New function.
12768 (pass_sprintf_length::handle_gimple_call): Call it.
12770 2017-05-14 Martin Sebor <msebor@redhat.com>
12772 PR middle-end/80669
12773 * builtins.c (expand_builtin_stpncpy): Simplify.
12775 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
12777 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
12778 * config/i386/i386.h
12779 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
12780 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
12781 (struct machine_function): Add new members call_ms2sysv,
12782 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
12783 (struct machine_frame_state): New fields sp_realigned and
12784 sp_realigned_offset.
12785 * config/i386/i386.c
12786 (enum xlogue_stub): New enum.
12787 (enum xlogue_stub_sets): New enum.
12788 (class xlogue_layout): New class.
12789 (struct ix86_frame): New fields stack_realign_allocate_offset,
12790 stack_realign_offset and outlined_save_offset. Modify comments to
12791 detail stack layout when using out-of-line stubs.
12792 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
12793 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
12794 -mcall-ms2sysv-xlogues.
12795 (stub_managed_regs): New static variable.
12796 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
12797 registers managed by out-of-line stub.
12798 (disable_call_ms2sysv_xlogues): New function.
12799 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
12800 m->call_ms2sysv when appropriate and compute frame layout for
12802 (sp_valid_at, fp_valid_at): New inline functions.
12803 (choose_basereg): New function.
12804 (choose_baseaddr): Add align parameter, use choose_basereg and modify
12806 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
12807 Use align parameter of choose_baseaddr to generated aligned SSE movs
12809 (pro_epilogue_adjust_stack): Modify to track
12810 machine_frame_state::sp_realigned.
12811 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
12812 (ix86_nsaved_sseregs): Likewise.
12813 (ix86_emit_save_regs): Likewise.
12814 (ix86_emit_save_regs_using_mov): Likewise.
12815 (ix86_emit_save_sse_regs_using_mov): Likewise.
12816 (get_scratch_register_on_entry): Likewise.
12817 (gen_frame_set): New function.
12818 (gen_frame_load): Likewise.
12819 (gen_frame_store): Likewise.
12820 (emit_outlined_ms2sysv_save): Likewise.
12821 (emit_outlined_ms2sysv_restore): Likewise.
12822 (ix86_expand_prologue): Modify stack re-alignment code and call
12823 emit_outlined_ms2sysv_save when appropriate.
12824 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
12825 parameter rtx_insn *insn, which allows the function to be used to only
12826 generate the notes.
12827 (ix86_expand_epilogue): Modify validity checks of frame and stack
12828 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
12829 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
12830 * config/i386/predicates.md
12831 (save_multiple): New predicate.
12832 (restore_multiple): Likewise.
12833 * config/i386/sse.md
12834 (save_multiple<mode>): New pattern.
12835 (save_multiple_realign<mode>): Likewise.
12836 (restore_multiple<mode>): Likewise.
12837 (restore_multiple_and_return<mode>): Likewise.
12838 (restore_multiple_leave_return<mode>): Likewise.
12839 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
12841 2017-05-14 Julia Koval <julia.koval@intel.com>
12843 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
12844 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
12845 (__builtin_ia32_xsetbv): New builtins.
12846 * config/i386/i386.c (ix86_expand_special_args_builtin):
12848 (ix86_expand_builtin): Special expand for new intrinsics.
12849 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
12850 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
12851 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
12853 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12855 * cfganal.c (inverted_post_order_compute): Change argument type
12857 * cfganal.h (inverted_post_order_compute): Adjust prototype.
12858 * df-core.c (rest_of_handle_df_initialize): Adjust.
12859 (rest_of_handle_df_finish): Likewise.
12860 (df_analyze_1): Likewise.
12861 (df_analyze): Likewise.
12862 (loop_inverted_post_order_compute): Change argument to be a vec *.
12863 (df_analyze_loop): Adjust.
12864 (df_get_n_blocks): Likewise.
12865 (df_get_postorder): Likewise.
12866 * df.h (struct df_d): Change field to be a vec.
12867 * lcm.c (compute_laterin): Adjust.
12868 (compute_available): Likewise.
12869 * lra-lives.c (lra_create_live_ranges_1): Likewise.
12870 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
12871 * tree-ssa-pre.c (compute_antic): Likewise.
12873 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12875 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
12876 (depth_first_search::depth_first_search): Change structure init
12877 function to this constructor.
12878 (depth_first_search::add_bb): Rename function to this member.
12879 (depth_first_search::execute): Likewise.
12880 (flow_dfs_compute_reverse_finish): Adjust.
12882 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12884 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
12885 (longest_simple_path): Likewise.
12886 * shrink-wrap.c (spread_components): Likewise.
12887 (disqualify_problematic_components): Likewise.
12888 (emit_common_heads_for_components): Likewise.
12889 (emit_common_tails_for_components): Likewise.
12890 (insert_prologue_epilogue_for_components): Likewise.
12892 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12894 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
12897 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12899 * df-core.c (df_set_blocks): Start using auto_bitmap.
12900 (df_compact_blocks): Likewise.
12901 * df-problems.c (df_rd_confluence_n): Likewise.
12902 * df-scan.c (df_insn_rescan_all): Likewise.
12903 (df_process_deferred_rescans): Likewise.
12904 (df_update_entry_block_defs): Likewise.
12905 (df_update_exit_block_uses): Likewise.
12906 (df_entry_block_bitmap_verify): Likewise.
12907 (df_exit_block_bitmap_verify): Likewise.
12908 (df_scan_verify): Likewise.
12909 * lra-constraints.c (lra_constraints): Likewise.
12910 (undo_optional_reloads): Likewise.
12911 (lra_undo_inheritance): Likewise.
12912 * lra-remat.c (calculate_gen_cands): Likewise.
12913 (do_remat): Likewise.
12914 * lra-spills.c (assign_spill_hard_regs): Likewise.
12915 (spill_pseudos): Likewise.
12916 * tree-ssa-pre.c (bitmap_set_and): Likewise.
12917 (bitmap_set_subtract_values): Likewise.
12919 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12921 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
12922 management with auto_bitmap.
12923 (fix_inter_tick): Likewise.
12924 (fix_recovery_deps): Likewise.
12925 * ira.c (add_store_equivs): Likewise.
12926 (find_moveable_pseudos): Likewise.
12927 (split_live_ranges_for_shrink_wrap): Likewise.
12928 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
12929 (rtx_reuse_manager::seen_def_p): Likewise.
12930 (rtx_reuse_manager::set_seen_def): Likewise.
12931 * print-rtl.h (class rtx_reuse_manager): Likewise.
12933 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12935 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
12937 (migrate_btr_def): Likewise.
12938 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
12939 * df-core.c (loop_post_order_compute): Likewise.
12940 (loop_inverted_post_order_compute): Likewise.
12941 * hsa-common.h: Likewise.
12942 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
12943 * init-regs.c (initialize_uninitialized_regs): Likewise.
12944 * ipa-inline.c (resolve_noninline_speculation): Likewise.
12945 (inline_small_functions): Likewise.
12946 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
12947 * ira.c (combine_and_move_insns): Likewise.
12948 (build_insn_chain): Likewise.
12949 * loop-invariant.c (find_invariants): Likewise.
12950 * lower-subreg.c (propagate_pseudo_copies): Likewise.
12951 * predict.c (tree_predict_by_opcode): Likewise.
12952 (predict_paths_leading_to): Likewise.
12953 (predict_paths_leading_to_edge): Likewise.
12954 (estimate_loops_at_level): Likewise.
12955 (estimate_loops): Likewise.
12956 * shrink-wrap.c (try_shrink_wrapping): Likewise.
12957 (spread_components): Likewise.
12958 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
12959 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
12960 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
12961 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
12962 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
12963 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
12964 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
12965 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
12966 (mark_threaded_blocks): Likewise.
12967 (thread_through_all_blocks): Likewise.
12968 * tree-ssa.c (verify_ssa): Likewise.
12969 (execute_update_addresses_taken): Likewise.
12970 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
12972 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12974 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
12976 (post_order_compute): Likewise.
12977 (inverted_post_order_compute): Likewise.
12978 (pre_and_rev_post_order_compute_fn): Likewise.
12980 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12982 * genrecog.c (int_set::int_set): Explicitly construct our
12983 auto_vec base class.
12984 * vec.h (auto_vec::auto_vec): New constructor.
12986 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12988 * bitmap.h (class auto_bitmap): New constructor taking
12989 bitmap_obstack * argument.
12991 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12993 * bitmap.h (class auto_bitmap): Change type of m_bits to
12994 bitmap_head, and adjust ctor / dtor and member operators.
12996 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
12998 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
12999 when returned register mode doesn't match original mode.
13001 2017-05-12 Jeff Law <law@redhat.com>
13002 Jakub Jelinek <jakub@redhat.com>
13004 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
13005 we look for cc setter after the compare-elim changes.
13006 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
13007 within the vector to match what compare-elim now expects.
13008 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
13009 (xorsi3_flags, one_cmplsi2_flags): Likewise.
13011 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
13012 after the compare-elim changes.
13013 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
13014 the vector to match what compare-elim now expects.
13015 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
13016 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
13017 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
13018 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
13019 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
13021 * config/visium/visium.c (single_set_and_flags): Fix where
13022 we look for cc setter after the compare-elim changes.
13023 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
13024 with the vector to match what compare-elim now expects.
13025 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
13026 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
13027 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
13028 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
13029 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
13030 (neg<mode>2_insn_set_overflow): Likewise.
13032 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
13034 PR middle-end/79794
13035 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
13036 maybe_expand_insn call, set ops[0].target. If still set after call,
13037 set alt_rtl. Add extra arg to recursive calls.
13038 (extract_bit_field): Add alt_rtl argument. Pass to
13040 * expmed.h (extract_bit_field): Fix prototype.
13041 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
13042 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
13043 to extract_bit_field_calls.
13044 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
13045 Pass alt_rtl to extract_bit_field calls.
13046 * calls.c (store_unaligned_arguments_into_psuedos)
13047 load_register_parameters): Pass extra NULL to extract_bit_field calls.
13048 * optabs.c (maybe_legitimize_operand): Clear op->target when call
13050 * optabs.h (struct expand_operand): Add target bitfield.
13052 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13054 * compare-elim.c (try_eliminate_compare): Canonicalize
13055 operation with embedded compare to
13056 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
13057 (set (reg) (operation)].
13059 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
13061 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13064 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
13065 cost of adding a carry flag for ADC instruction.
13066 [case MINUS]: Ignore the cost of subtracting a carry flag
13067 for SBB instruction.
13069 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
13071 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
13073 * config/rs6000/bmiintrin.h: New file.
13074 * config/rs6000/bmi2intrin.h: New file.
13075 * config/rs6000/x86intrin.h: New file.
13077 2017-05-12 Jeff Law <law@redhat.com>
13079 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
13082 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
13084 PR middle-end/80707
13085 * tree-cfg.c: Remove cfg edges of unreachable case statements.
13087 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13089 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
13090 early expansion of vector divide builtins.
13091 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
13092 builtins identified as having unsigned arguments.
13094 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13096 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
13097 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
13098 expansion of vector logical operations (and, andc, or, xor,
13101 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13103 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
13104 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
13106 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13108 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
13109 early GIMPLE expansion of vector multiplies.
13111 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13113 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
13114 TARGET_HAVE_MOVT conditional.
13115 (movt splitter): Likewise.
13117 2017-05-12 Richard Biener <rguenther@suse.de>
13119 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
13120 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13121 Fold all stmts not inplace.
13123 2017-05-12 Richard Biener <rguenther@suse.de>
13125 PR tree-optimization/80713
13126 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
13127 inserted_exprs bit for not removed stmts.
13129 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
13131 PR middle-end/69921
13132 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
13133 parallelized" attribute for parallelized OpenACC kernels.
13134 * omp-offload.c (execute_oacc_device_lower): Use it.
13136 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
13137 Set "oacc kernels" attribute.
13138 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
13139 parameter. Adjust all users.
13140 (oacc_fn_attrib_kernels_p): Remove function.
13141 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
13142 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
13143 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
13144 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
13145 assert "oacc kernels" attribute is set.
13147 2017-05-11 Carl Love <cel@us.ibm.com>
13149 * config/rs6000/rs6000-c: Add support for built-in functions
13150 vector unsigned char vec_popcnt (vector signed char)
13151 vector unsigned char vec_popcnt (vector unsigned char)
13152 vector unsigned short vec_popcnt (vector signed short)
13153 vector unsigned short vec_popcnt (vector unsigned short)
13154 vector unsigned int vec_popcnt (vector signed int)
13155 vector unsigned int vec_popcnt (vector unsigned int)
13156 vector unsigned long long vec_popcnt (vector signed long long)
13157 vector unsigned long long vec_popcnt (vector unsigned long long)
13158 vector signed long long vec_slo (vector signed long long,
13159 vector signed char)
13160 vector signed long long vec_slo (vector signed long long,
13161 vector unsigned char)
13162 vector unsigned long long vec_slo (vector unsigned long long,
13163 vector signed char)
13164 vector unsigned long long vec_slo (vector unsigned long long,
13165 vector unsigned char)
13166 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
13167 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
13168 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
13169 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
13170 * doc/extend.texi: Update the built-in documentation file for the
13171 new built-in functions.
13173 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
13175 * attribs.h (sorted_attr_string): Move machine independent
13176 functions for target clone support from the i386 port to common
13177 code. Rename ix86_function_versions to common_function_versions.
13178 Rename make_name to make_unique_name.
13179 (common_function_versions): Likewise.
13180 (make_unique_name): Likewise.
13181 (make_dispatcher_decl): Likewise.
13182 (is_function_default_version): Likewise.
13183 * attribs.c (attr_strcmp): Likewise.
13184 (sorted_attr_string): Likewise.
13185 (common_function_versions): Likewise.
13186 (make_unique_name): Likewise.
13187 (make_dispatcher_decl): Likewise.
13188 (is_function_default_version): Likewise.
13189 * config/i386/i386.c (attr_strcmp): Likewise.
13190 (sorted_attr_string): Likewise.
13191 (ix86_function_versions): Likewise.
13192 (make_name): Likewise.
13193 (make_dispatcher_decl): Likewise.
13194 (is_function_default_version): Likewise.
13195 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
13197 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13200 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
13201 Account for direct move costs for vec_construct of integer
13204 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
13207 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
13208 (UNSPEC_STX_ATOMIC): Ditto.
13209 (loaddi_via_sse): New insn.
13210 (storedi_via_sse): Ditto.
13211 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
13212 Update corresponding peephole2 patterns.
13213 (atomic_storedi_fpu): Ditto.
13215 2017-05-11 Julia Koval <julia.koval@intel.com>
13217 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
13218 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
13220 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
13221 (__builtin_ia32_rsqrt14ss_mask): New builtins.
13222 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
13224 2017-05-11 Nathan Sidwell <nathan@acm.org>
13226 * graphite-poly.c: Include dumpfile.h.
13228 * dumpfle.h (dump_function): Declare here ...
13229 * tree-dump.h (dump_function): ... not here.
13230 * dumpfile.c: #include tree-cfg.h.
13231 (dump_function): Move here from ...
13232 * tree-dump.c (dump_function): ... here.
13233 * gimplify.c: #include splay-tree.h, not tree-dump.h.
13234 * graphite-poly.c: Don't include tree-dump.h.
13235 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
13236 * print-tree.c: Likewise.
13237 * stor-layout.c: Likewise.
13238 * tree-nested.c: Likewise.
13240 * dumpfile.c (dump_start): Use TDF_FLAGS.
13241 (dump_enable_all): Fix TDF_KIND check thinko.
13243 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
13245 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13246 array entries to represent two legal parameterizations of the
13247 overloaded __builtin_cmpb function, as represented by the
13248 P6_OV_BUILTIN_CMPB constant.
13249 (altivec_resolve_overloaded_builtin): Add special case handling
13250 for the __builtin_cmpb function, as represented by the
13251 P6_OV_BUILTIN_CMPB constant.
13252 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
13253 (BU_P6_64BIT_2): New macro.
13254 (BU_P6_OVERLOAD_2): New macro
13255 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
13256 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
13257 (CMPB): Add overload support to represent both 32-bit and 64-bit
13258 compare-bytes function.
13259 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13260 support for TARGET_CMPB.
13261 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
13262 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13263 documentation of the __builtin_cmpb overloaded built-in function.
13265 2017-05-11 Richard Biener <rguenther@suse.de>
13267 PR tree-optimization/80705
13268 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
13269 bases are not vectorizable.
13271 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13273 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
13274 when counting register pressure.
13276 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13278 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
13279 (iv_ca_more_deps): Renamed to ...
13280 (iv_ca_compare_deps): ... this.
13281 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
13283 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13285 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
13287 (determine_group_iv_costs): ... here.
13288 (find_inv_vars_cb): Record inv var if it's not recorded before.
13290 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13292 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
13293 (get_shiftadd_cost): Ditto.
13295 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13297 * tree-ssa-address.c: Include header file.
13298 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
13300 (add_to_parts): Refactor.
13301 (addr_to_parts): New parameter. Update use of move_hint_to_base.
13302 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
13305 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13307 PR tree-optimization/53090
13308 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
13310 (extract_cond_operands): Detect condition with IV on both sides
13311 and return COMP_IV_EXPR_2.
13312 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
13313 (rewrite_use_compare): Simplify by removing call to function
13314 extract_cond_operands.
13316 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13318 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
13319 (extract_cond_operands): Detect condition comparing against non-
13320 invariant bound and return appropriate enum value.
13321 (find_interesting_uses_cond): Update use of extract_cond_operands.
13322 Handle its return value accordingly.
13323 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
13325 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13327 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
13328 nonlinear iv_use computation in loop invariant sensitive way.
13330 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13332 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
13333 (find_iv_candidates): Call relate_compare_use_with_all_cands.
13335 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13337 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
13338 (dump_cand): Support iv_cand.inv_exprs.
13339 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
13341 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
13344 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13346 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
13348 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
13349 as local function. Include necessary header files.
13350 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
13352 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13354 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
13356 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13358 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
13359 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
13360 RSHIFT_EXPR and BIT_NOT_EXPR.
13362 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13364 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
13365 (adjust_setup_cost): New parameter supporting round up adjustment.
13366 (struct address_cost_data): Delete.
13367 (force_expr_to_var_cost): Don't bound cost with spill_cost.
13368 (split_address_cost, ptr_difference_cost): Delete.
13369 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
13370 (struct ainc_cost_data): New struct.
13371 (get_address_cost_ainc): New function.
13372 (get_address_cost, get_computation_cost): Reimplement.
13373 (determine_group_iv_cost_address): Record inv_expr for all uses of
13375 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
13376 (iv_ca_has_deps): Reimplemented to ...
13377 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
13379 (iv_ca_extend): Call iv_ca_more_deps.
13381 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13383 * tree-ssa-address.c (struct mem_address): Move to header file.
13384 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
13385 * tree-ssa-address.h (struct mem_address): Move from C file.
13386 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
13388 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13390 * tree-affine.h (aff_combination_type): New interface.
13391 (aff_combination_zero_p): Remove static.
13392 (aff_combination_const_p): New interface.
13393 (aff_combination_singleton_var_p): New interfaces.
13395 2017-05-11 Richard Biener <rguenther@suse.de>
13397 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13398 Skip unreachable blocks and destinations.
13399 (eliminate): Move stmt removal and fixup ...
13400 (fini_eliminate): ... here. Skip inserted exprs.
13401 (pass_pre::execute): Move fini_pre after fini_eliminate.
13402 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
13403 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
13404 PRE to get rid of dead code that has invalid SSA form and
13405 split critical edges again.
13407 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13409 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
13411 2017-05-11 Richard Biener <rguenther@suse.de>
13413 * passes.c (execute_function_todo): Verify loops if they are
13414 said to be up-to-date.
13415 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
13416 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
13418 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
13421 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
13422 handle calling assemble_external ourself.
13425 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
13426 modes with zero size. Enhance comment.
13428 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13430 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
13431 built-ins for vec_xl and vec_xst with short and char pointer
13434 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13436 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
13437 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
13438 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
13439 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
13440 (_mm_maskz_min_round_ss): New intrinsics.
13441 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
13442 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
13443 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
13444 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
13445 (__builtin_ia32_minss_mask_round): New builtins.
13446 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13447 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13448 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
13450 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
13451 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
13453 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
13456 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13458 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
13459 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
13460 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
13461 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
13462 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
13463 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
13464 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
13465 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13466 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
13467 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
13468 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
13469 (__builtin_ia32_mulss_mask_round): New builtins.
13470 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13471 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13472 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
13474 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
13475 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
13477 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
13480 2017-05-10 Julia Koval <julia.koval@intel.com>
13482 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
13483 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
13484 (_mm256_setr_m128i): New intrinsics.
13486 2017-05-10 Julia Koval <julia.koval@intel.com>
13488 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
13489 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
13490 (_mm_maskz_rcp14_ss): New intrinsics.
13491 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
13492 (__builtin_ia32_rcp14ss_mask): New builtins.
13493 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
13495 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
13497 PR tree-optimization/51513
13498 * tree-cfg.c (gimple_seq_unreachable_p): New function.
13499 (assert_unreachable_fallthru_edge_p): Use it.
13500 (group_case_labels_stmt): Likewise.
13501 * tree-cfg.h: Prototype it.
13502 * stmt.c: Include cfghooks.h and tree-cfg.h.
13503 (emit_case_dispatch_table) <gap_label>: New local variable.
13504 Use it to fill dispatch table gaps.
13505 Test for default_label before updating probabilities.
13506 (expand_case) <default_label>: Remove unneeded initialization.
13507 Test for unreachable default case statement and remove its edge.
13508 Set default_label accordingly.
13509 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
13511 2017-05-10 Carl Love <cel@us.ibm.com>
13513 * config/rs6000/rs6000-c: Add support for built-in functions
13514 vector signed char vec_neg (vector signed char)
13515 vector signed short int vec_neg (vector short int)
13516 vector signed int vec_neg (vector signed int)
13517 vector signed long long vec_neg (vector signed long long)
13518 vector float vec_neg (vector float)
13519 vector double vec_neg (vector double)
13520 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
13522 * config/rs6000/altivec.h: Add define for vec_neg
13523 * doc/extend.texi: Update the built-in documentation for the
13524 new built-in functions.
13526 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13528 PR tree-optimization/77644
13529 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
13531 2017-05-10 Nathan Sidwell <nathan@acm.org>
13533 * dumpfile.h (TDI_lang_all): New.
13534 (TDF_KIND): New. Renumber others
13535 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
13537 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
13539 (get_dump_file_name): Adjust suffix generation.
13540 (dump_enable_all): Use TDF_KIND.
13541 * doc/invoke.texi (-fdump-lang-all): Document.
13543 * dumpfile.h: Tabify.
13545 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
13548 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
13549 Move member access before delete.
13551 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
13553 * tree-inline.c (expand_call_inline): Split block at stmt
13556 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13559 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
13560 are now unused after splitting mov{sf,sd}_hardfloat.
13561 (f32_lr2): Likewise.
13562 (f32_lm): Likewise.
13563 (f32_lm2): Likewise.
13564 (f32_li): Likewise.
13565 (f32_li2): Likewise.
13566 (f32_lv): Likewise.
13567 (f32_sr): Likewise.
13568 (f32_sr2): Likewise.
13569 (f32_sm): Likewise.
13570 (f32_sm2): Likewise.
13571 (f32_si): Likewise.
13572 (f32_si2): Likewise.
13573 (f32_sv): Likewise.
13574 (f32_dm): Likewise.
13575 (f32_vsx): Likewise.
13576 (f32_av): Likewise.
13577 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
13578 For movsf, order stores so the VSX stores occur before the GPR
13579 store which encourages the register allocator to use a traditional
13580 FPR instead of a GPR. For movsd, order the stores so that the GPR
13581 store comes before the VSX stores to allow the power6 to work.
13582 This is due to the power6 not having a 32-bit integer store
13583 instruction from a FPR.
13584 (movsf_hardfloat): Likewise.
13585 (movsd_hardfloat): Likewise.
13587 2017-05-09 Martin Sebor <msebor@redhat.com>
13589 PR translation/80280
13590 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
13593 PR translation/80280
13594 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
13595 data member added in r247778.
13596 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
13598 2017-05-09 Nathan Sidwell <nathan@acm.org>
13600 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
13602 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
13605 2017-05-09 Marek Polacek <polacek@redhat.com>
13607 * doc/invoke.texi: Fix typo.
13609 2017-05-09 Richard Biener <rguenther@suse.de>
13611 * tree-vrp.c (vrp_val_is_max): Adjust comment.
13612 (vrp_val_is_min): Likewise.
13613 (set_value_range_to_value): Likewise.
13614 (set_value_range_to_nonnegative): Likewise.
13615 (gimple_assign_nonzero_p): Likewise.
13616 (gimple_stmt_nonzero_p): Likewise.
13617 (vrp_int_const_binop): Likewise. Remove unreachable case.
13618 (adjust_range_with_scev): Adjust comments.
13619 (compare_range_with_value): Likewise.
13620 (extract_range_from_phi_node): Likewise.
13621 (test_for_singularity): Likewise.
13623 2017-05-09 Richard Biener <rguenther@suse.de>
13625 * tree-vrp.c (get_single_symbol): Add assert that we don't
13626 get overflowed constants as invariant part.
13627 (compare_values_warnv): Add comment before the TREE_NO_WARNING
13628 checks. Use wi::cmp instead of recursing for integer constants.
13629 (compare_values): Just ignore whether we assumed undefined
13630 overflow instead of failing the compare.
13631 (extract_range_for_var_from_comparison_expr): Add comment before the
13632 TREE_NO_WARNING sets.
13633 (test_for_singularity): Likewise.
13634 (extract_range_from_comparison): Do not disable optimization
13635 when we assumed undefined overflow.
13636 (extract_range_basic): Remove init of unused var.
13638 2017-05-09 Richard Biener <rguenther@suse.de>
13640 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
13641 (extract_range_from_multiplicative_op_1): Adjust.
13642 (extract_range_from_binary_expr_1): Use int_const_binop.
13644 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13647 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
13648 rs6000_store_data_bypass_p in seven define_bypass directives and
13649 in several comments.
13650 * config/rs6000/rs6000-protos.h: Add prototype for
13651 rs6000_store_data_bypass_p function.
13652 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
13653 function implements slightly different (rs6000-specific) semantics
13654 than store_data_bypass_p, returning false rather than aborting
13655 with assertion error when arguments do not satisfy the
13656 requirements of store data bypass.
13657 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
13658 rs6000_store_data_bypass_p.
13660 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
13662 * config/xtensa/xtensa-protos.h
13663 (xtensa_initial_elimination_offset): New declaration.
13664 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13665 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
13666 macro definition, add case for FRAME_POINTER_REGNUM when
13667 FRAME_GROWS_DOWNWARD.
13668 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
13669 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
13670 xtensa_initial_elimination_offset.
13672 2017-05-08 Nathan Sidwell <nathan@acm.org>
13674 * doc/invoke.texi: Alphabetize -fdump options.
13676 2017-05-08 Martin Sebor <msebor@redhat.com>
13678 PR translation/80280
13679 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
13681 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
13683 * target.def (compute_frame_layout): New optional target hook.
13684 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
13685 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
13686 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
13688 * reload1.c (verify_initial_elim_offsets): Likewise.
13689 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
13690 (use_simple_return_p): Call arm_compute_frame_layout if needed.
13691 (arm_get_frame_offsets): Split up into this ...
13692 (arm_compute_frame_layout): ... and this function.
13694 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
13696 * config/aarch64/constraints.md (Usa): New constraint.
13697 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
13699 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
13701 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
13702 with_multilib_list after it has been checked.
13704 2017-05-08 Richard Biener <rguenther@suse.de>
13706 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
13707 (bitmap_set_subtract_values): Likewise.
13709 2017-05-08 Richard Biener <rguenther@suse.de>
13711 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
13712 (gimple_assign_nonzero): ... this and remove strict_overflow_p
13714 (gimple_stmt_nonzero_warnv_p): Rename to ...
13715 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
13717 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
13718 (extract_range_basic): Adjust, do not disable propagation on
13719 strict overflow sensitive simplification.
13720 (vrp_visit_cond_stmt): Likewise.
13722 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
13724 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
13725 body size unconditionally.
13727 2017-05-07 Jeff Law <law@redhat.com>
13730 2017-05-06 Jeff Law <law@redhat.com>
13731 PR tree-optimization/78496
13732 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
13735 PR tree-optimization/78496
13736 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
13737 (simplify_stmt_using_ranges): Call it.
13738 (vrp_dom_walker::before_dom_children): Extract equivalences
13739 from an ASSERT_EXPR with an equality comparison against a
13742 2017-05-06 Jeff Law <law@redhat.com>
13744 PR tree-optimization/78496
13745 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
13748 PR tree-optimization/78496
13749 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
13750 (simplify_stmt_using_ranges): Call it.
13751 (vrp_dom_walker::before_dom_children): Extract equivalences
13752 from an ASSERT_EXPR with an equality comparison against a
13755 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
13757 * lra-constraints.c (lra_copy_reg_equiv): New function.
13758 (split_reg): Use it to copy equivalence information from the
13759 original register to the spill register.
13761 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
13763 PR rtl-optimization/75964
13764 * simplify-rtx.c (simplify_const_relational_operation): Remove
13765 invalid handling of comparisons of integer ABS.
13767 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
13769 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
13770 initialize to zero.
13771 (init_regs): Remove declaration.
13772 (function_arg_advance_32): Initialize error_p as boolean variable.
13774 2017-05-05 Nathan Sidwell <nathan@acm.org>
13776 * store-motion.c (remove_reachable_equiv_notes): Reformat long
13777 lines. Use for (;;).
13779 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13781 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
13782 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
13783 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
13784 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
13785 VF=2 that require versioning.
13787 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13789 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
13792 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13794 * diagnostic.h (diagnostic_override_option_index): Convert from
13795 macro to inline function.
13797 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13799 * diagnostic.c (last_module_changed_p): New function.
13800 (set_last_module): New function.
13801 (diagnostic_report_current_module): Convert macro usage to
13802 the above functions.
13803 * diagnostic.h (diagnostic_context::last_module): Strengthen
13804 from const line_map * to const line_map_ordinary *.
13805 (diagnostic_last_module_changed): Delete macro.
13806 (diagnostic_set_last_module): Delete macro.
13808 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13810 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
13811 with diagnostic_report_diagnostic.
13812 (diagnostic_n_impl_richloc): Likewise.
13813 * diagnostic.h (report_diagnostic): Delete macro.
13814 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
13815 with diagnostic_report_diagnostic.
13816 * substring-locations.c (format_warning_va): Likewise.
13818 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13820 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
13821 save/restor of format_spec. Move option-printing code to...
13822 (print_option_information): ...this new function, and
13823 reimplement by simply printing to the pretty_printer,
13824 rather than appending to the format string.
13826 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13828 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
13829 handling logic into...
13830 (update_effective_level_from_pragmas): ...this new function.
13832 2017-05-04 Andrew Waterman <andrew@sifive.com>
13834 * config/riscv/riscv.opt (mstrict-align): New option.
13835 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
13836 (SLOW_UNALIGNED_ACCESS): Define.
13837 (riscv_slow_unaligned_access): Declare.
13838 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
13840 (riscv_slow_unaligned_access): New variable.
13841 (rocket_tune_info): Set slow_unaligned_access to true.
13842 (optimize_size_tune_info): Set slow_unaligned_access to false.
13843 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
13844 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
13845 (riscv_option_override): Set riscv_slow_unaligned_access.
13846 * doc/invoke.texi: Add -mstrict-align to RISC-V.
13848 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
13850 * config/riscv/riscv.md: Unify indentation.
13852 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13857 * config/rs6000/rs6000.md (u code attribute): Add FIX and
13859 (extendsi<mode>2): Add support for doing sign extension via
13860 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
13861 don't have ISA 3.0 instructions.
13862 (extendsi<mode>2 splitter): Likewise.
13863 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
13864 generate the normal insns since SImode can now go in vector
13865 registers. Disallow the special UNSPECs needed for previous
13866 machines to hide SImode being used. Add new insns
13867 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
13868 (fix_trunc<mode>si2_stfiwx): Likewise.
13869 (fix_trunc<mode>si2_internal): Likewise.
13870 (fixuns_trunc<mode>si2): Likewise.
13871 (fixuns_trunc<mode>si2_stfiwx): Likewise.
13872 (fctiw<u>z_<mode>_smallint): Likewise.
13873 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
13874 of floating point to 32-bit integer from doing a direct move to
13875 the GPR registers to do a store.
13876 (fctiwz_<mode>): Break long line.
13878 2017-05-05 Bin Cheng <bin.cheng@arm.com>
13880 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
13881 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
13882 (addr_list, addr_offset_valid_p): New.
13883 (split_address_groups): Check offset validity with above function.
13884 (gt-tree-ssa-loop-ivopts.h): Include header file.
13886 2017-05-05 Nathan Sidwell <nathan@acm.org>
13888 * config.gcc (arm*-*-*): Add missing 'fi'.
13890 2017-05-05 Steve Ellcey <sellcey@cavium.com>
13892 * doc/invoke.texi (-fopt-info): Explicitly say order of options
13893 included in -fopt-info does not matter.
13894 * doc/optinfo.texi (-fopt-info): Fix description of default
13895 behavour. Explicitly say order of options included in -fopt-info
13898 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13900 * config.gcc: Allow combinations of aprofile and rmprofile values for
13901 --with-multilib-list.
13902 * config/arm/t-multilib: New file.
13903 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
13904 variables. Remove setting of ISA and floating-point ABI in
13905 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
13906 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
13907 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
13908 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
13910 * config/arm/t-rmprofile: Likewise except for the matches changes.
13911 * doc/install.texi (--with-multilib-list): Document the combination of
13912 aprofile and rmprofile values and warn about pitfalls in doing that.
13914 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
13916 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
13917 (movdi_aarch64): Likewise.
13919 2017-05-05 Jakub Jelinek <jakub@redhat.com>
13921 PR tree-optimization/80632
13922 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
13924 (build_arrays): Initialize it for virtual phis.
13925 (fix_phi_nodes): Use it for virtual phis.
13927 PR tree-optimization/80558
13928 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
13929 [x, y] op z into [x op, y op z] for op & or | if conditions
13932 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
13933 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13936 * config/arm/arm.md (use_literal_pool): Remove.
13937 (64-bit immediate split): No longer takes cost into consideration
13938 if arm_disable_literal_pool is enabled.
13939 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
13940 used when arm_disable_literal_pool is enabled.
13941 (arm_max_const_double_inline_cost): Remove use of
13942 arm_disable_literal_pool.
13943 (push_minipool_fix): Add assert.
13944 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
13945 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
13946 (no_literal_pool_sf_immediate): New.
13948 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13950 PR tree-optimization/80613
13951 * tree-ssa-dce.c (propagate_necessity): Remove cases for
13952 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
13954 2017-05-05 Richard Biener <rguenther@suse.de>
13956 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
13958 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
13960 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
13961 of this flag from insn conditions due to removal from r247495.
13963 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
13965 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
13967 (arm_early_store_addr_dep_ptr): Likewise.
13968 * config/arm/aarch-common-protos.h
13969 (arm_early_load_addr_dep_ptr): Add prototype.
13970 (arm_early_store_addr_dep_ptr): Likewise.
13971 * config/arm/cortex-a53.md: Add new bypasses.
13973 2017-05-05 Jakub Jelinek <jakub@redhat.com>
13975 * tree.c (next_type_uid): Change type to unsigned.
13976 (type_hash_canon): Decrement back next_type_uid if
13977 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
13978 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
13981 2017-05-04 Martin Sebor <msebor@redhat.com>
13983 * builtins.c: Fix a trivial typo in a comment.
13985 PR middle-end/79234
13986 * builtins.c (check_sizes): Adjust to handle reading past the end.
13987 Avoid printing excessive upper bound of ranges. Use %E to print
13988 tree nodes instead of converting them to %wu.
13989 (expand_builtin_memchr): New function.
13990 (compute_dest_size): Rename...
13991 (compute_objsize): ...to this.
13992 (expand_builtin_memcpy): Adjust.
13993 (expand_builtin_mempcpy): Adjust.
13994 (expand_builtin_strcat): Adjust.
13995 (expand_builtin_strcpy): Adjust.
13996 (check_strncat_sizes): Adjust.
13997 (expand_builtin_strncat): Adjust.
13998 (expand_builtin_strncpy): Adjust and simplify.
13999 (expand_builtin_memset): Adjust.
14000 (expand_builtin_bzero): Adjust.
14001 (expand_builtin_memcmp): Adjust.
14002 (expand_builtin): Handle memcmp.
14003 (maybe_emit_chk_warning): Check strncat just once.
14005 2017-05-04 Martin Sebor <msebor@redhat.com>
14007 PR preprocessor/79214
14008 PR middle-end/79222
14009 PR middle-end/79223
14010 * builtins.c (check_sizes): Add inlining context and issue
14011 warnings even when -Wno-system-headers is set.
14012 (check_strncat_sizes): Same.
14013 (expand_builtin_strncat): Same.
14014 (expand_builtin_memmove): New function.
14015 (expand_builtin_stpncpy): Same.
14016 (expand_builtin): Handle memmove and stpncpy.
14018 2017-05-04 Bin Cheng <bin.cheng@arm.com>
14020 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
14021 which is not used any more.
14023 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14025 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
14027 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14029 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
14030 (cortexa53_tunings): Likewise.
14031 (cortexa57_tunings): Likewise.
14032 (cortexa72_tunings): Likewise.
14033 (cortexa73_tunings): Likewise.
14035 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14037 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
14038 Set loop alignment to 8.
14040 2017-05-04 Martin Sebor <msebor@redhat.com>
14042 PR translation/80280
14043 * builtins.c (expand_builtin_object_size): Add missing quoting to
14044 %D and like directives.
14045 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
14046 (hsa_type_for_tree_type): Same.
14047 (verify_function_arguments): Same.
14048 * symtab.c (symbol_table::change_decl_assembler_name): Same.
14049 * varasm.c (get_section): Same.
14052 2017-05-04 Martin Sebor <msebor@redhat.com>
14054 PR translation/80280
14055 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
14057 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14059 * config/aarch64/aarch64.c (generic_addrcost_table):
14060 Change HI/TI mode setting.
14062 2017-05-04 Martin Jambor <mjambor@suse.cz>
14064 PR tree-optimization/80622
14065 * tree-sra.c (comes_initialized_p): New function.
14066 (build_accesses_from_assign): Only set write lazily when
14067 comes_initialized_p is false.
14068 (analyze_access_subtree): Use comes_initialized_p.
14069 (propagate_subaccesses_across_link): Assert !comes_initialized_p
14070 instead of testing for PARM_DECL.
14072 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14074 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
14075 constraint on operand 0 to allow more general addressing modes.
14076 Adjust output template.
14077 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
14079 * config/aarch64/aarch64-protos.h
14080 (aarch64_address_valid_for_prefetch_p): Declare prototype.
14081 * config/aarch64/constraints.md (Dp): New address constraint.
14082 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
14085 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
14087 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
14088 update use of estimate_ipcp_clone_size_and_time.
14089 (estimate_local_effects): Update use of
14090 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
14091 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
14092 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
14093 Return nonspecialized time.
14095 2017-05-04 Richard Biener <rguenther@suse.de>
14097 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
14098 for the last VUSE which def dominates the PHI. Directly call
14100 (get_continuation_for_phi_1): Remove.
14102 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
14104 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
14105 to explain the use of truncating division. Cap the number of
14106 iterations to the maximum given by nb_iterations_upper_bound,
14109 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14111 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
14112 * configure: Regenerate.
14113 * config.in: Regenerate.
14114 * config/i386/driver-mingw32.c: new file.
14115 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
14116 * config.host: Link driver-mingw32.o on MinGW host.
14117 * doc/install.texi: Document new --enable-mingw-wildcard configure
14120 2017-05-04 Marek Polacek <polacek@redhat.com>
14122 PR tree-optimization/80612
14123 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
14125 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14126 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
14128 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
14129 (movt splitter): Likewise.
14130 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
14131 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
14132 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
14133 block for Thumb-1 with MOVT.
14134 (thumb2_legitimate_address_p): Move code block ...
14135 (can_avoid_literal_pool_for_label_p): ... into this new function.
14136 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
14138 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
14139 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
14140 "M-profile targets with the MOVT instruction".
14142 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14144 * config/arm/arm-builtins.c (arm_init_builtins): Rename
14145 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
14146 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
14148 2017-05-04 Martin Liska <mliska@suse.cz>
14150 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
14151 variable cond_code.
14153 2017-05-04 Richard Biener <rguenther@suse.de>
14155 * tree.c (array_at_struct_end_p): Handle arrays at struct
14156 end with flexarrays more conservatively. Refactor and treat
14157 arrays of arrays or aggregates more strict. Fix
14158 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
14159 * tree.c (array_at_struct_end_p): Adjust prototype.
14160 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
14161 * gimple-fold.c (get_range_strlen): Likewise.
14162 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
14164 2017-05-04 Richard Biener <rguenther@suse.de>
14166 PR tree-optimization/31130
14167 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
14169 (supports_overflow_infinity): Likewise.
14170 (is_negative_overflow_infinity): Likewise.
14171 (is_positive_overflow_infinity): Likewise.
14172 (is_overflow_infinity): Likewise.
14173 (stmt_overflow_infinity): Likewise.
14174 (overflow_infinity_range_p): Likewise.
14175 (usable_range_p): Remove as always returning true.
14176 (make_overflow_infinity): Remove.
14177 (negative_overflow_infinity): Likewise.
14178 (positive_overflow_infinity): Likewise.
14179 (avoid_overflow_infinity): Likewise.
14180 (set_value_range): Adjust accordingly.
14181 (set_value_range_to_nonnegative): Likewise, remove now unused
14182 overflow_infinity arg.
14183 (vrp_operand_equal_p): Adjust.
14184 (update_value_range): Likewise.
14185 (range_int_cst_singleton_p): Likewise.
14186 (operand_less_p): Likewise.
14187 (compare_values_warnv): Likewise.
14188 (extract_range_for_var_from_comparison_expr): Likewise.
14189 (vrp_int_const_binop): Likewise.
14190 (zero_nonzero_bits_from_vr): Likewise.
14191 (extract_range_from_multiplicative_op_1): Likewise.
14192 (extract_range_from_binary_expr_1): Likewise.
14193 (extract_range_from_unary_expr): Likewise.
14194 (extract_range_from_comparison): Likewise.
14195 (extract_range_basic): Likewise.
14196 (adjust_range_with_scev): Likewise.
14197 (compare_ranges): Likewise.
14198 (compare_range_with_value): Likewise.
14199 (dump_value_range): Likewise.
14200 (test_for_singularity): Likewise, remove strict_overflow_p parameter
14202 (simplify_cond_using_ranges): Adjust.
14204 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
14206 * brig-builtins.def: Added a builtin for class_f64.
14207 * builtin-types.def: Added a builtin type needed by class_f64.
14209 2017-05-03 Jason Merrill <jason@redhat.com>
14211 * timevar.def: Add TV_CONSTEXPR.
14213 2017-05-03 David Malcolm <dmalcolm@redhat.com>
14215 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
14217 2017-05-03 Martin Jambor <mjambor@suse.cz>
14219 * ipa-prop.c (ipa_update_after_lto_read): Removed.
14220 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
14221 * ipa-cp.c (ipcp_propagate_stage): Do not call
14222 ipa_update_after_lto_read.
14223 * ipa-inline.c (ipa_inline): Likewise.
14225 2017-05-03 Martin Jambor <mjambor@suse.cz>
14227 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
14228 tag. Added a default constructor and a destructor.
14229 (ipa_edge_args_sum_t): New class;
14230 (ipa_edge_args_sum): Declare.
14231 (ipa_edge_args_vector): Remove declaration.
14232 (IPA_EDGE_REF): Use ipa_edge_args_sum.
14233 (ipa_free_edge_args_substructures): Remove declaration.
14234 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
14235 (ipa_edge_args_info_available_for_edge_p): Likewise.
14236 * ipa-prop.c (ipa_edge_args_vector): Removed.
14237 (edge_removal_hook_holder): Likewise.
14238 (edge_duplication_hook_holder): Likewise.
14239 (ipa_edge_args_sum): New variable.
14240 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
14241 ipa_edge_args_vector.
14242 (ipa_free_edge_args_substructures): Likewise.
14243 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
14244 ipa_edge_args_vector.
14245 (ipa_edge_removal_hook): Turned into method
14246 ipa_edge_args_sum_t::remove.
14247 (ipa_edge_duplication_hook): Turned into method
14248 ipa_edge_args_sum_t::duplicate.
14249 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
14250 registering edge hooks.
14251 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
14252 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
14253 ipa_edge_args_sum instead of ipa_edge_args_vector.
14254 * ipa-profile.c (ipa_profile): Likewise.
14256 2017-05-03 Martin Jambor <mjambor@suse.cz>
14258 * symbol-summary.h (function_summary): New method exists.
14259 (function_summary::symtab_removal): Deallocate through release.
14260 (call_summary): New class.
14261 (gt_ggc_mx): New overload.
14262 (gt_pch_nx): Likewise.
14263 (gt_pch_nx): Likewise.
14265 2017-05-03 Jeff Law <law@redhat.com>
14267 PR tree-optimization/78496
14268 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
14269 from simplify_cond_using_ranges. Split off code to walk
14270 backwards through casts into ...
14271 (simplify_cond_using_ranges_2): New function.
14272 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
14273 (execute_vrp): After identifying jump threads, call
14274 simplify_cond_using_ranges_2.
14276 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
14279 * ipa-inline.h (inline_summary): Add ctor.
14280 (create_ggc): Do not use ggc_cleared_alloc.
14282 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
14283 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14285 * gcc.c (handle_braces): Support escaping in switch matching
14287 * doc/invoke.texi (Spec Files): Document it.
14288 Remove superfluous @code markup in items.
14290 2017-05-03 David Malcolm <dmalcolm@redhat.com>
14292 * diagnostic-show-locus.c (struct column_range): New struct.
14293 (get_affected_columns): New function.
14294 (get_printed_columns): New function.
14295 (struct correction): New struct.
14296 (correction::ensure_capacity): New function.
14297 (correction::ensure_terminated): New function.
14298 (struct line_corrections): New struct.
14299 (line_corrections::~line_corrections): New dtor.
14300 (line_corrections::add_hint): New function.
14301 (layout::print_trailing_fixits): Reimplement in terms of the new
14303 (selftest::test_overlapped_fixit_printing): New function.
14304 (selftest::diagnostic_show_locus_c_tests): Call it.
14306 2017-05-03 Nathan Sidwell <nathan@acm.org>
14308 Canonicalize canonical type hashing
14309 * tree.h (type_hash_canon_hash): Declare.
14310 * tree.c (type_hash_list, attribute_hash_list): Move into
14311 type_hash_canon_hash.
14312 (build_type_attribute_qual_variant): Break out hash code calc into
14313 type_hash_canon_hash.
14314 (type_hash_canon_hash): New. Generic type hash computation.
14315 (build_range_type_1, build_array_type_1, build_function_type,
14316 build_method_type_directly, build_offset_type, build_complex_type,
14317 make_vector_type): Call it.
14319 2017-05-03 Richard Biener <rguenther@suse.de>
14321 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
14322 When all DRs have unknown misaligned do not always peel
14323 when there is a store but apply the same costing model as if
14324 there were only loads.
14326 2017-05-03 Richard Biener <rguenther@suse.de>
14329 PR tree-optimization/80492
14330 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
14331 compare_base_decls returning dont-know properly.
14333 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
14335 * config/arm/iterators.md (CCSI): New mode iterator.
14336 (arch): New mode attribute.
14337 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
14338 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
14339 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
14340 code iterator for success result mode.
14341 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
14342 the corresponding new insn generators.
14344 2017-05-03 Bin Cheng <bin.cheng@arm.com>
14347 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14348 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14350 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
14352 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
14353 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
14354 (DDR_A): Wrap DDR argument in brackets.
14355 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
14356 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
14357 (DDR_REVERSED_P): Likewise.
14359 2017-05-03 Jakub Jelinek <jakub@redhat.com>
14361 PR tree-optimization/79472
14362 * tree-switch-conversion.c (struct switch_conv_info): Add
14363 contiguous_range and default_case_nonstandard fields.
14364 (collect_switch_conv_info): Compute contiguous_range and
14365 default_case_nonstandard fields, don't clear final_bb if
14366 contiguous_range and only the default case doesn't have the required
14368 (check_all_empty_except_final): Set default_case_nonstandard instead
14369 of failing if contiguous_range and the default case doesn't have empty
14371 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
14372 and only the default case doesn't have the required constants. Skip
14374 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
14375 if default_case_nonstandard.
14376 (build_constructors): Build constant 1 just once. Assert that default
14377 values aren't inserted in between cases if contiguous_range. Skip
14379 (build_arrays): Skip virtual phis.
14380 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
14381 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
14382 Handle virtual phis.
14383 (gen_inbound_check): Handle default_case_nonstandard case.
14384 (process_switch): Adjust check_final_bb caller. Call
14385 gather_default_values with the first non-default case instead of
14386 default case if default_case_nonstandard.
14388 2017-05-02 Nathan Sidwell <nathan@acm.org>
14390 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
14391 check. Fix formatting.
14393 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
14395 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
14396 errors when comparing specialized and unspecialized times.
14398 2017-05-02 David Malcolm <dmalcolm@redhat.com>
14400 * diagnostic-show-locus.c
14401 (layout::should_print_annotation_line_p): Make private.
14402 (layout::print_annotation_line): Make private.
14403 (layout::annotation_line_showed_range_p): Make private.
14404 (layout::show_ruler): Make private.
14405 (layout::print_source_line): Make private. Pass in line and
14406 line_width, rather than calling location_get_source_line. Drop
14408 (layout::print_leading_fixits): New method.
14409 (layout::print_any_fixits): Rename to...
14410 (layout::print_trailing_fixits): ...this, and make private.
14411 Don't print newline fixits.
14412 (diagnostic_show_locus): Move logic for printing one row into...
14413 (layout::print_line): ...this new function. Move the
14414 location_get_source_line call and error-handling from
14415 print_source_line to here. Call print_leading_fixits, and rename
14416 print_any_fixits to print_trailing_fixits.
14417 (selftest::test_fixit_insert_containing_newline): Update now that
14418 newlines are partially supported.
14419 (selftest::test_fixit_insert_containing_newline_2): New test.
14420 (selftest::test_fixit_replace_containing_newline): Update comments.
14421 (selftest::diagnostic_show_locus_c_tests): Call the new test.
14422 * edit-context.c (class added_line): New class.
14423 (class edited_line): Describe newline handling in comment.
14424 (edited_line::actually_edited_p): New method.
14425 (edited_line::print_content): Delete redundant decl.
14426 (edited_line::m_predecessors): New field.
14427 (edited_file::print_content): Call edited_line::print_content.
14428 (edited_file::print_diff): Update to support newlines.
14429 (edited_file::print_diff_hunk): Likewise.
14430 (edited_file::print_run_of_changed_lines): New function.
14431 (edited_file::print_diff_line): Convert to...
14432 (print_diff_line): ...this.
14433 (edited_file::get_effective_line_count): New function.
14434 (edited_line::edited_line): Initialize new field m_predecessors.
14435 (edited_line::~edited_line): Clean up m_predecessors.
14436 (edited_line::apply_fixit): Handle newlines.
14437 (edited_line::get_effective_line_count): New function.
14438 (edited_line::print_content): New function.
14439 (edited_line::print_diff_lines): New function.
14440 (selftest::test_applying_fixits_insert_containing_newline): New
14442 (selftest::test_applying_fixits_replace_containing_newline): New
14444 (selftest::insert_line): New function.
14445 (selftest::test_applying_fixits_multiple_lines): Add example of
14447 (selftest::edit_context_c_tests): Call the new tests.
14449 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14451 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
14452 parameter cand. Update dump information.
14453 (get_computation_cost): Update uses.
14455 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14457 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
14458 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
14459 (get_computation_at, rewrite_use_address): Update use of
14460 get_computation_aff.
14462 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14464 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
14465 (get_computation): Delete.
14466 (get_computation_cost): Implement like get_computation_cost_at.
14467 Use get_computation_at.
14468 (get_computation_cost_at): Delete.
14469 (rewrite_use_nonlinear_expr): Use get_computation_at.
14470 (rewrite_use_compare, remove_unused_ivs): Ditto.
14472 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14474 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
14476 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14478 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
14479 (ivopts_global_cost_for_size): Rename parameter and update uses.
14480 (iv_ca_recount_cost): Update uses.
14481 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
14482 candidates seperately in n_invs and n_cands.
14483 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
14485 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14487 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
14488 (find_inv_vars_cb): New.
14489 (find_depends): Renamed to ...
14490 (find_inv_vars): ... this.
14491 (add_candidate_1, force_var_cost): Call find_inv_vars.
14492 (split_address_cost, determine_group_iv_cost_cond): Ditto.
14494 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14496 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
14497 inv_vars. Add inv_exprs.
14498 (struct iv_cand): Rename depends_on to inv_vars.
14499 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
14500 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
14501 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
14502 (dump_cand): Dump inv_vars.
14503 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
14504 (record_invariant, find_depends, add_candidate_1): Ditto.
14505 (set_group_iv_cost, force_var_cost): Ditto.
14506 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
14507 (get_computation_cost_at, get_computation_cost): Ditto.
14508 (determine_group_iv_cost_generic): Ditto.
14509 (determine_group_iv_cost_address): Ditto.
14510 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
14511 (determine_group_iv_costs): Ditto.
14512 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
14513 (iv_ca_set_remove_invariants): Renamed to ...
14514 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
14515 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
14516 (iv_ca_set_add_invariants): Renamed to ...
14517 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
14518 (iv_ca_set_cp): Use iv_ca_set_add_invs.
14519 (iv_ca_has_deps): Support inv_vars and inv_exprs.
14520 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
14521 (create_new_ivs): Remove useless dump.
14523 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14525 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
14527 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
14528 (iv_ca_set_no_cp, create_new_iv): Ditto.
14530 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14532 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14534 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
14536 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
14537 function tree_check2.
14539 2017-05-02 Martin Liska <mliska@suse.cz>
14541 * doc/gcov.texi: Add missing preposition.
14542 * gcov.c (function_info::function_info): Properly fill up
14543 all member variables.
14545 2017-05-02 Tamar Christina <tamar.christina@arm.com>
14547 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
14549 2017-05-02 Tamar Christina <tamar.christina@arm.com>
14551 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
14553 2017-05-02 Martin Liska <mliska@suse.cz>
14556 * lto-streamer-in.c (lto_read_tree_1): Remove
14557 LTO_STREAMER_DEBUG.
14558 * lto-streamer.c (struct tree_hash_entry): Likewise.
14559 (struct tree_entry_hasher): Likewise.
14560 (tree_entry_hasher::hash): Likewise.
14561 (tree_entry_hasher::equal): Likewise.
14562 (lto_streamer_init): Likewise.
14563 (lto_orig_address_map): Likewise.
14564 (lto_orig_address_get): Likewise.
14565 (lto_orig_address_remove): Likewise.
14566 * lto-streamer.h: Likewise.
14567 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
14568 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
14570 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
14572 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
14573 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
14574 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
14575 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
14576 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
14577 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
14578 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
14579 (mm_maskz_sub_ss): New intrinsics.
14580 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14581 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
14582 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
14583 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
14584 (__builtin_ia32_subss_mask_round): New builtins.
14585 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14586 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
14587 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
14589 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
14590 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
14592 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
14595 2017-05-02 Martin Jambor <mjambor@suse.cz>
14597 PR tree-optimization/78687
14598 * tree-sra.c (access): New field parent.
14599 (process_subtree_disqualification): New function.
14600 (disqualify_candidate): Call it.
14601 (build_accesses_from_assign): Reset write flag if creating an
14603 (build_access_subtree): Fill in parent field and also prpagate
14604 down grp_write flag.
14605 (create_artificial_child_access): New parameter set_grp_write, set
14606 grp_write to its value.
14607 (propagate_subaccesses_across_link): Also propagate grp_write flag
14609 (propagate_all_subaccesses): Push the closest parent back to work
14610 queue if add_access_to_work_queue returned true.
14612 2017-05-02 Richard Biener <rguenther@suse.de>
14614 * common.opt (fstrict-overflow): Alias negative to fwrapv.
14615 * doc/invoke.texi (fstrict-overflow): Remove all traces of
14616 -fstrict-overflow documentation.
14617 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
14618 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
14619 flag_strict_overflow.
14620 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
14621 * lto-opts.c (lto_write_options): Do not stream it.
14622 * lto-wrapper.c (merge_and_complain): Do not handle it.
14623 * opts.c (default_options_table): Do not set -fstrict-overflow.
14624 (finish_options): Likewise do not clear it when sanitizing.
14625 * simplify-rtx.c (simplify_const_relational_operation): Do not
14626 test flag_strict_overflow.
14628 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
14630 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
14631 using enabled attribute.
14632 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
14633 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
14634 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
14635 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
14636 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
14637 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
14638 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
14639 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
14640 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
14641 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
14643 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
14645 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
14647 2017-05-02 Richard Biener <rguenther@suse.de>
14649 PR tree-optimization/80591
14651 2017-04-10 Richard Biener <rguenther@suse.de>
14653 * tree-ssa-structalias.c (find_func_aliases): Properly handle
14656 2017-05-02 Richard Biener <rguenther@suse.de>
14658 PR tree-optimization/80549
14659 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
14660 (cleanup_tree_cfg_noloop): Create forwarders to known loop
14661 headers if they do not have a preheader.
14663 2017-05-02 Martin Liska <mliska@suse.cz>
14666 * common.opt: Fix typo.
14667 * doc/invoke.texi: Likewise.
14669 2017-05-01 Jan Beulich <jbeulich@suse.com>
14671 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
14672 swapping, add (x,x,m,x,n) alternative.
14674 2017-05-01 Nathan Sidwell <nathan@acm.org>
14676 * calls.c (combine_pending_stack_adjustment_and_call): Remove
14677 unnecessary unadjusted_alignment check.
14679 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
14682 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
14684 * gimplify.c (gimplify_cilk_detach): New function.
14685 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
14686 * tree-core.h: Document EXPR_CILK_SPAWN.
14687 * tree.h (EXPR_CILK_SPAWN): Define.
14689 2017-05-01 David Malcolm <dmalcolm@redhat.com>
14691 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
14692 to use new fixit_hint representation, using the "replace" logic.
14693 (get_line_span_for_fixit_hint): Likewise.
14694 (layout::print_any_fixits): Likewise.
14695 (selftest::test_one_liner_many_fixits): Rename to...
14696 (selftest::test_one_liner_many_fixits_1): ...this, and update
14697 comment and expected output to reflect that the multiple fix-it
14698 hints are now consolidated into one insertion.
14699 (selftest::test_one_liner_many_fixits_2): New test.
14700 (selftest::test_diagnostic_show_locus_one_liner): Update for
14702 (selftest::test_fixit_consolidation): Update for fix-it API
14704 * diagnostic.c (print_parseable_fixits): Likewise.
14705 * edit-context.c (edited_line::m_line_events): Convert from
14706 auto_vec <line_event *> to auto_vec <line_event>.
14707 (class line_event): Convert from abstract base class to a concrete
14708 class, taking over the role of replace_event.
14709 (class insert_event): Delete.
14710 (class replace_event): Rename to class line_event. Convert to
14712 (edit_context::add_fixits): Reimplement.
14713 (edit_context::apply_insert): Delete.
14714 (edit_context::apply_replace): Rename to...
14715 (edit_context::apply_fixit): ...this. Convert to half-open range.
14716 (edited_file::apply_insert): Delete.
14717 (edited_file::apply_replace): Rename to...
14718 (edited_file::apply_fixit): ...this.
14719 (edited_line::~edited_line): Drop deletion of events.
14720 (edited_line::apply_insert): Delete.
14721 (edited_line::apply_replace): Rename to...
14722 (edited_line::apply_fixit): ...this. Convert to half-open range.
14723 Update for change to type of m_line_events.
14724 * edit-context.h (edit_context::apply_insert): Delete.
14725 (edit_context::apply_replace): Rename to...
14726 (edit_context::apply_fixit): ...this.
14728 2017-05-01 Martin Sebor <msebor@redhat.com>
14730 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
14733 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
14736 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
14737 __get_cpuid_max returns 0.
14738 (__get_cpuid_count): Ditto.
14740 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
14742 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
14743 replacement expression is another instance of one of its arguments.
14745 2017-05-01 Jakub Jelinek <jakub@redhat.com>
14748 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
14749 check for stack push/pop autoinc.
14750 * config/i386/i386.c (ix86_agi_dependent): Return false
14751 if the only reason why modified_in_p returned true is that
14752 addr is SP based and set_insn is a push or pop.
14754 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
14756 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
14759 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
14762 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
14763 (account_size_time): Use two predicates - exec_pred and
14765 (evaluate_conditions_for_known_args): Compute both clause and
14767 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
14768 (inline_summary_t::duplicate): Update.
14769 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
14771 (compute_inline_parameters): Likewise.
14772 (estimate_edge_size_and_time): Update caluclation of time.
14773 (estimate_node_size_and_time): Compute both time and nonspecialized
14775 (estimate_ipcp_clone_size_and_time): Update.
14776 (inline_merge_summary): Update.
14777 (do_estimate_edge_time): Update.
14778 (do_estimate_edge_size): Update.
14779 (do_estimate_edge_hints): Update.
14780 (inline_read_section, inline_write_summary): Stream both new predicates.
14781 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
14783 (compute_inlined_call_time): Cleanup.
14784 (big_speedup_p): Update.
14785 (edge_badness): Update.
14786 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
14787 (size_time_entry): Replace predicate by exec_predicate and
14788 nonconst_predicate.
14789 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
14790 (estimate_edge_time): Return also nonspec_time.
14791 (reset_edge_growth_cache): Update.
14793 2017-04-29 Jakub Jelinek <jakub@redhat.com>
14795 PR rtl-optimization/80491
14796 * ifcvt.c (noce_process_if_block): When looking for x setter
14797 with missing else_bb, don't check only the insn right before
14798 cond_earliest, but look for the last insn that x is modified in
14799 within the same bb.
14801 PR rtl-optimization/80491
14802 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
14804 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
14806 PR tree-optimization/80487
14807 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
14809 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14811 PR tree-optimization/79697
14812 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
14813 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
14814 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
14816 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
14817 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
14819 2017-04-28 Martin Sebor <msebor@redhat.com>
14821 PR tree-optimization/80523
14822 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
14823 (init_target_to_host_charmap, target_to_host, target_strtol10): New
14825 (maybe_warn, format_directive, parse_directive): Use new functions.
14826 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
14828 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
14830 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
14832 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
14834 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
14835 target_header_dir): Set correctly.
14836 * configure: Regenerated.
14837 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
14838 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
14839 instead of SYSTEM_HEADER_DIR.
14841 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
14843 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
14844 (estimate_local_effects): Likewise.
14845 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
14846 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
14847 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
14848 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
14849 do_estimate_edge_time, estimate_edge_time): Likewise.
14850 * ipa-inline-analysis.c (estimate_node_size_and_time,
14851 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
14852 (estimate_time_after_inlining): Remove.
14854 2017-04-28 Martin Liska <mliska@suse.cz>
14856 * doc/gcov.texi: Enhance documentation of gcov.
14858 2017-04-28 Martin Liska <mliska@suse.cz>
14860 * doc/gcov.texi: Sort options in alphabetic order.
14861 * doc/gcov-dump.texi: Likewise.
14862 * doc/gcov-tool.texi: Likewise.
14863 * gcov.c (print_usage): Likewise.
14864 * gcov-dump.c (print_usage): Likewise.
14865 * gcov-tool.c (print_merge_usage_message): Likewise.
14866 (print_rewrite_usage_message): Likewise.
14867 (print_overlap_usage_message): Likewise.
14869 2017-04-28 Martin Liska <mliska@suse.cz>
14871 PR gcov-profile/53915
14872 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
14874 2017-04-28 Martin Liska <mliska@suse.cz>
14876 PR gcov-profile/79891
14877 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
14878 is marked by compiler as living on a line.
14879 (get_cycles_count): Remove usage of the union.
14880 (output_intermediate_file): Likewise.
14881 (find_source): Fix GNU coding style.
14882 (accumulate_line_counts): Remove old non-all block mode.
14883 (output_lines): Remove usage of the union.
14884 * profile.c (output_location): Include all BBs, even if
14885 belonging to a same line (and file) as a previous BB.
14887 2017-04-28 Martin Liska <mliska@suse.cz>
14889 * gcov.c (process_args): Handle new argument 'w'.
14890 (read_graph_file): Assign ID to BBs.
14891 (output_branch_count): Display BB # if verbose flag is set.
14892 (output_lines): Likewise for arcs.
14893 (print_usage): Add '--verbose' option help.
14894 * doc/gcov.texi: Document --verbose (-w) option.
14896 2017-04-28 Martin Liska <mliska@suse.cz>
14898 * gcov.c (struct block_location_info): New struct.
14899 (process_file): Fill up the new structure.
14900 (read_graph_file): Replace usage of encoding by the newly added
14902 (add_line_counts): Likewise.
14903 (accumulate_line_counts): Remove usage of the union.
14904 (function_info::function_info): New function.
14905 (function_info::~function_info): Likewise.
14906 (process_file): Call delete instead of release_function.
14907 (release_function): Release the function.
14908 (release_structures): Call delete instead of release_function.
14909 (solve_flow_graph): Replace usage of num_blocks.
14910 (find_exception_blocks): Likewise.
14911 (output_lines): Fix GNU coding style.
14913 2017-04-28 Martin Liska <mliska@suse.cz>
14916 * coverage.c (coverage_remove_note_file): New function.
14917 * coverage.h: Declare the function.
14918 * toplev.c (finalize): Clean if an error has been seen.
14920 2017-04-28 Martin Liska <mliska@suse.cz>
14922 PR gcov-profile/80031
14923 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
14924 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
14925 * gcov.c (read_graph_file): Read just number of blocks.
14926 * profile.c (branch_prob): Do not stream 0 flags per a basic
14929 2017-04-28 Martin Liska <mliska@suse.cz>
14931 * gcov-dump.c (tag_*): Add new argument to declarations.
14932 (dump_gcov_file): Likewise.
14933 (tag_blocks): Add and use new argument depth.
14934 (tag_arcs): Likewise.
14935 (tag_lines): Likewise.
14936 (tag_counters): Likewise.
14937 (tag_summary): Likewise.
14938 (dump_working_sets): Use depth to do a proper indentation.
14940 2017-04-28 Jakub Jelinek <jakub@redhat.com>
14943 * cgraph.h (symtab_node::debug_symtab): No longer inline.
14944 * symtab.c (symtab_node::debug_symtab): Move definition here.
14946 2017-04-28 Richard Biener <rguenther@suse.de>
14948 * lto-streamer.h (LTO_major_version): Bump to 7.
14950 2017-04-28 Richard Biener <rguenther@suse.de>
14952 * tree-vrp.c (assert_info): New struct.
14953 (add_assert_info): New helper.
14954 (register_edge_assert_for_2): Refactor to add asserts to a vector
14956 (register_edge_assert_for_1): Likewise.
14957 (register_edge_assert_for): Likewise.
14958 (finish_register_edge_assert_for): New helper actually registering
14959 asserts where live on edge.
14960 (find_conditional_asserts): Adjust.
14961 (find_switch_asserts): Likewise.
14962 (evrp_dom_walker::try_find_new_range): Generalize.
14963 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
14965 2017-04-27 Marek Polacek <polacek@redhat.com>
14968 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
14969 arg10 and arg11 to itype.
14971 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
14973 * doc/extend.texi (Object Size Checking): Improve grammar.
14975 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
14978 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
14979 that the logic for permitting reciprocal estimates matches that
14982 2017-04-27 Jakub Jelinek <jakub@redhat.com>
14985 * tree.c (type_cache_hasher::equal): Only compare
14986 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
14987 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
14988 non-aggregate element types.
14989 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
14990 about the flag on ARRAY_TYPEs in the comment, formatting fix.
14992 2017-04-27 Richard Biener <rguenther@suse.de>
14994 PR middle-end/80533
14995 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
14996 stripping ARRAY_REFs from MEM_EXPR make sure we're not
14997 keeping a reference to a trailing array.
14999 2017-04-27 Richard Biener <rguenther@suse.de>
15001 PR middle-end/80539
15002 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
15003 being in loop-closed SSA form conservatively.
15004 (chrec_fold_multiply_poly_poly): Likewise.
15006 2017-04-27 Tamar Christina <tamar.christina@arm.com>
15008 PR middle-end/79665
15009 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
15010 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
15012 2017-04-27 Jakub Jelinek <jakub@redhat.com>
15015 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
15016 (aarch64_function_arg_alignment): Return unsigned int again, but still
15017 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
15018 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
15019 Don't emit -Wpsabi note.
15020 (aarch64_function_arg_boundary): Likewise.
15021 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
15024 2017-04-26 Nathan Sidwell <nathan@acm.org>
15026 * tree.h (crc32_unsigned_n): Declare.
15027 (crc32_unsigned, crc32_unsigned): Make inline.
15028 * tree.c (crc32_unsigned_bits): Replace with ...
15029 (crc32_unsigned_n): ... this.
15030 (crc32_unsigned, crc32_byte): Remove.
15031 (crc32_string): Remove unnecessary braces.
15033 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15035 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
15036 * ipa-inline-analysis.c (MAX_TIME): Remove.
15037 (account_size_time): Use sreal for time.
15038 (dump_inline_summary): Update.
15039 (estimate_function_body_sizes): Update.
15040 (estimate_edge_size_and_time): Update.
15041 (estimate_calls_size_and_time): Update.
15042 (estimate_node_size_and_time): Update.
15043 (inline_merge_summary): Update.
15044 (inline_update_overall_summary): Update.
15045 (estimate_time_after_inlining): Update.
15046 (inline_read_section): Update.
15047 (inline_write_summary): Update.
15048 * ipa-inline.c (compute_uninlined_call_time): Update.
15049 (compute_inlined_call_time): Update.
15050 (recursive_inlining): Update.
15051 (inline_small_functions): Update.
15052 (dump_overall_stats): Update.
15053 * ipa-inline.h: Include sreal.h.
15054 (size_time_entry): Turn time to sreal.
15055 (inline_summary): Turn self_time nad time to sreal.
15057 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15059 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
15061 (sreal::stream_out, sreal::stream_in): New.
15062 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
15064 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15066 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
15069 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
15072 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
15074 (dimode_scalar_chain::compute_convert_gain): Ditto.
15075 (dimode_scalar_chain::make_vector_copies): Ditto.
15076 (dimode_scalar_chain::convert_reg): Ditto.
15077 (dimode_scalar_chain::convert_insn): Ditto.
15078 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
15079 (VI248_AVX512BW_1): New mode iterator.
15080 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
15081 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
15084 2017-04-25 Martin Sebor <msebor@redhat.com>
15086 PR tree-optimization/80497
15087 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
15088 constants are representable in HOST_WIDE_INT.
15089 (parse_directive): Ditto.
15091 2017-04-25 Martin Sebor <msebor@redhat.com>
15094 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
15095 (new_zero_array): Adjust signature.
15096 (dom_info::dom_init): Used unsigned rather that size_t.
15097 (dom_info::dom_info): Same.
15099 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15100 Jakub Jelinek <jakub@redhat.com>
15103 * config/arm/arm.c: Include gimple.h.
15104 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
15105 returns negative, increment ncrn only if it returned positive.
15106 (arm_needs_doubleword_align): Return int instead of bool,
15107 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
15108 members, but if there is any such non-FIELD_DECL
15109 > PARM_BOUNDARY aligned decl, return -1 instead of false.
15110 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
15111 returns negative, increment nregs only if it returned positive.
15112 (arm_setup_incoming_varargs): Likewise.
15113 (arm_function_arg_boundary): Emit -Wpsabi note if
15114 arm_needs_doubleword_align returns negative, return
15115 DOUBLEWORD_ALIGNMENT only if it returned positive.
15117 2017-04-25 Marek Polacek <polacek@redhat.com>
15120 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
15121 first argument to type.
15123 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
15126 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
15127 type checks to test for compatibility instead of equality.
15129 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15130 Jakub Jelinek <jakub@redhat.com>
15133 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
15135 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
15136 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
15137 the alignment computation, but return their maximum in warn_alignment.
15138 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
15139 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
15141 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
15142 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
15145 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15147 * config/arc/simdext.md (dmpyh): Fix typo.
15149 2017-04-25 Richard Biener <rguenther@suse.de>
15151 PR tree-optimization/80492
15152 * alias.c (compare_base_decls): Handle registers with asm
15153 specification conservatively.
15154 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
15155 compare_base_decls returning dont-know properly.
15157 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15159 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
15160 (legitimate_offset_address_p): New function.
15161 (arc_legitimate_address_p): Use above function.
15163 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15165 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
15167 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15169 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
15170 ACCH registers whenever they are available.
15172 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15174 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
15175 double regs fix when not used.
15177 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15179 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
15181 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
15182 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
15184 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15186 * config/arc/arc.c (arc_output_addsi): Check for h-register class
15187 when emitting short ADD instructions.
15189 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15191 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
15193 (cmpsi_cc_c_insn): Likewise.
15194 (cbranchsi4_scratch): Compute proper instruction length using
15195 compact_hreg_operand.
15196 * config/arc/predicates.md (compact_hreg_operand): New predicate.
15198 2017-04-25 Richard Biener <rguenther@suse.de>
15200 PR middle-end/80509
15201 * passes.c (pass_manager::pass_manager): Initialize
15202 m_name_to_pass_map.
15204 2017-04-25 Richard Biener <rguenther@suse.de>
15206 PR tree-optimization/79201
15207 * tree-ssa-sink.c (statement_sink_location): Handle calls.
15209 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15212 * config/s390/vector.md: Split MEM->GPR vector moves for
15213 non-s_operand addresses.
15215 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15218 * config/s390/predicates.md (reload_const_wide_int_operand): New
15220 * config/s390/s390.md ("movti"): Remove d/P alternative.
15221 ("movti_bigconst"): New pattern definition.
15223 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15226 * s390-protos.h (s390_expand_cs_hqi): Removed.
15227 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
15228 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
15229 modes as well as CCZ1mode and CCZmode.
15230 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
15231 signature of s390_emit_compare_and_swap.
15232 (s390_expand_cs_hqi): Likewise, make static.
15233 (s390_expand_cs_tdsi): Generate an explicit compare before trying
15234 compare-and-swap, in some cases.
15235 (s390_expand_cs): Wrapper function.
15236 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
15238 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
15239 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
15240 patterns for small and large integers. Forbid symref memory operands.
15241 Move expander to s390.c. Require cc register.
15242 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
15243 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
15244 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
15245 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
15246 symref memory operands. Remove CC mode and call s390_match_ccmode
15248 ("atomic_exchange<mode>"): Allow and implement all integer modes.
15250 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15252 * config/s390/s390.md (define_peephole2): New peephole to help
15253 combining the load-and-test pattern with volatile memory.
15255 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15257 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
15258 with CCZmode for TARGET_Z196.
15260 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15262 PR rtl-optimization/80501
15263 * combine.c (make_compound_operation_int): Set subreg_code to SET
15264 even for AND with mask of the sign bit of mode.
15266 PR rtl-optimization/80500
15267 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
15268 sum's initial value.
15270 2017-04-25 Julian Brown <julian@codesourcery.com>
15271 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15273 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
15275 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
15277 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
15279 2017-04-25 Julian Brown <julian@codesourcery.com>
15280 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15282 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
15283 (thunderx2t99_sha): New Reservation.
15285 2017-04-25 Julian Brown <julian@codesourcery.com>
15286 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15288 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
15289 type for 1-element load.
15291 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
15293 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
15295 2017-04-24 Martin Jambor <mjambor@suse.cz>
15297 PR tree-optimization/80293
15298 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
15299 char arrays not totally scalarizable if it is false.
15300 (analyze_all_variable_accesses): Pass correct value in the new
15301 parameter. Add a statistics counter.
15303 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
15305 PR middle-end/79931
15306 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
15308 2017-04-24 Richard Biener <rguenther@suse.de>
15310 PR tree-optimization/80494
15311 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
15312 out for complex types.
15314 2017-04-24 Richard Biener <rguenther@suse.de>
15316 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
15317 * tree-ssa-sccvn.c (print_scc): Print SCC size.
15318 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
15319 (DFS): Adjust and never fail.
15320 (sccvn_dom_walker::fail): Remove.
15321 (sccvn_dom_walker::before_dom_children): Adjust.
15322 (run_scc_vn): Likewise and never fail.
15323 * tree-ssa-pre.c (pass_pre::execute): Adjust.
15324 (pass_fre::execute): Likewise.
15326 2017-04-24 Richard Biener <rguenther@suse.de>
15328 PR tree-optimization/79725
15329 * tree-ssa-sink.c (statement_sink_location): Return whether
15330 failure reason was zero uses. Move that check later.
15331 (sink_code_in_bb): Deal with zero uses by removing the stmt
15334 2017-04-24 Richard Biener <rguenther@suse.de>
15337 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
15338 pointer-based references.
15340 2017-04-24 Richard Biener <rguenther@suse.de>
15343 * pass_manager.h (pass_manager::operator new): Remove.
15344 (pass_manager::operator delete): Likewise.
15345 * passes.c (pass_manager::operator new): Remove.
15346 (pass_manager::operator delete): Likewise.
15347 (pass_manager::pass_manager): Zero individual pass members.
15349 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
15352 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
15353 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
15354 Check "XEXP (src, 1)" operand here.
15355 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
15356 Check "XEXP (src, 1)" operand here.
15357 (dimode_scalar_chain::make_vector_copies): Detect count register
15358 of a shift instruction. Zero extend count register from QImode
15359 to DImode to satisfy vector shift pattern count operand predicate.
15360 Substitute vector shift count operand with a DImode copy.
15361 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
15364 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
15366 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
15367 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15368 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
15369 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15370 (UNSPEC_NOREX_MEM): Remove definition.
15372 2017-04-21 Richard Biener <rguenther@suse.de>
15374 PR tree-optimization/79547
15375 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15376 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
15377 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
15378 without any constraints.
15380 2017-04-21 Richard Biener <rguenther@suse.de>
15382 PR tree-optimization/78847
15383 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
15385 2017-04-21 Richard Biener <rguenther@suse.de>
15387 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
15388 (build_distinct_type_copy): Likewise.
15389 (build_variant_type_copy): Likewise.
15390 * tree.c (build_qualified_type): Pass down mem-stat info.
15391 (build_distinct_type_copy): Likewise.
15392 (build_variant_type_copy): Likewise.
15394 2017-04-21 Richard Biener <rguenther@suse.de>
15396 PR tree-optimization/80237
15397 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
15399 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
15400 for a simplified result.
15402 2016-04-21 Richard Biener <rguenther@suse.de>
15404 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
15405 sth as strict as a simple_iv but a chrec without symbols and an
15406 operand defined in the loop we are peeling (and not some subloop).
15407 (propagate_constants_for_unrolling): Propagate all constants.
15409 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15412 * config/i386/i386.c (print_reg): Remove assert for disalowed
15413 regno values, call output_operand_lossage instead.
15415 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15418 * config/i386/constraints.md (Yc): New register constraint.
15419 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
15420 Use Yc constraint for alternative 2 of operand 0. Remove
15421 preferred_for_speed attribute.
15423 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15425 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
15426 lastprivate clauses in SIMT case.
15428 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
15430 * doc/invoke.texi (-Wextra-semi): Document new warning option.
15432 2017-04-20 Richard Biener <rguenther@suse.de>
15434 PR tree-optimization/57796
15435 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
15436 as N scalar stores.
15437 (vect_model_load_cost): Cost gathers as N scalar loads.
15439 2017-04-20 Richard Biener <rguenther@suse.de>
15441 * ggc-page.c (ggc_allocated_p): Rename to ...
15442 (safe_lookup_page_table_entry): ... this and return the lookup
15444 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
15446 2017-04-20 Richard Biener <rguenther@suse.de>
15448 PR tree-optimization/80453
15449 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
15450 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
15451 from the conditions.
15452 (vn_phi_eq): Pass them down.
15453 (vn_phi_lookup): Record them.
15454 (vn_phi_insert): Likewise.
15456 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
15458 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
15459 uninitialized variable warning to avoid buffer overrun.
15461 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15464 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
15465 is suppressed for '{ 0 }' in C.
15467 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15469 * BASE-VER: Set to 8.0.0.
15471 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
15473 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
15474 priority .init_array and .fini_array section with SECTION_NOTYPE
15477 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15479 PR middle-end/80423
15480 * tree.h (build_array_type): Add typeless_storage default argument.
15481 * tree.c (type_cache_hasher::equal): Also compare
15482 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
15483 (build_array_type): Add typeless_storage argument, set
15484 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
15486 (build_nonshared_array_type): Adjust build_array_type_1 caller.
15487 (build_array_type): Likewise. Add typeless_storage argument.
15489 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
15490 Jakub Jelinek <jakub@redhat.com>
15492 PR tree-optimization/80426
15493 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
15494 operation on symbolic operands, also compute the overflow for the
15495 invariant part when the operation degenerates into a negation.
15497 2017-04-19 Jakub Jelinek <jakub@redhat.com>
15500 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
15501 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
15504 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
15506 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
15509 * config/avr/avr.c (tree.h): Include it.
15510 (cgraph.h): Include it.
15511 (avr_encode_section_info): Don't warn for uninitialized progmem
15512 variable if it's just an alias.
15514 2017-04-19 Richard Biener <rguenther@suse.de>
15517 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
15518 when needed by AutoPGO.
15520 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
15523 * doc/lto.texi: Remove an extra 'that'.
15525 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
15527 PR rtl-optimization/80429
15528 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
15529 are only used in debug insns.
15531 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
15532 Vladimir Makarov <vmakarov@redhat.com>
15534 * config/sparc/predicates.md (input_operand): Add comment. Return
15535 true for any memory operand when LRA is in progress.
15536 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
15538 2017-04-18 Jeff Law <law@redhat.com>
15541 * mips.md ({return,simple_return}_internal): Do not overwrite
15544 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15546 PR tree-optimization/80443
15547 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
15548 instead of adding 1, subtract -1 and similarly instead of subtracting
15551 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
15553 PR rtl-optimization/80357
15554 * haifa-sched.c (tmp_bitmap): New variable.
15555 (model_recompute): Handle duplicate use records.
15556 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
15557 (free_global_sched_pressure_data): Free it.
15559 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
15562 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
15563 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15564 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15565 instead of SYSTEM_HEADER_DIR.
15567 2017-04-18 Jeff Law <law@redhat.com>
15569 PR middle-end/80422
15570 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
15571 predecessors after walking up the insn chain.
15573 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15576 * dwarf2out.c (modified_type_die): Try harder not to emit internal
15577 sizetype type into debug info.
15579 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15582 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
15583 unneeded test for TARGET_UPPER_REGS_SF.
15584 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
15586 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15589 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
15590 instead of gsi_after_labels.
15592 2017-04-18 Jeff Law <law@redhat.com>
15594 * regcprop.c (maybe_mode_change): Avoid creating copies of the
15598 2017-04-13 Jeff Law <law@redhat.com>
15599 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
15600 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
15602 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
15605 * config/avr/avr.c (intl.h): Include it.
15606 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
15608 2017-04-18 Martin Liska <mliska@suse.cz>
15610 PR gcov-profile/78783
15611 * gcov-tool.c (gcov_output_files): Validate that destination
15612 file is either removed by the tool or by a user.
15614 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
15615 Guy Benyei <guybe@mellanox.com>
15617 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
15618 block, and do not negate it, the stored id is already negative.
15620 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
15622 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
15624 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15627 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
15628 masks of options that should be turned off if the VSX vector
15629 options are turned off.
15630 (OTHER_P8_VECTOR_MASKS): Likewise.
15631 (OTHER_VSX_VECTOR_MASKS): Likewise.
15632 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
15633 rs6000_disable_incompatible_switches to validate no type switches
15635 (rs6000_incompatible_switch): New function to disallow turning on
15636 other vector options if -mno-vsx, -mno-power8-vector, or
15637 -mno-power9-vector are specified.
15639 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15641 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
15643 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15645 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
15646 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
15647 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
15648 (ARG_POINTER_CFA_OFFSET): Likewise.
15650 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15652 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
15653 conditions to take advantage of various optimizations.
15655 2017-04-13 Jeff Law <law@redhat.com>
15657 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
15658 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
15659 (zero_extendsidi2_dext): Likewise.
15661 2017-04-13 Jakub Jelinek <jakub@redhat.com>
15664 * fold-const.c (fold_ternary_loc): Revert
15665 use op0 instead of fold_convert_loc (loc, type, arg0) part of
15668 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
15670 PR rtl-optimization/80343
15671 * lra-remat.c (update_scratch_ops): Assign original hard reg to
15672 new scratch pseudo.
15674 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
15677 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
15678 to ubsan_encode_value.
15680 2017-04-13 Jeff Law <law@redhat.com>
15682 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
15683 appearing in DEBUG_INSNs.
15685 2017-04-13 Martin Liska <mliska@suse.cz>
15687 PR gcov-profile/80413
15688 * gcov-io.c (gcov_write_string): Copy to buffer just when
15689 allocated size is greater than zero.
15691 2017-04-13 Jakub Jelinek <jakub@redhat.com>
15694 * dwarf2out.c (decls_for_scope): Ignore declarations of
15695 current_function_decl in BLOCK_NONLOCALIZED_VARS.
15697 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
15700 * ipa-visibility.c (non_local_p): Fix typos.
15701 (localize_node): When localizing symbol in same comdat group,
15702 dissolve the group only when we know external symbols are going
15704 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
15706 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15708 PR tree-optimization/79390
15709 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
15710 order does not result in usable sequence, retry with reversed operand
15716 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
15717 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
15718 op0 instead of fold_convert_loc (loc, type, arg0).
15720 2017-04-12 Jeff Law <law@redhat.com>
15722 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
15723 has a delay slot in the generated code.
15725 * config/cris/cris.md (cris_preferred_reload_class): Return
15726 GENNONACR_REGS rather than GENERAL_REGS.
15728 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15731 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
15732 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
15733 signedness of the result type.
15735 2017-04-12 Richard Biener <rguenther@suse.de>
15736 Jeff Law <law@redhat.com>
15738 PR tree-optimization/80359
15739 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
15740 trim stores to TARGET_MEM_REFs.
15742 2017-04-12 Richard Biener <rguenther@suse.de>
15744 PR tree-optimization/79390
15745 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
15746 threading case even more.
15748 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
15751 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
15752 for quad_address_p for TImode, instead of just not indexed_address.
15754 2017-04-12 Richard Biener <rguenther@suse.de>
15755 Bernd Edlinger <bernd.edlinger@hotmail.de>
15757 PR middle-end/79671
15758 * alias.c (component_uses_parent_alias_set_from): Handle
15759 TYPE_TYPELESS_STORAGE.
15760 (get_alias_set): Likewise.
15761 * tree-core.h (tree_type_common): Add typeless_storage flag.
15762 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
15763 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
15764 for types containing members with TYPE_TYPELESS_STORAGE.
15765 (place_field): Likewise.
15766 (layout_type): Likewise for ARRAY_TYPE.
15767 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
15768 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
15769 TYPE_TYPELESS_STORAGE.
15770 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
15772 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15775 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
15776 first argument to type.
15778 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15782 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
15783 CONST0_RTX (mode) rather than const0_rtx where appropriate.
15784 (rs6000_expand_binop_builtin): Likewise.
15785 (rs6000_expand_ternop_builtin): Likewise; also add missing
15786 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
15787 vshasigma built-ins.
15788 * doc/extend.texi: Document that vec_xxpermdi's third argument
15789 must be a constant.
15791 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
15793 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
15794 Use shift_const cost parameter when calculating gain of STV shifts.
15796 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
15798 PR rtl-optimization/70478
15799 * lra-constraints.c (process_alt_operands): Check memory for
15800 disfavoring memory insn operand.
15802 2017-04-11 Jakub Jelinek <jakub@redhat.com>
15804 PR middle-end/80100
15805 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
15806 left shift in unsigned HOST_WIDE_INT type.
15808 PR rtl-optimization/80385
15809 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
15810 (not (neg X)) into (plus X -1) for complex or non-integral modes.
15813 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
15814 if they have any depend clauses.
15816 2017-04-11 Martin Liska <mliska@suse.cz>
15819 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
15820 * ipa-split.c (split_function): Create a local comdat symbol
15821 if caller is in a comdat group.
15823 2017-04-11 Martin Liska <mliska@suse.cz>
15826 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
15829 2017-04-11 Martin Sebor <msebor@redhat.com>
15831 PR middle-end/80364
15832 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
15833 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
15835 (directive::set_width, directive::set_precision, format_character):
15837 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
15840 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
15843 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
15844 conflict, set target->arch_name instead of target->cpu_name.
15846 2017-04-11 Richard Biener <rguenther@suse.de>
15848 PR tree-optimization/80374
15849 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
15850 build_zero_cst, remove fold_convertible_p check again.
15852 2017-04-11 Martin Liska <mliska@suse.cz>
15855 * ubsan.c (instrument_object_size): Do not instrument register
15858 2017-04-11 Jakub Jelinek <jakub@redhat.com>
15861 * config/i386/i386-builtin-types.def
15862 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
15863 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
15864 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
15865 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
15866 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
15867 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
15868 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
15869 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
15870 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
15871 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
15872 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
15873 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
15874 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
15875 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
15876 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
15877 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
15878 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
15879 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
15880 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
15881 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
15882 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
15883 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
15884 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
15885 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
15886 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
15887 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
15888 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
15889 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
15890 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
15891 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
15892 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
15893 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
15894 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
15895 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
15896 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
15897 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
15898 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
15899 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
15900 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
15901 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
15902 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
15903 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
15904 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
15905 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
15906 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
15907 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
15909 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
15910 flag to second_arg_count, handle 4 argument function type _COUNT
15911 aliases, handle second_arg_count on second argument rather than last.
15913 2017-04-10 Jeff Law <law@redhat.com>
15915 PR tree-optimization/80374
15916 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
15917 record anything if we can not convert integer_zero_node to the
15920 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
15923 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15924 Enhance special handling given to the TARGET_P9_MINMAX option in
15925 relation to certain other options.
15927 2017-04-10 Bin Cheng <bin.cheng@arm.com>
15929 PR tree-optimization/80153
15930 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
15931 remove POINTER_PLUS_EXPR's base part directly, rather than through
15934 2017-04-10 Richard Biener <rguenther@suse.de>
15935 Bin Cheng <bin.cheng@arm.com>
15937 PR tree-optimization/80153
15938 * tree-affine.c (aff_combination_to_tree): Get base pointer from
15939 the first element of pointer type aff_tree. Build result expr in
15941 (add_elt_to_tree): Convert to type unconditionally. Remove other
15942 fold_convert calls.
15943 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
15944 (rewrite_use_nonlinear_expr): Check invariant using iv information.
15946 2017-04-10 Richard Biener <rguenther@suse.de>
15948 * tree-ssa-structalias.c (find_func_aliases): Properly handle
15951 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
15953 PR rtl-optimization/70478
15954 * lra-constraints.c (curr_small_class_check): New.
15955 (update_and_check_small_class_inputs): New.
15956 (process_alt_operands): Update curr_small_class_check. Disfavor
15957 alternative insn memory operands. Check available regs for small
15960 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
15963 * config/mips/mips.opt (-mvirt): Update description.
15964 * doc/invoke.texi (-mvirt): Likewise.
15966 2017-04-10 Richard Biener <rguenther@suse.de>
15968 PR middle-end/80362
15969 * fold-const.c (fold_binary_loc): Look at unstripped ops when
15970 looking for NEGATE_EXPR in -A / -B to A / B folding.
15972 2017-04-10 Martin Liska <mliska@suse.cz>
15974 PR gcov-profile/80224
15975 * gcov.c (print_usage): Fix usage string.
15976 (get_gcov_intermediate_filename): Remove.
15977 (output_gcov_file): Use both for normal and intermediate format.
15978 (generate_results): Do not initialize special file for
15979 intermediate format.
15981 2017-04-10 Richard Biener <rguenther@suse.de>
15983 PR tree-optimization/80304
15984 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
15987 2017-04-10 Nathan Sidwell <nathan@acm.org>
15990 * config/rs6000/rs6000.c (rs6000_vector_type): New.
15991 (rs6000_init_builtins): Use it.
15993 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15995 * config/arm/arm.md (<mrc>): Add mode to SET source.
15996 (<mrrc>): Likewise.
15998 2017-04-10 Richard Biener <rguenther@suse.de>
16000 PR middle-end/80344
16001 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
16003 2017-04-10 Jakub Jelinek <jakub@redhat.com>
16006 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
16007 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
16008 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
16009 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
16010 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
16011 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
16012 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
16013 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
16014 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
16015 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
16016 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
16017 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
16018 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
16019 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
16020 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
16021 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
16022 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
16023 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
16024 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
16025 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
16026 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
16027 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
16028 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
16030 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
16032 PR rtl-optimization/70478
16033 * lra-constraints.c: Reverse the last patch.
16035 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
16037 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
16038 Add comment for WCHAR_T.
16040 2017-04-08 Martin Liska <mliska@suse.cz>
16043 2017-04-07 Martin Liska <mliska@suse.cz>
16046 * ipa-split.c (split_function): Add function part to a same comdat
16049 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16052 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
16054 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
16056 * rs6000/rs6000.c (vec_load_pendulum): Rename...
16057 (vec_pairing): ...to this.
16058 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
16059 (rs6000_sched_init): Adjust for name change.
16060 (struct rs6000_sched_context): Likewise.
16061 (rs6000_init_sched_context): Likewise.
16062 (rs6000_set_sched_context): Likewise.
16064 2017-04-07 Jakub Jelinek <jakub@redhat.com>
16070 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
16072 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
16073 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
16074 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
16076 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
16078 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
16080 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
16082 PR rtl-optimization/70703
16083 * ira-color.c (update_conflict_hard_regno_costs): Use
16084 int64_t instead of HOST_WIDE_INT.
16086 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
16088 PR rtl-optimization/70478
16089 * lra-constraints.c (process_alt_operands): Disfavor alternative
16090 insn memory operands.
16092 2017-04-07 Jeff Law <law@redhat.com>
16094 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
16095 CALL and NOTE_INSN_CALL_ARG_LOCATION.
16097 2017-04-07 Martin Liska <mliska@suse.cz>
16100 * config/aarch64/aarch64.c (aarch64_process_target_attr):
16101 Show error message instead of an ICE.
16103 2017-04-07 Martin Liska <mliska@suse.cz>
16106 * ipa-split.c (split_function): Add function part to a same comdat
16109 2017-04-07 Richard Biener <rguenther@suse.de>
16111 PR middle-end/80341
16112 * tree.c (get_unwidened): Also handle ! for_type case for
16114 * convert.c (do_narrow): Split out from ...
16115 (convert_to_integer_1): ... here. Do not pass final truncation
16116 type to get_unwidened for TRUNC_DIV_EXPR.
16118 2017-04-07 Richard Biener <rguenther@suse.de>
16120 * tree-affine.c (wide_int_ext_for_comb): Take type rather
16122 (aff_combination_const): Adjust.
16123 (aff_combination_scale): Likewise.
16124 (aff_combination_add_elt): Likewise.
16125 (aff_combination_add_cst): Likewise.
16126 (aff_combination_convert): Likewise.
16127 (add_elt_to_tree): Likewise. Remove unused argument.
16128 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
16130 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
16132 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
16134 * config/arm/arm.c (arm_default_short_enums): Use
16135 ARM_DEFAULT_SHORT_ENUMS.
16136 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
16138 2017-04-06 Jakub Jelinek <jakub@redhat.com>
16141 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
16142 members with redundant out-of-class redeclaration.
16144 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
16147 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
16148 * config/i386/i386.md (*zero_extendsidi2):
16149 Add (?*x,*x) and (?*v,*v) alternatives.
16151 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
16154 * config/i386/i386.c (ix86_expand_builtin)
16155 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
16156 mode from insn data. Convert operands to insn operand mode.
16157 Copy operands that don't satisfy insn predicate to a register.
16159 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
16161 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
16164 2017-04-06 Richard Biener <rguenther@suse.de>
16166 PR tree-optimization/80334
16167 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
16168 preserve alignment of accesses.
16170 2017-04-06 Richard Biener <rguenther@suse.de>
16172 PR tree-optimization/80262
16173 * tree-sra.c (build_ref_for_offset): Preserve address-space
16175 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
16176 Drop useless address-space information on MEM_REF offsets.
16178 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
16180 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
16182 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
16184 PR rtl-optimization/70703
16185 * ira-color.c (update_conflict_hard_regno_costs): Use
16186 HOST_WIDE_INT instead of long.
16188 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
16191 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
16192 not defined for x86_64 target. Add -mmmx target option when __SSE2__
16194 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
16195 for x86_64 target. Handle -m3dnowa option.
16197 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
16199 PR rtl-optimization/70703
16200 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
16201 (update_conflict_hard_regno_costs): Use long instead of unsigned
16202 arithmetic for cost calculation.
16204 2017-04-05 Jakub Jelinek <jakub@redhat.com>
16205 Bernd Edlinger <bernd.edlinger@hotmail.de>
16208 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
16211 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
16214 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
16215 ptr_mode with Pmode throughout.
16216 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
16217 into probe_stack_range and use DImode.
16219 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
16222 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
16223 call_eh_return is true.
16225 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16227 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
16228 Initialize last_match_fntype_index.
16230 2017-04-05 Jakub Jelinek <jakub@redhat.com>
16233 * tree-nvr.c: Include internal-fn.h.
16234 (pass_return_slot::execute): Ignore internal calls without
16237 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16238 Richard Biener <rguenther@suse.de>
16241 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
16242 captures used multiple times, except for the last use.
16243 * generic-match-head.c: Include gimplify.h.
16245 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16247 PR tree-optimization/79390
16248 * target.h (struct noce_if_info): Declare.
16249 * targhooks.h (default_noce_conversion_profitable_p): Declare.
16250 * target.def (noce_conversion_profitable_p): New target hook.
16251 * ifcvt.h (struct noce_if_info): New type, moved from ...
16252 * ifcvt.c (struct noce_if_info): ... here.
16253 (noce_conversion_profitable_p): Renamed to ...
16254 (default_noce_conversion_profitable_p): ... this. No longer
16256 (noce_try_store_flag_constants, noce_try_addcc,
16257 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
16258 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
16259 instead of noce_conversion_profitable_p.
16260 * config/i386/i386.c: Include ifcvt.h.
16261 (ix86_option_override_internal): Don't override
16262 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
16263 (ix86_noce_conversion_profitable_p): New function.
16264 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
16265 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
16266 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
16267 * doc/tm.texi: Regenerated.
16269 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16271 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
16274 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16277 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
16278 instructions for small multiply cores.
16280 2017-04-04 Jeff Law <law@redhat.com>
16282 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
16284 (mips_expand_vec_perm_const): Initialize elements in orig_perm
16285 that are not set by the loop over the elements.
16287 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16290 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
16291 int mode, convert_modes it to mode as unsigned, otherwise use
16292 lowpart_subreg to mode rather than SImode.
16293 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
16294 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
16295 Use DImode instead of SImode for the shift count operand.
16296 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
16299 2017-04-04 Richard Biener <rguenther@suse.de>
16301 PR middle-end/80281
16302 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
16303 arithmetic done for the negate or the plus. Simplify.
16304 (A - (-B) -> A + B): Likewise.
16305 * fold-const.c (split_tree): Make sure to not negate pointers.
16307 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
16309 PR rtl-optimization/60818
16310 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
16311 a compare of comparisons with the thing compared if this results
16312 in a different machine mode.
16314 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
16316 * alias.c (base_alias_check): Fix typo in comment.
16317 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
16318 * cgraphunit.c (symbol_table::compile): Likewise.
16319 * collect2.c (maybe_run_lto_and_relink): Likewise.
16320 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
16321 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
16322 * config/avr/avr.c (avr_map_op_t): Likewise.
16323 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
16324 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
16325 * config/epiphany/epiphany.md (movcc): Likewise.
16326 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
16327 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
16329 * config/mips/mips.c (mips_save_restore_reg): Likewise.
16330 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
16331 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
16332 * config/sh/sh.c (sh_rtx_costs): Likewise.
16333 * fold-const.c (fold_truth_andor): Likewise.
16334 * genautomata.c (collapse_flag): Likewise.
16335 * gengtype.h (struct type::u::s): Likewise.
16336 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
16337 * input.c (FORMAT_AMOUNT): Likewise.
16338 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
16339 (known_aggs_to_agg_replacement_list): Likewise.
16340 * ipa-inline-analysis.c: Likewise.
16341 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
16342 * ipa-polymorphic-call.c
16343 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
16344 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
16345 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
16347 * modulo-sched.c (apply_reg_moves): Likewise.
16348 * omp-expand.c (build_omp_regions_1): Likewise.
16349 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
16350 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
16351 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
16352 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
16353 * value-prof.c: Likewise.
16354 * var-tracking.c (val_reset): Likewise.
16356 2017-04-03 Richard Biener <rguenther@suse.de>
16358 PR tree-optimization/80275
16359 * fold-const.c (split_address_to_core_and_offset): Handle
16362 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
16364 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
16365 descriptors is at least equal to that of functions.
16367 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16369 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
16371 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16374 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
16375 (mov<IMOD4:mode>): New expander.
16376 (*mov<IMOD4:mode>_internal): New insn and split pattern.
16378 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
16380 PR rtl-optimization/79405
16381 * fwprop.c (propagations_left): New variable.
16382 (forward_propagate_into): Decrement it.
16383 (fwprop_init): Initialize it.
16384 (fw_prop): If the variable has reached zero, stop propagating.
16385 (fwprop_addr): Ditto.
16387 2017-03-31 Jakub Jelinek <jakub@redhat.com>
16390 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
16391 a FUNCTION_DECL, pass it as decl instead of origin to
16394 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
16396 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
16399 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
16402 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
16403 TARGET_VSX_SMALL_INTEGER.
16405 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16407 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16408 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
16410 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
16412 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
16413 extraction from odd-numbered MSA register.
16415 2017-03-31 Jakub Jelinek <jakub@redhat.com>
16417 PR middle-end/80173
16418 * expmed.c (store_bit_field_1): Don't attempt to create
16419 a word subreg out of hard registers wider than word if they
16420 have HARD_REGNO_NREGS of 1 for their mode.
16422 PR middle-end/80163
16423 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
16424 conversions to integer types wider than word and pointer.
16427 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
16428 (rtx_equal_for_cselib_p): Pass 0 to it.
16429 * cselib.c (cselib_hasher::equal): Likewise.
16430 (rtx_equal_for_cselib_1): Add depth argument. If depth
16431 is 128, don't look up VALUE locs and punt. Increment
16432 depth in recursive calls when walking VALUE locs.
16434 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
16436 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
16437 (make_gcov_file_name): Use the canonical path name for generating
16439 (read_line): Fix handling of files with ascii null bytes.
16441 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
16443 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
16444 to initialise a vector register instead
16445 of using a const_int.
16447 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16449 PR translation/80189
16450 * gimplify.c (omp_default_clause): Use %qs instead of %s in
16451 diagnostic messages.
16453 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
16456 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
16457 (dfp_diex_<mode>): Update mode of operand 1.
16458 * doc/extend.texi (dxex, dxexq): Document change to return type.
16459 (diex, diexq): Document change to argument type.
16461 2017-03-30 Martin Jambor <mjambor@suse.cz>
16464 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
16465 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
16466 it reflects the signature changes performed at the callee side.
16467 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
16468 to cgraph_build_function_type_skip_args.
16469 (build_function_decl_skip_args): Adjust call to the above function.
16471 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16474 * config/i386/sse.md
16475 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
16476 register as dest whenever it is a MEM not rtx_equal_p to the
16477 corresponding dup operand, and when forcing into reg move the
16478 reg into the memory afterwards.
16479 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
16480 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
16481 for the force_reg mode.
16482 (avx512vl_vextractf128<mode>): Use register as dest either
16483 always when a MEM, or when it is a MEM not rtx_equal_p to the
16484 corresponding dup operand, or even not when it is a CONST_VECTOR
16485 depending on the mode and lo vs. hi.
16486 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
16488 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
16489 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
16490 Likewise. Require that operands[2] is even.
16491 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
16492 Remove extraneous parens. Require that operands[2] is a multiple
16494 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
16495 operands[0] is a MEM if <mask_applied>, the predicates/constraints
16496 disallow memory then.
16498 2017-03-30 Richard Biener <rguenther@suse.de>
16500 PR tree-optimization/77498
16501 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
16502 to non-constants over backedges.
16504 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
16506 PR rtl-optimization/80233
16507 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
16508 as last_combined_insn. Do not test for BARRIER_P separately.
16510 2017-03-29 Andreas Schwab <schwab@suse.de>
16513 * calls.c (prepare_call_address): Convert funexp to Pmode before
16514 copying to temp reg.
16516 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16518 PR tree-optimization/80158
16519 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
16520 Handle possible future case of more than one alternate
16522 (replace_rhs_if_not_dup): Likewise.
16523 (replace_one_candidate): Likewise.
16525 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
16527 PR rtl-optimization/80193
16528 * ira.c (ira): Do not check allocation for LRA.
16530 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
16532 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
16533 (nvptx_output_simt_exit): Declare.
16534 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
16535 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
16536 (init_softstack_frame): Move initialization of crtl->is_leaf to...
16537 (nvptx_declare_function_name): ...here. Emit declaration of local
16538 memory space buffer for omp_simt_enter insn.
16539 (nvptx_output_unisimt_switch): New.
16540 (nvptx_output_softstack_switch): New.
16541 (nvptx_output_simt_enter): New.
16542 (nvptx_output_simt_exit): New.
16543 * config/nvptx/nvptx.h (struct machine_function): New fields
16544 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
16545 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
16546 (UNSPECV_SIMT_EXIT): Ditto.
16547 (omp_simt_enter_insn): New insn.
16548 (omp_simt_enter): New expansion.
16549 (omp_simt_exit): New insn.
16550 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
16552 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
16553 (expand_GOMP_SIMT_ENTER_ALLOC): New.
16554 (expand_GOMP_SIMT_EXIT): New.
16555 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
16556 (GOMP_SIMT_ENTER_ALLOC): Ditto.
16557 (GOMP_SIMT_EXIT): Ditto.
16558 * target-insns.def (omp_simt_enter): New insn.
16559 (omp_simt_exit): Ditto.
16560 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
16562 (lower_rec_simd_input_clauses): Implement SIMT privatization.
16563 (lower_rec_input_clauses): Likewise.
16564 (lower_lastprivate_clauses): Handle SIMT privatization.
16566 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
16567 (ompdevlow_adjust_simt_enter): New.
16568 (find_simtpriv_var_op): New.
16569 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
16570 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
16572 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
16573 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
16574 (copy_decl_for_dup_finish): Ditto.
16576 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
16578 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
16581 * config/i386/i386.c (ix86_option_override_internal): Always
16582 allow -mpreferred-stack-boundary=3 for 64-bit targets.
16584 2017-03-28 Bin Cheng <bin.cheng@arm.com>
16586 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
16588 2017-03-28 Bin Cheng <bin.cheng@arm.com>
16590 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
16591 mark new edge's irreducible flag accordign to it.
16592 (vect_do_peeling): Check loop preheader edge's irreducible flag
16593 and pass it to function slpeel_add_loop_guard.
16595 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
16597 PR tree-optimization/80218
16598 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
16599 Update block frequencies and counts.
16601 2017-03-28 Richard Biener <rguenther@suse.de>
16603 PR tree-optimization/78644
16604 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
16605 of a simplification result we may not use it at all.
16607 2017-03-28 Richard Biener <rguenther@suse.de>
16610 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
16611 without arguments, generate default definition of a SSA name.
16613 2017-03-28 Richard Biener <rguenther@suse.de>
16615 PR middle-end/80222
16616 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
16617 TYPE_REF_CAN_ALIAS_ALL references.
16618 * fold-const.c (fold_indirect_ref_1): Likewise.
16620 2017-03-28 Martin Liska <mliska@suse.cz>
16623 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
16624 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
16626 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
16627 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
16629 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
16630 (EXTRA_SPECS): Define.
16631 (SUBTARGET_EXTRA_SPECS): Likewise.
16632 (SUBTARGET_CPP_SPEC): Likewise.
16633 * config/arc/elf.h (EXTRA_SPECS): Renamed to
16634 SUBTARGET_EXTRA_SPECS.
16635 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
16637 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
16639 * config/arc/simdext.md (vst64_insn): Update pattern.
16640 (vld32wh_insn): Likewise.
16641 (vld32wl_insn): Likewise.
16642 (vld64_insn): Likewise.
16643 (vld32_insn): Likewise.
16645 2017-03-28 Marek Polacek <polacek@redhat.com>
16648 * fold-const.c (fold_comparison): Use protected_set_expr_location
16649 instead of SET_EXPR_LOCATION.
16651 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
16653 * tree.c (add_expr): Avoid name lookup warning.
16655 2017-03-27 Jeff Law <law@redhat.com>
16657 PR tree-optimization/80216
16658 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
16659 function name. Limit recursion depth.
16660 (record_temporary_equivalences): Corresponding changes.
16662 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
16664 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
16667 2017-03-27 Jakub Jelinek <jakub@redhat.com>
16670 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
16672 * cfgcleanup.c (reg_note_cfa_p): New array.
16673 (insns_have_identical_cfa_notes): New function.
16674 (old_insns_match_p): Don't cross-jump in between /f
16675 and non-/f instructions. If both i1 and i2 are frame related,
16676 verify all CFA notes, their order and content.
16678 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
16681 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
16682 HImode and SImode with zero extend to DImode to one insn.
16683 (bswap<mode>2_extenddi): Likewise.
16684 (bswapsi2_extenddi): Likewise.
16685 (bswaphi2_extendsi): Likewise.
16686 (bswaphi2): Combine bswap HImode and SImode into one insn.
16687 Separate memory insns from swapping register.
16688 (bswapsi2): Likewise.
16689 (bswap<mode>2): Likewise.
16690 (bswaphi2_internal): Delete, no longer used.
16691 (bswapsi2_internal): Likewise.
16692 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
16693 store, and gpr<-gpr swap insns.
16694 (bswap<mode>2_store): Likewise.
16695 (bswaphi2_reg): Register only splitter, combine with the splitter.
16696 (bswaphi2 splitter): Likewise.
16697 (bswapsi2_reg): Likewise.
16698 (bswapsi2 splitter): Likewise.
16699 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
16700 the insns into load, store, and register/register insns.
16701 (bswapdi2_ldbrx): Likewise.
16702 (bswapdi2_load): Likewise.
16703 (bswapdi2_store): Likewise.
16704 (bswapdi2_reg): Likewise.
16706 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
16708 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
16709 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
16711 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
16714 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
16716 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
16717 special handling for target option conflicts between dform
16718 options (-mpower9-dform, -mpower9-dform-vector,
16719 -mpower9-dform-scalar) and -mno-direct-move.
16721 2017-03-27 Richard Biener <rguenther@suse.de>
16723 PR tree-optimization/80181
16724 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
16726 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16728 * config/arc/predicates.md (move_double_src_operand): Replace the
16729 call to move_double_src_operand with a call to address_operand.
16731 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16733 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
16734 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
16735 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
16737 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16739 * config/arc/predicates.md (long_immediate_loadstore_operand):
16740 Consider scaled addresses cases.
16742 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16744 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
16745 restored when in interrupt.
16746 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
16747 doesn't have delay slot.
16749 2017-03-27 Richard Biener <rguenther@suse.de>
16752 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
16753 inlined thunk clones.
16755 2017-03-27 Jakub Jelinek <jakub@redhat.com>
16758 * asan.c (instrument_derefs): Copy over last operand from
16759 original COMPONENT_REF to the new COMPONENT_REF with
16760 DECL_BIT_FIELD_REPRESENTATIVE.
16761 * ubsan.c (instrument_object_size): Likewise.
16763 2017-03-27 Richard Biener <rguenther@suse.de>
16765 PR tree-optimization/80170
16766 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
16767 sure DR/SCEV didnt fold in constants we do not see when looking
16768 at the reference base alignment.
16770 2017-03-27 Richard Biener <rguenther@suse.de>
16772 PR middle-end/80171
16773 * gimple-fold.c (fold_ctor_reference): Properly guard against
16774 NULL return value from canonicalize_constructor_val.
16776 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
16779 * config/i386/i386.c (ix86_expand_builtin)
16780 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
16781 flags reg setting and flags reg using instructions.
16782 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
16783 clobbering instructions to zero extend op2.
16785 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
16787 * doc/install.texi (Configuration) <--with-aix-soname>:
16788 Update link to AIX ld.
16790 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
16792 PR rtl-optimization/80160
16793 PR rtl-optimization/80159
16794 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
16795 reg_alternate_class into account.
16797 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
16800 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
16801 to consider in curr_insn_transform.
16803 2017-03-24 Jakub Jelinek <jakub@redhat.com>
16805 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
16806 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
16807 and emit_mode_inner.
16809 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16811 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
16812 argument to the overloaded builtin variants. Use the new flag to
16813 deprecate certain builtin variants.
16814 * config/s390/s390-builtin-types.def: Add new builtin types.
16815 * config/s390/s390-builtins.h: Support new flags field for
16816 overloaded builtins.
16817 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
16818 (s390_macro_to_expand): Enable vector float data type.
16819 (s390_cpu_cpp_builtins_internal): Indicate support of the new
16820 builtins by incrementing the __VEC__ version number.
16821 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
16823 (s390_resolve_overloaded_builtin): Emit error messages depending
16824 on the builtin flags.
16825 * config/s390/s390.c (s390_expand_builtin): Support additional
16826 flags argument. Change error message to match the messages
16827 emitted in s390-c.c.
16828 * config/s390/s390.md: New UNSPEC_* constants.
16829 (op_type): Add new instruction types.
16830 * config/s390/vecintrin.h: Add new builtins and test data class
16832 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
16833 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
16834 (VEC_INEXACT, VEC_NOINEXACT): New constants.
16835 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
16836 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
16837 ("vec_mergel<mode>"): V_HW -> VEC_HW.
16839 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
16840 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
16841 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
16842 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
16844 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
16845 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
16846 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
16847 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
16849 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
16850 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
16851 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
16852 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
16853 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
16854 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
16855 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
16857 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
16858 ("vec_scatter_element<V_HW_4:mode>_DI")
16859 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
16860 ("vec_fpint<mode>", "vflls")
16861 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
16862 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
16863 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
16864 ("*vec_cmphe<mode>_cc"): ... these.
16866 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
16867 mode constant instead of magic value.
16869 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16871 * config/s390/s390.c (s390_expand_vec_compare): Support other
16872 vector floating point modes than just V2DF.
16873 (s390_expand_vcond): Likewise.
16874 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
16875 (s390_cannot_change_mode_class): Prevent mode changes between TF
16876 and V1TF in vector registers.
16877 * config/s390/s390.md (DF, SF): New mode attributes.
16878 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
16879 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
16880 SFmode support for VRs.
16881 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
16883 (VFT, VF_HW): New mode iterators.
16884 (vw, sdx): New mode attributes.
16885 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
16886 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
16887 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
16888 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
16889 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
16890 also the new vector floating point modes. Renaming to ...
16892 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
16893 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
16894 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
16895 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
16896 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
16897 ("vec_unordered<mode>"): ... these.
16899 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
16900 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
16901 ("*vec_extendv2df"): New insn definitions.
16903 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16905 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
16906 ("mulditi3_2", "*muldi3_sign"): New patterns.
16907 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
16908 rename the pattern definition.
16910 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16912 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
16914 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
16916 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16918 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
16919 instruction if possible.
16920 * config/s390/vector.md (vec_halfnumelts): New mode
16922 ("*vec_vllezlf<mode>"): New pattern.
16924 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16926 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
16927 ("popcountv4si2", "popcountv2di2"): Rename to ...
16928 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
16929 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
16931 ("popcount<mode>2_vxe"): New pattern.
16933 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16935 * common/config/s390/s390-common.c (processor_flags_table): Add
16937 * config.gcc: Add arch12.
16938 * config/s390/driver-native.c (s390_host_detect_local_cpu):
16939 Default to arch12 for unknown CPU model numbers.
16940 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
16941 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
16942 PROCESSOR_max sanity check.
16943 * config/s390/s390-opts.h (enum processor_type): Add
16945 * config/s390/s390.c (processor_table): Add arch12.
16946 (s390_expand_builtin): Add check for B_VXE flag.
16947 (s390_issue_rate): Add PROCESSOR_ARCH12.
16948 (s390_get_sched_attrmask): Likewise.
16949 (s390_get_unit_mask): Likewise.
16950 (s390_sched_score): Enable z13 scheduling for arch12.
16951 (s390_sched_reorder): Likewise.
16952 (s390_sched_variable_issue): Likewise.
16953 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
16955 (s390_tune_attr): Use z13 scheduling also for arch12.
16956 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
16957 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
16958 (TARGET_VXE_P): New macros.
16959 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
16960 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
16961 * config/s390/s390.opt: Add arch12 as processor_type.
16963 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16965 * config/s390/s390.md
16966 ("fixuns_truncdddi2", "fixuns_trunctddi2")
16967 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
16968 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
16970 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
16971 Rename expanders to ...
16973 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
16974 ("fixuns_truncdddi2_emu"): ... these.
16976 ("fixuns_trunc<mode>si2_emu"): New expander.
16978 ("*fixuns_truncdfdi2_z13"): Rename to ...
16979 ("*fixuns_truncdfdi2_vx"): ... this.
16981 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16983 * config/s390/2964.md: Remove the single element vector compare
16984 instructions which are no longer used.
16985 * config/s390/s390.c (s390_select_ccmode): Remove handling of
16987 (s390_canonicalize_comparison): Remove handling of DFmode
16989 (s390_expand_vec_compare_scalar): Remove function.
16990 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
16991 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
16993 ("*cmp<mode>_ccs"): Add wfcdb instruction.
16995 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16997 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
16999 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
17000 will anyway by matched by mov<mode>_64dfp.
17002 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17004 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
17005 vlef/vstef. Add missing operand to vleif.
17007 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17009 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
17010 pair for all vector types with 64 bit elements.
17011 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
17012 * config/s390/vector.md (V_HW_64): ... here.
17013 (V_128_NOSINGLE): New mode iterator.
17014 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
17015 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
17016 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
17017 ("*vec_load_pairv2di"): Change to ...
17018 ("*vec_load_pair<mode>"): ... this one.
17020 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17022 * config/s390/constraints.md: Add comments.
17023 (jKK): Reject element sizes > 8 bytes.
17024 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
17026 * config/s390/s390.md: Add the s_operand checks formerly in
17027 s390_split_ok_p to various splitters where they are still
17029 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
17030 for 128 bit vectors. Plus two splitters.
17032 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17034 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
17037 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17040 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
17041 error if the boundary argument is not constant.
17043 2017-03-24 Jakub Jelinek <jakub@redhat.com>
17045 PR rtl-optimization/80112
17046 * loop-doloop.c (doloop_condition_get): Don't check condition
17047 if cmp isn't SET with IF_THEN_ELSE src.
17049 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17051 PR tree-optimization/80158
17052 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
17053 replacing a candidate statement, also replace it for the
17054 candidate's alternate interpretation.
17055 (replace_rhs_if_not_dup): Likewise.
17056 (replace_one_candidate): Likewise.
17058 2017-03-24 Richard Biener <rguenther@suse.de>
17060 PR tree-optimization/80167
17061 * graphite-isl-ast-to-gimple.c
17062 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
17064 (translate_isl_ast_to_gimple::get_rename): Likewise.
17066 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
17068 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
17069 handling of certain combinations of target options, including the
17070 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
17071 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
17073 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17076 * config/arm/arm.md (*load_multiple): Add reload_completed to
17077 matching condition.
17079 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17080 Richard Biener <rguenth@suse.de>
17082 PR tree-optimization/79908
17083 PR tree-optimization/80136
17084 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
17085 been cast away, gimplify_and_add suffices.
17087 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
17089 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
17091 2017-03-23 Richard Biener <rguenther@suse.de>
17093 PR tree-optimization/80032
17094 * gimplify.c (gimple_push_cleanup): Forced unconditional
17095 cleanups still have to go to the conditional_cleanups
17098 2017-03-22 Jakub Jelinek <jakub@redhat.com>
17100 PR tree-optimization/80072
17101 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
17103 (next_operand_entry_id): Change type to unsigned int.
17104 (sort_by_operand_rank): Make sure to return the right return value
17105 even if unsigned fields are bigger than INT_MAX.
17106 (struct oecount): Change cnt and id type to unsigned int.
17107 (oecount_hasher::equal): Formatting fix.
17108 (oecount_cmp): Make sure to return the right return value
17109 even if unsigned fields are bigger than INT_MAX.
17110 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
17113 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
17114 TREE_READONLY on result if writing it more than once.
17117 * doc/invoke.texi (-fsanitize=thread): Document that with
17118 -fnon-call-exceptions atomics are not able to throw
17122 * tsan.c: Include tree-eh.h.
17123 (instrument_builtin_call): Call maybe_clean_eh_stmt or
17124 maybe_clean_or_replace_eh_stmt where needed.
17125 (instrument_memory_accesses): Add cfg_changed argument.
17126 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
17127 if it returned true.
17128 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
17130 PR rtl-optimization/63191
17131 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
17132 wrapper function, moved the whole old content into ...
17133 (ix86_delegitimize_address_1): ... this. New inline function.
17134 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
17135 true as last argument instead of ix86_delegitimize_address.
17137 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
17139 * config/aarch64/aarch64.c (generic_branch_cost): Copy
17140 cortexa57_branch_cost.
17142 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
17144 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
17146 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17149 * doc/md.texi (Constraints): Document wA constraint.
17150 * config/rs6000/constraints.md (wA): New.
17151 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
17152 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
17153 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
17154 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
17156 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
17159 * gimplify.c (is_oacc_declared): New function.
17160 (oacc_default_clause): Use it to set default flags for acc declared
17161 variables inside parallel regions.
17162 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
17163 declared variables.
17164 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
17165 declare attribute to any decl as necessary.
17167 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
17170 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
17171 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
17172 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
17173 (arm_arch_lpae): This.
17174 * config/arm/arm.c (arm_arch7ve): Rename into ...
17175 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
17176 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
17179 2017-03-22 Martin Liska <mliska@suse.cz>
17182 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
17183 error message instead of an ICE.
17185 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17187 * doc/extend.texi (6.11 Additional Floating Types): Revise.
17189 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
17191 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
17193 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
17196 2017-03-21 Martin Sebor <msebor@redhat.com>
17198 * doc/extend.texi: Use "cannot" instead of "can't."
17199 * doc/hostconfig.texi: Same.
17200 * doc/install.texi: Same.
17201 * doc/invoke.texi: Same.
17202 * doc/loop.texi: Same.
17203 * doc/md.texi: Same.
17204 * doc/objc.texi: Same.
17205 * doc/rtl.texi: Same.
17206 * doc/tm.texi: Same.
17207 * doc/tm.texi.in: Same.
17208 * doc/trouble.texi: Same.
17210 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
17213 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
17214 (collect_checksum_attributes): Set it.
17215 (die_checksum_ordered): Use it.
17217 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17219 PR tree-optimization/79908
17220 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
17221 change: For a VA_ARG whose LHS has been cast away, use
17222 force_gimple_operand to construct the side effects.
17224 2017-03-21 David Malcolm <dmalcolm@redhat.com>
17226 PR translation/80001
17227 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
17228 more amenable to translation.
17229 (oacc_loop_auto_partitions): Likewise.
17231 2017-03-21 Marek Polacek <polacek@redhat.com>
17232 Martin Sebor <msebor@redhat.com>
17234 PR tree-optimization/80109
17235 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
17236 on INTEGRAL_TYPE_P.
17238 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17239 Segher Boessenkool <segher@kernel.crashing.org>
17242 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
17243 check reg_used_between_p between insn and one of succ or succ2
17244 depending on if succ is artificial insn not inserted into insn
17247 2017-03-21 Martin Liska <mliska@suse.cz>
17249 PR gcov-profile/80081
17250 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
17251 * doc/gcc.texi: Include gcov-dump stuff.
17252 * doc/gcov-dump.texi: New file.
17254 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
17256 PR rtl-optimization/79150
17257 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
17258 conditional jump, if the jump is the last insn of the loop.
17260 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17261 Richard Biener <rguenth@suse.de>
17263 PR tree-optimization/79908
17264 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
17265 been cast away, use force_gimple_operand to construct the side
17268 2017-03-21 Martin Liska <mliska@suse.cz>
17270 PR libfortran/79956
17271 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
17274 2017-03-21 Brad Spengler <spender@grsecurity.net>
17277 * plugin.c (htab_hash_plugin): New function.
17278 (add_new_plugin): Use it and adjust.
17279 (parse_plugin_arg_opt): Adjust.
17280 (init_one_plugin): Likewise.
17282 2017-03-21 Richard Biener <rguenther@suse.de>
17284 PR tree-optimization/80032
17285 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
17286 if set force the cleanup to happen unconditionally.
17287 (gimplify_target_expr): Push inserted clobbers with force_uncond
17288 to avoid them being removed by control-dependent DCE.
17290 2017-03-21 Richard Biener <rguenther@suse.de>
17292 PR tree-optimization/80122
17293 * tree-inline.c (copy_bb): Do not expans va-arg packs or
17294 va_arg_pack_len when the inlined call stmt requires pack
17296 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
17298 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17301 * tsan.c (instrument_builtin_call): If the memory model argument
17302 is not a constant, assume it is valid.
17305 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
17308 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17310 PR rtl-optimization/79910
17311 * combine.c (can_combine_p): Do not allow combining an I0 or I1
17312 if its dest is used by an insn before I2 (other than the combined
17313 insns themselves, which are properly handled already).
17315 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17318 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17320 * combine.c (record_used_regs): New static function.
17321 (try_combine): Handle situations where there is an additional
17322 instruction between I2 and I3 which needs to have a LOG_LINK
17326 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17328 * combine.c (try_combine): Delete redundant i1 test. Call
17329 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17331 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17334 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
17335 alternatives 13/14.
17337 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17339 PR tree-optimization/80054
17340 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
17341 the optimization if a PHI or any of its arguments is not dominated
17342 by the candidate's basis. Use gphi* rather than gimple* as
17344 (replace_profitable_candidates): Clean up a gimple* variable that
17345 should be a gphi* variable.
17347 2017-03-20 Martin Sebor <msebor@redhat.com>
17350 * doc/extend.texi (attribute constructor): Document present limitation.
17352 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
17355 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
17356 __POWER9_VECTOR__ #ifdef control, change template definition to
17357 use Power9-specific built-in function.
17358 (vec_any_eq): Likewise.
17359 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
17360 to control outcomes from this test.
17361 (vector_ae_<mode>p): For VEC_F modes, likewise.
17363 2017-03-20 Ian Lance Taylor <iant@google.com>
17365 * config/i386/i386.c (ix86_function_regparm): Save an extra
17366 register for -fsplit-stack with DECL_STATIC_CHAIN.
17368 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17371 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
17372 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
17374 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17376 * config/riscv/riscv.c (riscv_print_operand): Use "fence
17378 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
17381 2017-03-20 Marek Polacek <polacek@redhat.com>
17384 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
17386 2017-03-20 Richard Biener <rguenther@suse.de>
17388 PR tree-optimization/80113
17389 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
17390 allocate extra SSA name for PHI def.
17391 (add_close_phis_to_outer_loops): Likewise.
17392 (add_close_phis_to_merge_points): Likewise.
17393 (copy_loop_close_phi_args): Likewise.
17394 (copy_cond_phi_nodes): Likewise.
17396 2017-03-20 Martin Liska <mliska@suse.cz>
17398 PR middle-end/79753
17399 * tree-chkp.c (chkp_build_returned_bound): Do not build
17400 returned bounds for a LHS that's not a BOUNDED_P type.
17402 2017-03-20 Martin Liska <mliska@suse.cz>
17406 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
17407 COMPLEX_CST and VECTOR_CST.
17409 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17412 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
17413 target operand. A new splitter adds the clobber statement in case
17414 the target operand is dead anyway.
17416 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
17418 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
17419 to age-old versions of binutils and glibc.
17421 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
17423 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
17425 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17427 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
17429 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17431 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
17432 requirement for binutils 2.13.
17434 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17436 * combine.c (try_combine): Delete redundant i1 test. Call
17437 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17439 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
17441 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
17442 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
17444 <riscv64-*-elf>: Re-arrange section
17445 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
17446 <riscv32-*-linux>: Likewise.
17447 <riscv64-*-elf>: Likewise
17448 <riscv64-*-linux>: Likewise.
17450 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
17453 * aarch64.opt(verbose-cost-dump): Fix typo.
17455 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
17458 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
17459 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
17461 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17463 * reload.c (find_reloads): When reloading a nonoffsettable address,
17464 use RELOAD_OTHER for it and its address reloads.
17466 PR rtl-optimization/79910
17467 * combine.c (record_used_regs): New static function.
17468 (try_combine): Handle situations where there is an additional
17469 instruction between I2 and I3 which needs to have a LOG_LINK
17472 2017-03-17 Jeff Law <law@redhat.com>
17474 PR tree-optimization/71437
17475 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
17476 conditional in the hash table first.
17477 (vrp_dom_walker::before_dom_children): Extract condition from
17478 ASSERT_EXPR. Record condition, its inverion and any implied
17479 conditions as well.
17481 2017-03-17 Marek Polacek <polacek@redhat.com>
17482 Markus Trippelsdorf <markus@trippelsdorf.de>
17484 PR tree-optimization/80079
17485 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
17488 2017-03-17 Richard Biener <rguenther@suse.de>
17490 PR middle-end/80075
17491 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
17492 Properly verify the LHS before the RHS possibly claims to be
17494 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
17497 2017-03-17 Martin Jambor <mjambor@suse.cz>
17499 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
17500 (List of -O2 options): Likewise.
17501 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
17504 2017-03-17 Tom de Vries <tom@codesourcery.com>
17506 * gcov-dump.c (print_usage): Print bug_report_url.
17508 2017-03-17 Richard Biener <rguenther@suse.de>
17510 PR middle-end/80050
17511 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
17512 (parser::peek): Likewise.
17514 2017-03-17 Richard Biener <rguenther@suse.de>
17516 PR tree-optimization/80048
17517 * sese.c (free_sese_info): Properly release rename_map and
17518 copied_bb_map elements.
17520 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
17522 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
17523 Add linked-list forward and backlinks. Insert on
17524 construction, remove on destruction.
17525 (class pass_store_merging): Add m_stores_head field.
17526 (pass_store_merging::terminate_and_process_all_chains):
17527 Iterate over m_stores_head list.
17528 (pass_store_merging::terminate_all_aliasing_chains):
17530 (pass_store_merging::execute): Check for debug stmts first.
17531 Push new chains onto the m_stores_head stack.
17533 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
17536 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
17537 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
17538 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
17540 2017-03-16 Jeff Law <law@redhat.com>
17542 PR tree-optimization/71437
17543 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
17544 member function. Implementation moved into after_dom_children
17545 member function and into the threader's thread_outgoing_edges
17547 (dom_opt_dom_walker::after_dom_children): Simplify by moving
17548 some code into new thread_outgoing_edges.
17549 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
17550 definition. Simplify marker handling (do it here). Assume we always
17551 have the available expression and the const/copies tables.
17552 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
17554 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
17555 * tree-vrp.c (equiv_stack): No longer file scoped.
17556 (vrp_dom_walker): New class.
17557 (vrp_dom_walker::before_dom_children): New member function.
17558 (vrp_dom_walker::after_dom_children): Likewise.
17559 (identify_jump_threads): Setup domwalker. Use it rather than
17560 walking edges in a random order by hand. Simplify setup/finalization.
17561 (finalize_jump_threads): Remove.
17562 (vrp_finalize): Do not call identify_jump_threads here.
17563 (execute_vrp): Do it here instead and call thread_through_all_blocks
17566 PR tree-optimization/71437
17567 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
17569 (simplify_stmt_for_jump_threading): Add basic_block argument. All
17571 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
17572 (dom_opt_dom_walker::thread_across_edge): Remove
17573 handle_dominating_asserts argument. All callers changed.
17574 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
17575 changes. Remove calls to lhs_of_dominating_assert. Other
17576 uses of handle_dominating_asserts turn into unconditional code
17577 (simplify_control_stmt_condition_1): Likewise.
17578 (simplify_control_stmt_condition): Likewise.
17579 (thread_through_normal_block, thread_across_edge): Likewise.
17580 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
17581 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
17582 object if it is not an SSA_NAME.
17583 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
17584 before calling into the VRP specific simplifiers.
17585 (identify_jump_threads): Remove handle_dominating_asserts
17588 2017-03-16 Jakub Jelinek <jakub@redhat.com>
17591 * tree-diagnostic.c (default_tree_printer): No longer static.
17592 * tree-diagnostic.h (default_tree_printer): New prototype.
17594 2017-03-16 Tamar Christina <tamar.christina@arm.com>
17596 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
17597 Change ins into fmov.
17599 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17601 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
17602 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
17603 Use h_con constraint for operand 1.
17604 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
17605 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
17607 2017-03-15 Jeff Law <law@redhat.com>
17609 PR tree-optimization/71437
17610 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
17611 (record_temporary_equivalences): Use it.
17613 PR tree-optimization/71437
17614 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
17615 tree-ssa-scopedtables.
17616 (lookup_avail_expr, build_and_record_new_cond): Likewise.
17617 (record_conditions, record_cond, vuse_eq): Likewise.
17618 (record_edge_info): Adjust to API tweak of record_conditions.
17619 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
17620 (record_temporary_equivalences, optimize_stmt): Likewise.
17621 (eliminate_redundant_computations): Likewise.
17622 (record_equivalences_from_stmt): Likewise.
17623 * tree-ssa-scopedtables.c: Include options.h and params.h.
17624 (vuse_eq): New function, moved from tree-ssa-dom.c
17625 (build_and_record_new_cond): Likewise.
17626 (record_conditions): Likewise. Accept vector of conditions rather
17627 than edge_equivalence structure for first argument.
17628 for the first argument.
17629 (avail_exprs_stack::lookup_avail_expr): New member function, moved
17630 from tree-ssa-dom.c.
17631 (avail_exprs_stack::record_cond): Likewise.
17632 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
17633 from tree-ssa-dom.c.
17634 (avail_exprs_stack): Add new member functions lookup_avail_expr
17636 (record_conditions): Declare.
17638 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
17641 * lra-constraints.c (process_alt_operands): Increase reject for
17642 reloading an input/output operand.
17644 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
17647 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
17648 insns to convert from signed/unsigned char/short to IEEE 128-bit
17650 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
17652 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
17655 * config/i386/i386.c (ix86_vector_duplicate_value): Create
17656 subreg of inner mode for values already in registers.
17658 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
17660 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
17661 iteration reg is used after the loop.
17663 2017-03-14 Martin Sebor <msebor@redhat.com>
17665 PR tree-optimization/79800
17666 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
17667 precision in negative-positive range.
17668 (format_floating): Call non-const overload with adjusted precision.
17670 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17673 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
17676 2017-03-14 Martin Sebor <msebor@redhat.com>
17678 PR middle-end/80020
17679 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
17680 * builtins.def (aligned_alloc): Use it.
17683 * Makefile.in (GTFILES): Add calls.c.
17684 * calls.c: Include "gt-calls.h".
17686 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
17688 PR rtl-optimization/79728
17689 * regs.h (struct target_regs): New field
17690 x_contains_allocatable_regs_of_mode.
17691 (contains_allocatable_regs_of_mode): New macro.
17692 * reginfo.c (init_reg_sets_1): Initialize it, and change
17693 contains_reg_of_mode so it includes global regs as well.
17694 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
17695 rather than contains_regs_of_mode.
17697 2017-03-14 Martin Liska <mliska@suse.cz>
17699 * doc/invoke.texi: Document options that can't be combined with
17700 -fcheck-pointer-bounds.
17702 2017-03-14 Martin Liska <mliska@suse.cz>
17704 PR middle-end/79831
17705 * doc/invoke.texi (-Wchkp): Document the option.
17707 2017-03-14 Martin Liska <mliska@suse.cz>
17709 * Makefile.in: Install gcov-dump.
17711 2017-03-14 Martin Liska <mliska@suse.cz>
17713 * multiple_target.c (expand_target_clones): Bail out for
17714 an invalid attribute.
17716 2017-03-14 Richard Biener <rguenther@suse.de>
17718 * alias.c (struct alias_set_entry): Pack properly.
17719 * cfgloop.h (struct loop): Likewise.
17720 * cse.c (struct set): Likewise.
17721 * ipa-utils.c (struct searchc_env): Likewise.
17722 * loop-invariant.c (struct invariant): Likewise.
17723 * lra-remat.c (struct cand): Likewise.
17724 * recog.c (struct change_t): Likewise.
17725 * rtl.h (struct address_info): Likewise.
17726 * symbol-summary.h (function_summary): Likewise.
17727 * tree-loop-distribution.c (struct partition): Likewise.
17728 * tree-object-size.c (struct object_size_info): Likewise.
17729 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
17730 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
17731 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
17732 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
17733 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
17734 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
17735 (struct _stmt_vec_info): Likewise.
17737 2017-03-14 Martin Liska <mliska@suse.cz>
17740 * multiple_target.c (create_dispatcher_calls): Check that
17741 a target can create a function dispatcher.
17743 2017-03-14 Martin Liska <mliska@suse.cz>
17746 * multiple_target.c (expand_target_clones): Drop local.local
17747 flag for default implementation.
17749 2017-03-14 Richard Biener <rguenther@suse.de>
17751 PR tree-optimization/80030
17752 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
17754 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
17756 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
17757 gcc_fallthrough() instead of __attribute__((fallthrough));
17759 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
17761 * doc/gcc.texi: Remove "up" link to (DIR).
17762 * doc/gccint.texi: Ditto.
17764 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
17766 * doc/install.texi (Specific) <avr>: Remove reference to
17769 2017-03-13 Jeff Law <law@redhat.com>
17771 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
17772 attribute rather than comments.
17774 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
17775 match_scratch operand is highest.
17777 2017-03-13 Martin Liska <mliska@suse.cz>
17779 PR middle-end/78339
17780 * ipa-pure-const.c (warn_function_noreturn): If the declarations
17781 is a CHKP clone, use original declaration.
17783 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17785 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
17786 (arc_conditional_register_usage): Use a different allocation order
17787 when optimizing for size.
17788 * common/config/arc/arc-common.c (arc_option_optimization_table):
17789 Section anchors default on when optimizing for size.
17791 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17793 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
17795 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17797 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
17798 * config/arc/arc.md (cpu_facility): Add cd variant.
17799 (*movqi_insn): Add code density variant.
17800 (*movhi_insn): Likewise.
17801 (*movqi_insn): Likewise.
17802 (*addsi3_mixed): Likewise.
17803 (subsi3_insn): Likewise.
17805 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17807 * config/arc/arc.md (movsi_cond_exec): Update constraint.
17809 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17811 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
17812 expressions with MINUS and UNARY ops.
17814 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17817 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
17819 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
17820 between vec_select and vector argument.
17821 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
17822 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
17823 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
17824 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
17825 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
17826 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
17828 2017-03-13 Richard Biener <rguenther@suse.de>
17831 * params.def (vect-max-peeling-for-alignment): Fix typo.
17833 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17835 * doc/install.texi (Specific) <mips-*-*>: Remove description of
17836 issue that only occurred with binutils below 2.18.
17838 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17840 * doc/install.texi (Specific) <cris-axis-elf>: No longer
17841 refer to binutils 2.11/2.12 minimum.
17843 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17845 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
17846 ftp.kernel.org and simplify binutils requirement.
17848 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
17850 * doc/invoke.texi (Warning Options): Fix spelling of link-time
17852 (Optimize Options): Ditto. Also remove redundancy.
17854 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17856 PR translation/79848
17857 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
17859 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
17860 to G_ to avoid double translation.
17862 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17864 PR translation/79923
17865 * auto-profile.c (get_combined_location): Convert leading
17866 character of diagnostics to lower case and remove trailing period.
17867 (read_profile): Likewise for various diagnostics.
17868 * config/arm/arm.c (arm_option_override): Remove trailing period
17869 from various diagnostics.
17870 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
17871 (msp430_expand_delay_cycles): Likewise.
17873 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17876 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
17877 full command-line argument, rather than just "str".
17878 (aarch64_validate_march): Likewise.
17879 (aarch64_validate_mtune): Likewise.
17881 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
17883 PR rtl-optimization/78911
17884 * lra-assigns.c (must_not_spill_p): New function.
17885 (spill_for): Use it.
17887 2017-03-10 Jakub Jelinek <jakub@redhat.com>
17889 PR tree-optimization/79981
17890 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
17891 ATOMIC_COMPARE_EXCHANGE ifn result.
17892 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
17893 IFN_ATOMIC_COMPARE_EXCHANGE.
17895 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17898 * opts.c (parse_sanitizer_options): Add missing question mark to
17899 "did you mean" message.
17901 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17903 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
17905 (VMULEUH_UNS): Likewise.
17906 (VMULOUB_UNS): Likewise.
17907 (VMULOUH_UNS): Likewise.
17908 * config/rs6000/rs6000.c (builtin_function_type): Remove
17909 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
17911 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17914 * read-rtl-function.c (function_reader::read_rtx_operand): Update
17915 x with result of extra_parsing_for_operand_code_0.
17916 (function_reader::extra_parsing_for_operand_code_0): Convert
17917 return type from void to rtx, returning x. When reading
17918 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
17919 larger size containing struct block_symbol.
17921 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
17923 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
17924 -mfloat128-hardware without -m64.
17926 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
17929 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
17930 entries to the case statement that marks unsigned arguments to
17931 overloaded functions.
17933 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
17935 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
17936 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
17938 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
17941 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
17942 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
17944 2017-03-10 Martin Liska <mliska@suse.cz>
17948 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
17949 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
17952 2017-03-10 Olivier Hainque <hainque@adacore.com>
17954 * tree-switch-conversion (array_value_type): Start by resetting
17955 candidate type to it's main variant.
17957 2017-03-10 Jakub Jelinek <jakub@redhat.com>
17959 PR rtl-optimization/79909
17960 * combine.c (try_combine): Use simplify_replace_rtx on individual
17961 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
17962 of the whole CALL_INSN_FUNCTION_USAGE.
17964 PR tree-optimization/79972
17965 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
17966 get_range_info on SSA_NAMEs. Formatting fixes.
17968 2017-03-10 Richard Biener <rguenther@suse.de>
17969 Jakub Jelinek <jakub@redhat.com>
17971 PR tree-optimization/77975
17972 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
17973 edge to be constant.
17974 (get_val_for): For constant x return it. Formatting fix.
17975 (loop_niter_by_eval): Avoid pointless looping if the next iteration
17976 would use the same bases as the current one.
17978 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17980 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
17981 instead of vec_select for V1TImode.
17982 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
17984 (VSX_LE_128): Add V1TI to this mode iterator.
17985 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
17986 (*vsx_le_perm_store_<mode>): Likewise.
17987 (pre-reload splitter for VSX stores): Likewise.
17988 (post-reload splitter for VSX stores): Likewise.
17989 (*vsx_xxpermdi2_le_<mode>): Likewise.
17990 (*vsx_lxvd2x2_le_<mode>): Likewise.
17991 (*vsx_stxvd2x2_le_<mode>): Likewise.
17993 2017-03-09 Michael Eager <eager@eagercon.com>
17995 Correct failures with --enable-checking=yes,rtl.
17997 * config/microblaze/microblaze.c (microblaze_expand_shift):
17998 Replace GET_CODE test with CONST_INT_P and INTVAL test with
17999 test for const0_rtx.
18000 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
18001 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
18003 2017-03-09 Richard Biener <rguenther@suse.de>
18005 PR tree-optimization/79977
18006 * graphite-scop-detection.c (scop_detection::merge_sese):
18007 Handle the case of extra exits to blocks dominating the entry.
18009 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
18011 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
18014 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
18016 PR rtl-optimization/79949
18017 * lra-constraints.c (process_alt_operands): Check memory when
18018 trying to predict a cycle. Print about the overall increase.
18020 2017-03-09 Richard Biener <rguenther@suse.de>
18022 PR middle-end/79971
18023 * gimple-expr.c (useless_type_conversion_p): Preserve
18024 TYPE_SATURATING for fixed-point types.
18026 2017-03-09 Richard Biener <rguenther@suse.de>
18029 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
18030 alignment of BLKmode params.
18032 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18035 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
18036 (VALL_NO_V2Q): Likewise.
18038 * config/aarch64/aarch64-simd.md
18039 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
18041 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
18042 VALL_NO_V2Q mode iterator.
18043 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
18045 2017-03-09 Martin Liska <mliska@suse.cz>
18047 PR tree-optimization/79631
18048 * tree-chkp-opt.c (chkp_is_constant_addr): Call
18049 tree_int_cst_sign_bit just for INTEGER constants.
18051 2017-03-09 Martin Liska <mliska@suse.cz>
18055 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
18058 2017-03-09 Marek Polacek <polacek@redhat.com>
18061 * tree.c (inchash::add_expr): Handle TREE_VEC.
18063 2017-03-09 Martin Liska <mliska@suse.cz>
18066 (chkp_narrow_size_and_offset): New function.
18067 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
18068 (void chkp_parse_bit_field_ref): New function.
18069 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
18070 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
18072 2017-03-09 Martin Liska <mliska@suse.cz>
18075 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
18076 (chkp_find_bounds_1): Remove gcc_unreachable.
18078 2017-03-09 Jakub Jelinek <jakub@redhat.com>
18081 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
18082 BUILT_IN_SYNC*, determine the access type from the size suffix and
18083 always build a MEM_REF with that type. Handle forgotten
18084 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
18087 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
18088 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
18089 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
18090 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
18091 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
18092 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
18093 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
18094 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
18095 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
18096 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
18097 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
18098 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
18099 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
18100 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
18101 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
18102 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
18103 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
18104 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
18105 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
18106 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
18107 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
18108 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
18109 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
18110 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
18111 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
18112 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
18113 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
18114 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
18115 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
18116 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
18117 definitions outside of __OPTIMIZE__ guarded section.
18120 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
18121 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
18122 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
18123 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
18126 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18128 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
18129 ("vfenez<mode>"): Add missing constraints.
18131 2017-03-08 Martin Sebor <msebor@redhat.com>
18134 * config/nds32/nds32.c (nds32_option_override):
18135 Fix misspelled diagnostic.
18137 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18140 * gimplify.c (gimplify_omp_for): Replace index var in outer
18141 taskloop statement with an artificial variable and add
18142 OMP_CLAUSE_PRIVATE clause for it.
18144 2017-03-08 Richard Biener <rguenther@suse.de>
18146 PR tree-optimization/79955
18147 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
18148 for accesses that are completely outside of the variable.
18150 2017-03-08 Andrew Haley <aph@redhat.com>
18152 PR tree-optimization/79943
18153 * tree-ssa-loop-split.c (compute_new_first_bound): When
18154 calculating the new upper bound, (END-BEG) should be added, not
18157 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18159 * config/avr/avr.md (setmemhi): Make sure match_dup
18160 operand number comes before match_scratch.
18162 2017-03-08 Richard Biener <rguenther@suse.de>
18164 PR tree-optimization/79920
18165 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
18166 with ncopies == 1 to ...
18167 (vect_transform_slp_perm_load): ... here. Properly compute
18168 all element loads by iterating VF times over the group. Do
18169 not handle ncopies (computed in a broken way) in
18170 vect_create_mask_and_perm.
18172 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18175 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
18176 is a uniform vector, use uniform_vector_p return value instead of
18177 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
18179 2017-03-07 Marek Polacek <polacek@redhat.com>
18181 PR middle-end/79809
18182 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
18183 (alloca_call_type): Likewise.
18185 2017-03-07 Martin Liska <mliska@suse.cz>
18187 * gcov.c (process_args): Put comment to correct location.
18189 2017-03-07 Martin Liska <mliska@suse.cz>
18191 PR middle-end/68270
18192 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
18193 Use array_at_struct_end_p instead of DECL_CHAIN (field).
18194 (chkp_narrow_bounds_for_field): Likewise.
18195 (chkp_parse_array_and_component_ref): Pass one more argument to
18198 2017-03-07 Richard Biener <rguenther@suse.de>
18200 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
18203 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
18205 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
18206 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
18208 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18211 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
18212 to end of description.
18213 (PARAM_MAX_STORES_TO_MERGE): Likewise.
18215 2017-03-07 Jakub Jelinek <jakub@redhat.com>
18217 PR rtl-optimization/79901
18218 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
18220 (*avx512f_<code><mode>3<mask_name>): ... this.
18221 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
18222 iterator instead of VI8_AVX2_AVX512BW.
18224 PR rtl-optimization/79901
18225 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
18226 min/max expander, expand it using expand_vec_cond_expr.
18229 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
18232 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18235 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
18236 to void * for PCH reasons.
18237 * dwarf2out.c (output_loc_operands, output_die): Cast
18238 v.val_vec.array to unsigned char *.
18240 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
18243 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
18246 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
18248 PR rtl-optimization/79571
18249 * lra-constraints.c (process_alt_operands): Calculate static
18250 reject and subtract it from overall when only addresses will be
18253 2017-03-06 Julia Koval <julia.koval@intel.com>
18256 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
18257 incoming stack boundary to 128 for 64-bit targets.
18259 2017-03-06 Richard Biener <rguenther@suse.de>
18261 PR tree-optimization/79894
18262 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
18263 to NULL after folding it.
18265 2017-03-06 Richard Biener <rguenther@suse.de>
18267 PR tree-optimization/79824
18268 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
18269 check disabling peeling for gaps.
18271 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
18273 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
18274 attributes): Document gettimeofday.
18276 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18278 * config/s390/s390.c (s390_option_override_internal): Set
18279 PARAM_MIN_VECT_LOOP_BOUND
18281 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18283 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
18284 * config/s390/s390.md: Likewise.
18286 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18289 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
18290 (<avx2_avx512>_perm<mode>): Rename to ...
18291 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
18293 (<avx512>_perm<mode>_mask): Rename to ...
18294 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
18296 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
18297 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
18298 instead of VI8F_256_512.
18299 (avx512f_perm<mode>): New define_expand.
18300 (avx512f_perm<mode>_mask): Likewise.
18301 (avx512f_perm<mode>_1<mask_name>): New define_insn.
18302 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
18304 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18306 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
18307 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
18309 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
18311 2017-03-06 Martin Liska <mliska@suse.cz>
18314 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
18315 when having a SSA NAME w/o VAR_DECL assigned to it.
18317 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18319 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
18320 msa_dpsub_<su>_d): Fix MODE for vec_select.
18322 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18324 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
18326 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
18328 2017-03-06 Richard Biener <rguenther@suse.de>
18330 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
18331 * plugin.c (register_plugin_info): Likewise.
18332 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
18334 2017-03-05 Jakub Jelinek <jakub@redhat.com>
18336 * config/i386/sse.md (sse_storehps, sse_storelps,
18337 avx_<castmode><avxsizesuffix>_<castmode>,
18338 avx512f_<castmode><avxsizesuffix>_<castmode>,
18339 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
18340 in condition that at least one operand is not a MEM.
18342 2017-03-03 Jakub Jelinek <jakub@redhat.com>
18344 PR middle-end/79805
18345 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
18346 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
18348 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
18349 gimple_call_nothrow_p flag based on whether original builtin can throw.
18350 If it can, emit following stmts on the fallthrough edge.
18351 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
18352 don't create new bb if inserting just debug stmts on the edge, try to
18353 insert them on the fallthru bb or just reset debug stmts.
18355 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
18358 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
18359 restore recog_data (including the operand rtxes inside it) around
18360 the call to get_insn_template.
18362 2017-03-03 Martin Sebor <msebor@redhat.com>
18364 PR tree-optimization/79699
18365 * context.c (context::~context): Free MPFR caches to avoid
18366 a memory leak on program exit.
18368 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18370 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
18371 Use wide_int::ulow () instead of .elt (0).
18373 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18375 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
18376 (*pushxf): Limit oF constraint to 32bit targets and add oC
18377 constraint for 64bit targets.
18378 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
18379 (*pushdf): Change rmF constraint to rmC.
18381 2017-03-03 Martin Liska <mliska@suse.cz>
18383 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
18384 Remove unused variable.
18386 2017-03-03 Jakub Jelinek <jakub@redhat.com>
18389 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
18390 is a memory operand, increase num_memory.
18391 (ix86_expand_args_builtin): Likewise.
18393 2017-03-03 Jan Hubicka <jh@suse.cz>
18396 * ipa-devirt.c (maybe_record_node): Properly handle
18397 __cxa_pure_virtual visibility.
18399 2017-03-03 Martin Liska <mliska@suse.cz>
18401 PR tree-optimization/79803
18402 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
18404 (pass_loop_prefetch::execute): Disabled optimization if an
18405 assumption about L1 cache size is not met.
18407 2017-03-03 Martin Liska <mliska@suse.cz>
18409 PR rtl-optimization/79574
18410 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
18411 (hash_scan_set): Likewise.
18412 (dump_hash_table): Likewise.
18413 (hoist_code): Likewise.
18415 2017-03-03 Richard Biener <rguenther@suse.de>
18417 * fixed-value.c (fixed_from_string): Restore use of elt (1)
18418 in place of uhigh ().
18419 (fixed_convert_from_real): Likewise.
18421 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18424 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
18426 2017-03-03 Richard Biener <rguenther@suse.de>
18428 PR middle-end/79818
18429 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
18430 TYPE_OVERFLOW_UNDEFINED check.
18432 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18434 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
18436 (vector_ae_<mode>_p): Likewise.
18437 (vector_nez_<mode>_p): Likewise.
18438 (vector_ne_v2di_p): Likewise.
18439 (vector_ae_v2di_p): Likewise.
18440 (vector_ne_<mode>_p): Likewise.
18441 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
18443 (vsx_tsqrt<mode>2_fe): Likewise.
18445 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
18448 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
18450 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18452 PR rtl-optimization/79780
18453 * cprop.c (one_cprop_pass): When second and further conditional trap
18454 in a single basic block is turned into an unconditional trap, turn it
18455 into a deleted note to avoid RTL verification failures.
18457 2017-03-02 Richard Biener <rguenther@suse.de>
18459 * fold-const.c (const_binop): Use ulow () instead of elt (0).
18461 2017-03-02 Richard Biener <rguenther@suse.de>
18463 PR tree-optimization/79345
18465 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
18466 param and abort the walk, returning -1 if it is hit.
18467 (walk_aliased_vdefs): Take a limit param and pass it on.
18468 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
18469 defaulting to 0 and return a signed int.
18470 * tree-ssa-uninit.c (struct check_defs_data): New struct.
18471 (check_defs): New helper.
18472 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
18473 about uninitialized memory.
18474 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
18475 bogus uninitialized warning.
18476 (fixed_convert_from_real): Likewise.
18478 2017-03-02 Bin Cheng <bin.cheng@arm.com>
18480 PR tree-optimization/66768
18481 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
18482 iv_use if base object can't be determined.
18484 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18486 PR tree-optimization/79345
18487 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
18488 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
18489 (get_pattern_stats): Initialize it.
18490 * genemit.c (gen_expand): Verify match_scratch numbers come after
18491 match_operand/match_dup numbers.
18492 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
18493 match_scratch numbers.
18494 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
18496 * config/s390/s390.md (trunctdsd2): Likewise.
18498 2017-03-02 Richard Biener <rguenther@suse.de>
18500 * wide-int.h (wide_int_storage::operator=): Implement in terms
18503 2017-03-02 Richard Biener <rguenther@suse.de>
18505 PR tree-optimization/79777
18506 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
18507 the to insert expression to sth existing.
18509 2017-03-01 Martin Sebor <msebor@redhat.com>
18511 PR middle-end/79692
18512 * gimple-ssa-sprintf.c
18513 (directive::known_width_and_precision): New function.
18514 (format_integer): Use it.
18515 (get_mpfr_format_length): Consider the full range of precision
18516 when computing %g output with the # flag. Set the likely byte
18517 count to 3 rather than 1 when precision is indeterminate.
18518 (format_floating): Correct the lower bound of precision.
18520 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18522 * doc/invoke.texi: Document default code model for 64-bit Linux.
18524 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18527 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
18528 udiv rather than div since input pattern is unsigned.
18530 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
18532 * config/i386/i386.c (print_reg): Warn for values of
18533 unsupported size in integer register.
18535 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
18538 * config/rs6000/predicates.md (current_file_function_operand): Do
18539 not allow self calls to be local if the function is replaceable.
18541 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
18544 * config/rs6000/altivec.h (vec_ctz and others): Change the
18545 preprocessor macro that controls conditional compilation from
18546 _ARCH_PWR9 to __POWER9_VECTOR__.
18547 (vec_all_ne): Change parameterization of __altivec_scalar_pred
18548 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
18549 control (instead of _ARCH_PWR9 control) so that template
18550 definition uses power9-specific function.
18551 (vec_any_eq): Likewise.
18552 (vec_all_ne): Change macro definition to use a power9-specific
18553 expansion under #ifdef __POWER9_VECTOR__ control (instead of
18554 _ARCH_PWR9 control).
18555 (vec_any_eq) Likewise.
18556 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
18557 expansion for CMPNEF to remove support for xvcmpnesp instruction.
18558 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
18559 support for xvcmpnedp instruction.
18560 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
18561 macro expansion so that Power9 implementation of vec_all_ne does
18562 not use the AltiVec predicate framework.
18563 (VCMPNEH_P): Likewise.
18564 (VCMPNEW_P): Likewise.
18565 (VCMPNED_P): Likewise.
18566 (VCMPNEFP_P): Likewise.
18567 (VCMPNEDP_P): Likewise.
18568 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
18569 implementation of vec_any_eq to not use AltiVec predicate
18571 (VCMPAEH_P): Likewise.
18572 (VCMPAEW_P): Likewise.
18573 (VCMPAED_P): Likewise.
18574 (VCMPAEFP_P): Likewise.
18575 (VCMPAEDP_P): Likewise.
18576 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
18577 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
18578 not use the AltiVec predicate framework.
18579 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
18580 of vec_any_eq to not use AltiVec predicate framework.
18581 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
18582 support for predefined __POWER9_VECTOR__ macro to indicate that
18583 Power9 instruction selection is enabled.
18584 (altivec_overloaded_builtins): Remove extraneous
18585 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
18586 function argument types RS6000_BTI_bool_V16QI and
18587 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
18588 entry for overloaded function argument types RS6000_BTI_bool_V4SI
18589 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
18590 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
18591 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
18592 Power9 for implementations of vec_cmpne. Change the signature for
18593 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
18594 (representing vec_all_ne) to remove the previously described first
18595 argument of type RS6000_BTI_INTSI, as this was an artifact of
18596 reliance on the AltiVec predicate framework, which is no longer
18597 used in the implementation of these functions. Add
18598 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
18599 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
18600 since, unlike the AltiVec predicate framework implementation, we
18601 do not share function descriptors between vec_alle and vec_anyeq.
18602 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
18603 set of modes that receive special treatment even when
18604 TARGET_P9_VECTOR is true. The special treatment emits code that
18605 does not depend on Power9 instructions.
18606 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
18607 define_expand to not rely on AltiVec predicate framework.
18608 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
18610 (vector_ne_v2di_p): Change this define_expand to not rely on
18611 AltiVec predicate framework.
18612 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
18614 (vector_ne_<mode>_p): Change this define_expand to not rely on
18615 AltiVec predicate framework.
18616 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
18618 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
18619 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
18620 define_insn pattern.
18621 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
18622 define_insn pattern because the xvcmpne<VSs>. instruction is not
18624 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
18625 instruction is not supported.
18627 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18629 * config/nvptx/nvptx.c: Include intl.h.
18631 2017-03-01 Martin Jambor <mjambor@suse.cz>
18634 * ipa-prop.h (ipa_bits): Removed field known.
18635 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
18636 to pointers. Adjusted their comments to warn about their sharing.
18637 (ipcp_transformation_summary): Change bits to a vector of pointers.
18638 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
18639 (ipa_get_ipa_bits_for_value): Declare.
18640 * tree-vrp.h (value_range): Mark as GTY((for_user)).
18641 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
18642 (ipa_bits_hash_table): Likewise.
18643 (ipa_vr_ggc_hash_traits): Likewise.
18644 (ipa_vr_hash_table): Likewise.
18645 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
18646 being pointers and vr_known being removed.
18647 (ipa_set_jf_unknown): Likewise.
18648 (ipa_get_ipa_bits_for_value): New function.
18649 (ipa_set_jfunc_bits): Likewise.
18650 (ipa_get_value_range): New overloaded functions.
18651 (ipa_set_jfunc_vr): Likewise.
18652 (ipa_compute_jump_functions_for_edge): Use the above functions to
18653 construct bits and vr parts of jump functions.
18654 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
18655 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
18657 (ipcp_grow_transformations_if_necessary): Also allocate
18658 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
18660 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
18661 them. Fix too long lines.
18662 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
18663 vr_known being removed.
18664 (ipa_read_jump_function): Use new setter functions to construct bits
18665 and vr parts of jump functions or set them to NULL.
18666 (write_ipcp_transformation_info): Adjust for bits being pointers.
18667 (read_ipcp_transformation_info): Likewise.
18668 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
18670 Include gt-ipa-prop.h.
18671 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
18673 (ipcp_store_bits_results): Likewise.
18674 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
18675 Do not write to existing jump functions but use a temporary instead.
18677 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18680 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
18681 attempt to use its first operand as BIT_FIELD_REF base.
18683 2017-03-01 Richard Biener <rguenther@suse.de>
18685 PR middle-end/79721
18686 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
18687 interpolating formula in wrapping arithmetic.
18688 (chrec_apply): Convert chrec_evaluate return value to wanted type.
18690 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18692 PR tree-optimization/79734
18693 * tree-vect-generic.c (expand_vector_condition): Optimize
18694 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
18695 Handle VEC_COND_EXPR where comparison has different inner width from
18696 type's inner width.
18698 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
18700 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
18701 markup, and similar issues. Remove @opindex entries for things
18702 that aren't options. Add missing -mmpy-option entries.
18704 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18706 PR tree-optimization/79737
18707 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
18708 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
18709 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
18710 instead of byte_size. Formatting fix.
18711 (shift_bytes_in_array_right): Formatting fix.
18713 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
18716 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
18717 condition on optimize for the leaf function test.
18719 2017-02-28 Martin Liska <mliska@suse.cz>
18722 * read-rtl-function.c (function_reader::handle_unknown_directive):
18723 Bail out when one uses -flto.
18725 2017-02-28 Martin Liska <mliska@suse.cz>
18727 * common.opt: Replace space with tabular for options of <number>
18729 * config/i386/i386.opt: Show <number> value for
18730 -mlarge-data-threshold.
18731 * opts.c (print_filtered_help): Do not display number in hexadecimal
18734 2017-02-28 Martin Liska <mliska@suse.cz>
18736 * common.opt: Fix --help=option -Q for options which are of
18739 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
18741 * config/i386/i386.c (print_reg): Error out for values
18742 of 8-bit size in invalid integer register.
18744 2017-02-28 Martin Sebor <msebor@redhat.com>
18746 PR tree-optimization/79691
18747 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
18749 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18752 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
18753 gcc_unreachable with output_operand_lossage.
18755 2017-02-28 Richard Biener <rguenther@suse.de>
18757 PR tree-optimization/79740
18758 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
18760 (visit_nary_op): Insert the nary into the hashtable if we
18761 pattern-matched sth.
18762 * tree-ssa-pre.c (eliminate_insert): Robustify.
18764 2017-02-28 Richard Biener <rguenther@suse.de>
18766 PR middle-end/79731
18767 * fold-const.c (decode_field_reference): Reject out-of-bound
18770 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18772 * config/i386/i386.c: Include intl.h.
18773 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
18774 instead of just cond ? "..." : "...".
18775 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
18776 * coverage.c (read_counts_file): Likewise.
18777 * omp-offload.c: Include intl.h.
18778 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
18779 of just cond ? "..." : "...".
18780 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
18781 of just cond ? "..." : "...".
18783 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
18786 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
18788 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
18789 'tune for' CPU name.
18790 * config/arm/arm-cpu-data.h: Regenerated.
18792 2017-02-28 Richard Biener <rguenther@suse.de>
18794 PR tree-optimization/79732
18795 * tree-inline.c (expand_call_inline): Do not shadow var.
18797 2017-02-28 Richard Biener <rguenther@suse.de>
18799 PR tree-optimization/79723
18800 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
18801 address-space properly.
18803 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
18805 * doc/optinfo.texi (Optimization groups): Fix option used for
18807 * doc/invoke.texi (-fopt-info): Document "omp".
18808 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
18809 (OPTGROUP_ALL): Add OPTGROUP_OMP.
18810 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
18811 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
18812 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
18814 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
18816 * dumpfile.c (optgroup_options): Instead of "openmp", associate
18817 OPTGROUP_OMP with "omp".
18819 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
18822 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
18823 for arithmetic shift of unsigned V2DI.
18825 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
18827 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
18828 arc/linux.h headers.
18829 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
18830 (LINK_SPEC): Likewise.
18831 (ARC_TLS_EXTRA_START_SPEC): Likewise.
18832 (EXTRA_SPECS): Likewise.
18833 (STARTFILE_SPEC): Likewise.
18834 (ENDFILE_SPEC): Likewise.
18835 (LIB_SPEC): Likewise.
18836 (TARGET_SDATA_DEFAULT): Likewise.
18837 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
18838 (MULTILIB_DEFAULTS): Likewise.
18839 (DWARF2_UNWIND_INFO): Likewise.
18840 * config/arc/big.h: New file.
18841 * config/arc/elf.h: Likewise.
18842 * config/arc/linux.h: Likewise.
18843 * config/arc/t-uClibc: Remove.
18845 2017-02-27 Bin Cheng <bin.cheng@arm.com>
18847 PR tree-optimization/77536
18848 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
18849 (tree_transform_and_unroll_loop): Use above function to compute the
18850 estimated niter of unrolled loop and use it when scaling profile.
18851 Also use count info rather than frequency if it's non-zero.
18852 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
18853 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
18854 (vect_transform_loop): Call above function.
18856 2017-02-27 Richard Biener <rguenther@suse.de>
18858 PR tree-optimization/45397
18859 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
18860 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
18861 (visit_nary_op): Add pattern matching for CSEing sign-changed
18862 or truncated operations with wider ones.
18864 2017-02-27 Richard Biener <rguenther@suse.de>
18866 PR tree-optimization/79690
18867 * tree-vect-stmts.c (vectorizable_store): Use vector type
18868 built from the DR with address-space.
18870 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
18872 * doc/invoke.texi (Optimize Options): Refine the description
18873 of asan-use-after-return.
18875 2017-02-25 Alan Modra <amodra@gmail.com>
18877 PR rtl-optimization/79584
18878 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
18879 base, not ad->base_term, the reg within base. Remove assertion
18880 that ad->base == ad->base_term. Replace gen_int_mode using
18881 bogus mode with const0_rtx.
18883 2017-02-25 Jakub Jelinek <jakub@redhat.com>
18885 PR middle-end/79396
18886 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
18887 FMA_EXPR like tcc_binary or tcc_unary.
18889 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
18892 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
18894 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
18895 (output_loc_operands): Handle DW_OP_call_ref and
18896 DW_OP_GNU_variable_value.
18897 (struct variable_value_struct): New type.
18898 (struct variable_value_hasher): Likewise.
18899 (variable_value_hash): New variable.
18900 (string_types): Remove.
18901 (copy_loc_descr): New function.
18902 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
18903 (prepend_loc_descr_to_each): New function.
18904 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
18905 instead of add_loc_descr_to_each if the first argument is single
18906 location list and the second has multiple.
18907 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
18908 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
18909 when looking for variable value which doesn't have other location info.
18910 (loc_list_from_tree): Formatting fix.
18911 (gen_array_type_die): Simplify DW_AT_string_length handling.
18912 (adjust_string_types): Remove.
18913 (gen_subprogram_die): Don't call adjust_string_types nor test/set
18914 string_types. Call resolve_variable_values.
18915 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
18916 (resolve_addr_in_expr): Likewise. Add A argument.
18917 (copy_deref_exprloc): Remove deref argument. Adjust for the
18918 original expression being DW_OP_GNU_variable_value with optionally
18919 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
18920 optionally after it.
18921 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
18922 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
18923 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
18924 (variable_value_hasher::hash, variable_value_hasher::equal): New
18926 (resolve_variable_value_in_expr, resolve_variable_value,
18927 resolve_variable_values, note_variable_value_in_expr,
18928 note_variable_value): New functions.
18929 (dwarf2out_early_finish): Call note_variable_value on all toplevel
18932 2017-02-24 Jakub Jelinek <jakub@redhat.com>
18935 * opts.h (handle_generated_option): Add GENERATED_P argument.
18936 * opts-common.c (handle_option): Adjust function comment.
18937 (handle_generated_option): Add GENERATED_P argument, pass it to
18939 (control_warning_option): Pass false to handle_generated_option
18941 * opts.c (maybe_default_option): Pass true to handle_generated_option
18943 * optc-gen.awk: Likewise.
18945 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18947 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
18948 a REG, look at the REG it is a SUBREG of.
18949 (splitter for cmpeqsi_t): Ditto.
18951 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18953 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
18954 the special USEs with the pattern of the insn, not the insn itself.
18956 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
18959 * doc/invoke.texi: Document -mload-store-pairs.
18961 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18962 Sandra Loosemore <sandra@codesourcery.com>
18964 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
18965 argument isn't a CONST_INT.
18966 (nios2_alternate_compare_const): Assert op is a CONST_INT.
18967 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
18968 (nios2_validate_compare): Bypass alternate compare logic if *op2
18969 is not a CONST_INT.
18970 (ldstwm_operation_p): Return false if first_base is not a REG or
18971 if first_offset is not a CONST_INT.
18973 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18975 * config/cris/cris.md: Use correct operand in a define_peephole2.
18977 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18979 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
18981 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18983 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
18984 this_insn if it is an INSN or JUMP_INSN.
18985 (force_offsettable): Look at base, not at addr.
18986 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
18987 on things that aren't necessarily CONST_INTs.
18989 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
18991 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
18992 -mfpmath=sse is the default also for x86-32 targets with SSE2
18993 instruction set when @option{-ffast-math} is enabled
18995 2017-02-24 Jeff Law <law@redhat.com>
18997 PR rtl-optimizatoin/79286
18998 * ira.c (update_equiv_regs): Drop may_trap_p exception to
19001 2017-02-24 Richard Biener <rguenther@suse.de>
19003 PR tree-optimization/79389
19004 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
19007 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
19009 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
19010 function comment to reflect reality.
19011 (loop_exits_before_overflow): Fix typo in function description.
19013 2017-02-24 Richard Biener <rguenther@suse.de>
19015 PR tree-optimization/79389
19016 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
19017 properly that a threading opportunity exists. Detect conditional
19018 copy/constant propagation opportunities.
19020 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
19022 * config/visium/visium.md (type): Add trap.
19023 (b): New mode attribute.
19024 (*btst): Rename into...
19025 (*btst<mode>): ...this and adjust.
19026 (*cbranchsi4_btst_insn): Rename into...
19027 (*cbranch<mode>4_btst_insn): ...this and adjust.
19028 (trap): New define_insn.
19030 2017-02-23 Jakub Jelinek <jakub@redhat.com>
19032 PR tree-optimization/79389
19033 * ifcvt.c (struct noce_if_info): Add rev_cond field.
19034 (noce_reversed_cond_code): New function.
19035 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
19036 reversed_comparison_code. Formatting fix.
19037 (noce_try_store_flag): Test rev_cond != NULL in addition to
19038 reversed_comparison_code.
19039 (noce_try_store_flag_constants): Likewise.
19040 (noce_try_store_flag_mask): Likewise.
19041 (noce_try_addcc): Use rev_cond if non-NULL instead of
19042 reversed_comparison_code.
19043 (noce_try_cmove_arith): Likewise. Formatting fixes.
19044 (noce_try_minmax, noce_try_abs): Clear rev_cond.
19045 (noce_find_if_block): Initialize rev_cond.
19046 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
19047 instead of false as last argument never attempt to reverse it
19050 2017-02-23 Bin Cheng <bin.cheng@arm.com>
19052 PR tree-optimization/79663
19053 * tree-predcom.c (combine_chains): Process refs in reverse order
19054 only for ZERO length chains, and add explaining comment.
19056 2017-02-23 Jeff Law <law@redhat.com>
19058 PR tree-optimization/79578
19059 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
19060 in call to operand_equal_p.
19062 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
19065 * config/i386/cpuid.h: Fix another undefined behavior.
19067 2017-02-23 Richard Biener <rguenther@suse.de>
19069 PR tree-optimization/79683
19070 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
19071 vector types for data-refs.
19073 2017-02-23 Martin Liska <mliska@suse.cz>
19075 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
19077 2017-02-23 Jakub Jelinek <jakub@redhat.com>
19079 PR middle-end/79665
19080 * internal-fn.c (get_range_pos_neg): Moved to ...
19081 * tree.c (get_range_pos_neg): ... here. No longer static.
19082 * tree.h (get_range_pos_neg): New prototype.
19083 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
19084 are known to be in between 0 and signed maximum inclusive, try to
19085 expand both unsigned and signed divmod and use the cheaper one from
19088 2017-02-22 Jeff Law <law@redhat.com>
19090 PR tree-optimization/79578
19091 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
19092 to compare base operands.
19094 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
19097 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
19098 gpc_reg_operand instead of fpr_reg_operand.
19100 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
19102 * config/mips/mips.c (mips_return_in_memory): Force FP
19103 vector types to be returned in memory for o32 ABI.
19105 2017-02-22 Jakub Jelinek <jakub@redhat.com>
19107 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
19108 instead of DW_TAG_member for static data member declarations and don't
19109 set no_linkage_name for static inline data members.
19110 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
19113 2017-02-22 Martin Liska <mliska@suse.cz>
19115 * doc/invoke.texi: Replace inequality signs with square brackets
19118 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19120 PR tree-optimization/68644
19121 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
19123 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
19126 * lra-constraints.c (simplify_operand_subreg): Handle
19127 WORD_REGISTER_OPERATIONS targets.
19129 2017-02-22 Jakub Jelinek <jakub@redhat.com>
19132 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
19133 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
19134 elimination by swapping fld*.
19136 2017-02-22 Richard Biener <rguenther@suse.de>
19138 PR tree-optimization/79673
19139 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
19140 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
19141 irrelevant address-space qualifiers and avoiding a
19142 ADDR_SPACE_CONVERT_EXPR from fold_convert.
19144 2017-02-22 Richard Biener <rguenther@suse.de>
19146 PR tree-optimization/79666
19147 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
19148 to not symbolically negate if that may introduce undefined
19151 2017-02-22 Martin Liska <mliska@suse.cz>
19154 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
19155 * data-streamer-out.c (streamer_write_gcov_count_stream):
19157 * value-prof.c (stream_out_histogram_value): Make assert more
19158 precise based on type of counter.
19160 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
19163 * config/i386/i386.md (standard_x87sse_constant_load splitter):
19164 Use nonimmediate_operand instead of memory_operand for operand 1.
19165 (float-extend standard_x87sse_constant_load splitter): Ditto.
19167 2017-02-21 Jeff Law <law@redhat.com>
19169 PR tree-optimization/79621
19170 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
19171 blocks with edges to themselves.
19173 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19176 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
19177 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
19178 Use gimple_call_builtin_p.
19181 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
19182 on temporarily removed DEBUG_INSNs.
19184 PR tree-optimization/79649
19185 * tree-loop-distribution.c (classify_partition): Give up on
19186 non-generic address space loads/stores.
19188 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
19190 * doc/loop.texi (Loop manipulation): Remove nonexistent
19191 tree_ssa_loop_version from the documentation.
19192 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
19194 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19197 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
19198 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
19199 * config/rs6000/rs6000.c: Include except.h.
19200 (rs6000_expand_split_stack_prologue): Call
19201 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
19203 2017-02-21 Martin Jambor <mjambor@suse.cz>
19206 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
19207 have been analyzed.
19209 2017-02-21 Martin Jambor <mjambor@suse.cz>
19211 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
19212 for backward compatibility only.
19213 * doc/invoke.texi (Option Summary): Remove all references to
19214 -fipa-cp-alignment.
19216 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
19220 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19222 * lra-constraints.c (curr_insn_transform): Handle
19223 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19225 2017-02-21 Martin Liska <mliska@suse.cz>
19227 * config/i386/i386.opt: Replace -masm-dialect with -masm.
19229 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
19231 PR translation/79638
19232 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
19234 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
19237 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
19238 (arm_function_ok_for_sibcall): Return false for an indirect call by
19239 descriptor if all the argument registers are used.
19240 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
19241 alignment of the function.
19243 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19245 PR tree-optimization/61441
19246 * simplify-rtx.c (simplify_const_unary_operation): For
19247 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
19248 the sNaN unmodified.
19250 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19252 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19253 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19254 instead of SYSTEM_HEADER_DIR.
19256 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
19257 Martin Liška <mliska@suse.cz>
19259 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
19260 Fix typos and grammar, use active voice, and clarify.
19262 2017-02-20 Marek Polacek <polacek@redhat.com>
19264 PR middle-end/79537
19265 * gimplify.c (gimplify_expr): Handle unused *&&L;.
19268 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
19270 2017-02-20 Jakub Jelinek <jakub@redhat.com>
19273 * config/i386/i386.c (ix86_expand_builtin): Handle
19274 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
19275 ix86_builtins_isa[fcode].isa as a requirement of those
19276 flags and any other flag in the bitmask.
19277 (ix86_init_mmx_sse_builtins): Use 0 instead of
19278 ~OPTION_MASK_ISA_64BIT as mask.
19279 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
19280 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
19281 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
19282 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
19284 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19287 * lra-constraints.c (split_reg): Check requested split mode
19288 is supported by the register.
19290 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19292 * lra-constraints.c (simplify_operand_subreg): Remove early
19295 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19298 * lra-constraints.c (curr_insn_transform): Tighten condition
19299 for converting SUBREG reloads from OP_OUT to OP_INOUT.
19301 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19304 * lra-constraints.c (curr_insn_transform): Handle
19305 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19307 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
19310 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
19312 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
19314 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
19317 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
19320 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19322 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
19325 2017-02-18 Jakub Jelinek <jakub@redhat.com>
19327 * final.c (last_columnnum, override_columnnum): New variables.
19328 (final_start_function): Set last_columnnum, pass it to begin_prologue
19329 hook and pass 0 to dwarf2out_begin_prologue.
19330 (final_scan_insn): Update override_columnnum. Pass last_columnnum
19331 to source_line debug hook.
19332 (notice_source_line): Compute last_columnnum and for debug_column_info
19333 return true on column changes.
19334 * debug.h (struct gcc_debug_hooks): Add column argument to
19335 source_line and begin_prologue hooks.
19336 (debug_nothing_int_charstar_int_bool): Remove prototype.
19337 (debug_nothing_int_int_charstar,
19338 debug_nothing_int_int_charstar_int_bool): New prototypes.
19339 (dwarf2out_begin_prologue): Add column argument.
19340 * debug.c (do_nothing_debug_hooks): Adjust source_line and
19341 begin_prologue hooks.
19342 (debug_nothing_int_charstar_int_bool): Remove.
19343 (debug_nothing_int_int_charstar,
19344 debug_nothing_int_int_charstar_int_bool): New functions.
19345 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
19346 through to dwarf2out_source_line.
19347 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
19348 (dwarf2out_source_line): Add column argument, emit it if requested.
19349 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
19351 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19352 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19353 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
19354 through to dwarf2out_begin_prologue.
19355 (vmsdbgout_source_line): Add column argument, pass it through to
19356 dwarf2out_source_line.
19357 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
19358 dbxout_source_line caller.
19359 (dbxout_source_line): Add column argument.
19361 * common.opt (gno-column-info, gcolumn-info): New options.
19362 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
19363 (check_die): Also test for multiple DW_AT_decl_column attributes.
19364 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
19365 DW_AT_decl_column if requested.
19366 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
19368 (gen_variable_die): Likewise.
19369 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
19370 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
19373 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
19374 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
19375 (ix86_handle_option): Handle OPT_m3dnowa.
19376 * doc/invoke.texi (-m3dnowa): Document.
19377 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
19378 -m3dnowa instead of -m3dnow -march=athlon.
19381 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
19382 instead of gcc_assert for K, r and R code checks. Formatting fixes.
19384 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19387 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
19388 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
19389 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
19390 generator for vsx_xxpermdi_<mode>_be.
19391 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
19392 force big-endian semantics.
19393 (vsx_xxpermdi_<mode>_be): New define_expand with same
19394 implementation as previous version of vsx_xxpermdi_<mode>.
19396 2017-02-17 Jakub Jelinek <jakub@redhat.com>
19398 PR tree-optimization/79327
19399 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
19400 variable, its initialization and use.
19402 2017-02-17 Julia Koval <julia.koval@intel.com>
19404 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
19405 (OPTION_MASK_ISA_PKU_UNSET): New.
19406 (ix86_handle_option): Handle -mrdpid.
19407 * config/i386/cpuid.h (bit_RDPID): New.
19408 * config/i386/driver-i386.c (host_detect_local_cpu):
19409 Detect RDPID feature.
19410 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
19411 * config/i386/i386-c.c (ix86_target_macros_internal):
19413 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
19414 (ix86_valid_target_attribute_inner_p): Add "rdpid".
19415 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
19416 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
19417 * config/i386/i386.md (define_insn "rdpid"): New.
19418 * config/i386/i386.opt Add -mrdpid.
19419 * config/i386/immintrin.h (_rdpid_u32): New.
19421 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
19423 PR rtl-optimization/79541
19424 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
19425 instead of transforming it into USE.
19427 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
19429 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
19430 If HONOR_SNANS (SFmode) force the input to a register.
19431 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
19432 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
19433 an frsp or similar insn.
19435 2017-02-17 Martin Liska <mliska@suse.cz>
19437 PR rtl-optimization/79577
19438 * params.def (selsched-max-sched-times): Increase minimum to 1.
19440 2017-02-17 Martin Liska <mliska@suse.cz>
19442 PR rtl-optimization/79574
19443 * gcse.c (want_to_gcse_p): Prevent integer overflow.
19445 2017-02-17 Martin Liska <mliska@suse.cz>
19447 PR tree-optimization/79529
19448 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
19449 ssa_defined_default_def_p to handle cases which are implicitly
19451 * tree-ssa.c (ssa_defined_default_def_p): New function.
19452 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
19453 which are implicitly defined.
19454 * tree-ssa.h (ssa_defined_default_def_p): Declare.
19456 2017-02-17 Richard Biener <rguenther@suse.de>
19458 PR middle-end/79576
19459 * params.def (max-ssa-name-query-depth): Limit to 10.
19461 2017-02-17 Richard Biener <rguenther@suse.de>
19463 PR tree-optimization/79552
19464 * tree-ssa-structalias.c (visit_loadstore): Properly verify
19467 2017-02-17 Richard Biener <rguenther@suse.de>
19470 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
19472 2017-02-17 Marek Polacek <polacek@redhat.com>
19474 PR middle-end/79536
19475 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
19476 (fold_negate_expr): New wrapper.
19478 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
19480 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
19481 Correct terminology and de-emphasize pre-standard behavior.
19483 2017-02-16 Alan Modra <amodra@gmail.com>
19485 PR rtl-optimization/79286
19486 * ira.c (def_dominates_uses): New function.
19487 (update_equiv_regs): Don't create an equivalence for insns that
19488 may trap where the register def does not dominate the use.
19490 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
19492 PR rtl-optimization/78127
19493 * lra.c (lra): Call lra_eliminate before finish the loop after
19496 2017-02-16 Richard Biener <rguenther@suse.de>
19498 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
19500 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
19501 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
19502 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
19503 (isl_val_int_from_wi): New function.
19504 (extract_affine_gmp): Rename to ...
19505 (extract_affine_wi): ... this, take a widest_int.
19506 (extract_affine_int): Just wrap extract_affine_wi.
19507 (add_param_constraints): Use isl_val_int_from_wi.
19508 (add_loop_constraints): Likewise, and extract_affine_wi.
19510 2017-02-15 Jeff Law <law@redhat.com>
19512 PR middle-end/79521
19513 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
19514 ira_init_register_move_cost_if_necessary.
19516 2017-02-15 Martin Sebor <msebor@redhat.com>
19518 PR middle-end/32003
19519 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
19520 removed in a prior commit.
19522 2017-02-15 Bin Cheng <bin.cheng@arm.com>
19524 PR tree-optimization/79347
19525 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
19526 counters during peeling.
19528 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
19530 * Makefile.in (site.exp): Remove "set ISLVER".
19532 2017-02-15 Jakub Jelinek <jakub@redhat.com>
19535 * real.c (real_from_integer): Call real_convert even for decimal.
19537 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
19540 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
19542 2017-02-14 Andrew Pinski <apinski@cavium.com>
19544 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
19545 cores and change the partno/implementer to be correct.
19546 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
19547 the 'B" as the implementer.
19548 * config/aarch64/aarch64-tune.md: Regenerate.
19550 2017-02-14 Carl Love <cel@us.ibm.com>
19552 * config/rs6000/rs6000.c: Add case statement entry to make the
19553 xvcvuxdsp built-in argument unsigned.
19554 * config/rs6000/vsx.md: Fix the source and return operand types so they
19555 match the instruction definitions from the ISA document. Fix typo
19556 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
19559 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
19562 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
19563 member early_clobber_alts.
19564 * lra-lives.c (reg_early_clobber_p): New.
19565 (process_bb_lives): Use it.
19566 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
19567 (debug_operand_data): Initialize early_clobber_alts.
19568 (setup_operand_alternative): Set up early_clobber_alts.
19569 (collect_non_operand_hard_regs): Ditto. Pass early clobber
19570 alternatives to new_insn_reg.
19571 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
19573 (lra_update_insn_regno_info): Pass the new arg.
19575 2017-02-14 Jakub Jelinek <jakub@redhat.com>
19577 PR middle-end/79505
19578 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
19579 (new_oacc_loop_raw): Don't clear already cleared fields.
19582 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
19583 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
19584 _mm512_prefetch_i64gather_ps): New inline functions and macros.
19586 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
19589 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
19591 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
19594 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
19595 the extra instruction to the right place to store 128-bit constant
19598 2017-02-14 Martin Sebor <msebor@redhat.com>
19600 PR middle-end/79448
19601 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
19602 warning for strings of unknown length.
19604 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
19606 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
19608 2017-02-14 Jeff Law <law@redhat.com>
19611 * ira-costs.c (scan_one_insn): Initialize register move costs
19612 for pseudos seen in USE/CLOBBER insns.
19614 PR tree-optimization/79095
19615 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
19616 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
19617 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
19618 if the operands are known to be not equal, then the resulting range
19620 (intersect_ranges): If the new range is ~[0,0] and the old range is
19621 wide, then prefer ~[0,0].
19622 * tree-vrp.c (overflow_comparison_p_1): New function.
19623 (overflow_comparison_p): New function.
19624 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
19625 if NAME is used in an overflow test.
19626 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
19627 overflow check that can be expressed as an equality test, then adjust
19628 ops to be that equality test.
19630 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19632 * config/s390/s390-builtin-types.def: Remove flags argument.
19633 * config/s390/s390.c (s390_init_builtins): Likewise.
19635 2017-02-14 Martin Liska <mliska@suse.cz>
19637 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
19638 vector. Fix trailing white spaces.
19640 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
19642 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
19645 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19647 PR rtl-optimization/68664
19648 * config/arm/arm.c (arm_sched_can_speculate_insn):
19649 New function. Declare prototype.
19650 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
19652 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19654 PR rtl-optimization/68664
19655 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
19657 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
19659 2017-02-14 Amit Pawar <amit.pawar@amd.com>
19661 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
19662 max skip bytes for function, loop and jump.
19664 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19666 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
19667 ABS_EXPR for gimple dump.
19669 2017-02-14 Jakub Jelinek <jakub@redhat.com>
19672 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
19674 PR tree-optimization/79408
19675 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
19676 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
19677 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
19678 also if rhs1 is INTEGER_CST.
19680 2017-02-14 Richard Biener <rguenther@suse.de>
19682 PR middle-end/79432
19683 * tree-into-ssa.c (insert_phi_nodes): When the function can
19684 have abnormal edges rewrite SSA names with broken use-def
19685 dominance out of SSA and register them for PHI insertion.
19687 2017-02-13 Martin Sebor <msebor@redhat.com>
19689 PR middle-end/79496
19690 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
19691 clearing info.nowrite flag when snprintf size argument is a range.
19693 2017-02-13 Jakub Jelinek <jakub@redhat.com>
19695 * cprop.c (cprop_jump): Add missing space in string literal.
19696 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
19697 (get_constraint_for_component_ref): Likewise.
19698 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
19699 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
19700 * lra-constraints.c (process_alt_operands): Likewise.
19701 * ipa-inline.c (inline_small_functions): Likewise.
19702 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
19703 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19704 * trans-mem.c (diagnose_tm_1_op): Likewise.
19705 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
19706 (grid_parallel_clauses_gridifiable): Likewise.
19708 * config/nvptx/mkoffload.c (process): Add space in between
19711 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
19712 "MOD4_SSE_REGS" and "ALL_REGS".
19714 * spellcheck.c (test_data): Add , in between "foo" and "food".
19716 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19719 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
19720 boundary crossing check and subsequent code generation agree.
19722 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19724 * config/aarch64/aarch64.c (has_memory_op): Delete.
19725 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
19728 2017-02-13 Jakub Jelinek <jakub@redhat.com>
19730 PR rtl-optimization/79388
19731 PR rtl-optimization/79450
19732 * combine.c (distribute_notes): When removing TEM_INSN for which
19733 corresponding dest has last value recorded, invalidate that last
19736 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19738 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
19739 of explicit '@'. Add missing assembly comment marker on branch costs
19742 2017-02-13 Nathan Sidwell <nathan@acm.org>
19744 * gengtype-lex.l (<in_struct>): Add '/'.
19746 2017-02-13 Martin Liska <mliska@suse.cz>
19749 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
19751 2017-02-13 Richard Biener <rguenther@suse.de>
19753 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
19755 * configure: Re-generate.
19756 * config.in: Likewise.
19757 * graphite-dependences.c: Simplify as if
19758 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
19759 * graphite-isl-ast-to-gimple.c: Likewise.
19760 * graphite-optimize-isl.c: Likewise.
19761 * graphite-poly.c: Likewise.
19762 * graphite-sese-to-poly.c: Likewise.
19763 * graphite.h: Likewise.
19764 * toplev.c: Include isl/version.h and use isl_version () for
19765 printing the ISL version.
19766 * doc/install.texi: Update ISL requirement.
19768 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
19770 * doc/standards.texi (Standards): Update reference to
19773 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
19775 * doc/extend.texi (Named Address Spaces): sourceware.org now
19777 * doc/install.texi (Binaries): Ditto.
19780 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
19782 * doc/cpp.texi: Replace "stringify"/"stringification" with C
19783 standard terminology "stringize"/"stringizing" throughout.
19784 * doc/cppinternals.texi: Likewise.
19786 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
19788 * doc/extend.texi: Fix some spelling mistakes and typos.
19789 * doc/invoke.texi: Likewise.
19791 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
19794 * params.def (inline-min-speedup) Change from 10 to 8.
19796 2017-02-11 Jakub Jelinek <jakub@redhat.com>
19798 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
19801 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
19804 * ipa-inline-analysis.c (get_minimal_bb): New function.
19805 (record_modified): Use it.
19806 (remap_edge_change_prob): Handle also ancestor functions.
19808 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
19810 * doc/contrib.texi (Contributors): Remove broken link into
19811 the Mauve CVS repository.
19813 2017-02-11 Jakub Jelinek <jakub@redhat.com>
19815 PR middle-end/79454
19816 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
19817 result computation whenever lhs doesn't have vector mode, not
19818 just when it has BLKmode.
19820 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
19822 * doc/makefile.texi (profiledbootstrap): Refer to the
19823 installation instructions only in textual form.
19825 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19828 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
19830 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
19832 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
19833 (Specific): Update mingw-w64 reference.
19835 (Specific): Remove broken link to Renesas RX processor.
19837 2017-02-10 Richard Biener <rguenther@suse.de>
19839 * toplev.c (process_options): Do not mention obsolete graphite
19840 options when printing sorry message about missing graphite support.
19841 Mention -floop-nest-optimize.
19843 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
19845 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
19846 (vtst_p16): Likewise.
19847 (vtstq_p8): Likewise.
19848 (vtstq_p16): Likewise.
19850 (vtstq_p64): Likewise.
19851 * config/arm/arm_neon.h (vgetq_lane_p64): New.
19852 (vset_lane_p64): New.
19853 (vsetq_lane_p64): New.
19855 2017-02-10 Jakub Jelinek <jakub@redhat.com>
19857 PR tree-optimization/79411
19858 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
19859 stmt operands are SSA_NAMEs used in abnormal phis.
19860 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
19863 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
19866 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
19869 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
19871 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
19873 2017-02-09 Jakub Jelinek <jakub@redhat.com>
19875 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
19879 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
19880 not arbitrary TREE_CONSTANT.
19883 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
19884 "omp declare target link" attribute unless is_global_var.
19885 * omp-offload.c (find_link_var_op): Likewise.
19887 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
19888 Chung-Lin Tang <cltang@codesourcery.com>
19890 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
19892 (gimplify_adjust_omp_clauses): Don't delete TILE.
19893 (gimplify_omp_for): Deal with TILE.
19894 * internal-fn.c (expand_GOACC_TILE): New function.
19895 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
19897 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
19898 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
19900 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
19901 avoid DIV for outermost collapse var.
19902 (expand_oacc_for): Insert tile element loop as needed. Adjust.
19903 Remove out of date comments, fix whitespace.
19904 * omp-general.c (omp_extract_for_data): Deal with tiling.
19905 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
19906 adjust OLF_DIM_BASE value.
19907 (struct omp_for_data): Add tiling field.
19908 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
19909 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
19910 for auto loops. Remove default auto determining, moved to
19911 oacc_loop_fixed_partitions.
19912 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
19913 stmts, add e_mask field.
19914 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
19915 (oacc_thread_numbers): Use oacc_dim_call.
19916 (oacc_xform_tile): New.
19917 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
19918 (finish_oacc_loop): Adjust for ifns vector.
19919 (oacc_loop_discover_walk): Append loop abstraction sites to list,
19920 add case for GOACC_TILE fns.
19921 (oacc_loop_xform_loop): Delete.
19922 (oacc_loop_process): Iterate over call list directly, and add
19923 handling for GOACC_TILE fns.
19924 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
19926 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
19927 vector partitioning to outer loops. Assign 2 partitions to loops
19928 when available. Add TILE handling.
19929 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
19930 (execite_oacc_device_lower): Process GOACC_TILE fns,
19931 ignore unknown specs.
19932 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
19933 * tree.c (omp_clause_num_ops): Adjust TILE ops.
19934 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
19936 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
19938 * configure.ac (ACX_BUGURL): Update.
19939 * configure: Regenerate.
19941 2017-02-09 Richard Biener <rguenther@suse.de>
19943 PR tree-optimization/69823
19944 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
19945 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
19947 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
19949 * config/arc/arc-c.def: Add __NPS400__ definition.
19950 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
19951 (TARGET_NPS400): Define.
19953 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
19955 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
19957 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
19958 pointer, arch_info.
19959 (arc_cpu_types): Fill the arch_info field with a pointer into the
19960 arc_arch_types table.
19961 (arc_selected_cpu): Declare.
19962 * config/arc/arc.c (arc_selected_cpu): Make global.
19963 (arc_selected_arch): Delete.
19964 (arc_base_cpu): Delete.
19965 (arc_override_options): Remove references to deleted variables,
19966 update access to arch information.
19967 (ARC_OPT): Update access to arch information.
19968 (ARC_OPTX): Likewise.
19969 * config/arc/arc.h (arc_base_cpu): Remove declaration.
19970 (TARGET_ARC600): Update access to arch information.
19971 (TARGET_ARC601): Likewise.
19972 (TARGET_ARC700): Likewise.
19973 (TARGET_EM): Likewise.
19974 (TARGET_HS): Likewise.
19975 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
19978 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
19981 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
19982 condition/operands for integer GE/LE/GEU/LEU operations.
19984 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
19986 PR translation/79397
19987 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
19990 2017-02-08 Martin Jambor <mjambor@suse.cz>
19993 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
19994 whether allocation happened.
19995 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
19996 nothing was allocated.
19998 2017-02-08 Jakub Jelinek <jakub@redhat.com>
20000 PR tree-optimization/79408
20001 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
20002 constant, but SSA_NAME with a known integer range, use the minimum
20003 of that range instead of op1 to determine if modulo can be replaced
20004 with its first operand.
20006 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20008 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
20010 2017-02-08 Richard Biener <rguenther@suse.de>
20012 PR tree-optimization/71824
20013 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
20014 Check all loops contained in the merged region.
20016 2017-02-07 Andrew Pinski <apinski@cavium.com>
20018 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
20020 2017-02-07 Andrew Pinski <apinski@cavium.com>
20022 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
20023 (thunderxt88): Likewise.
20024 (thunderxt81): Disable LSE and change v8.1 to v8.
20025 (thunderxt83): Likewise.
20027 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20028 Richard Biener <rguenther@suse.de>
20030 PR middle-end/79399
20031 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
20032 type from int to size_t.
20033 * ira-costs.c (struct_costs_size): Change type from int to size_t.
20035 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20037 PR rtl-optimization/79386
20038 * cprop.c (bypass_conditional_jumps): Initialize
20039 bypass_last_basic_block already before splitting bbs after
20040 unconditional traps...
20041 (bypass_conditional_jumps): ... rather than here.
20044 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
20045 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
20046 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
20047 fix -masm=intel patterns.
20049 2017-02-07 Richard Biener <rguenther@suse.de>
20051 PR tree-optimization/79256
20052 PR middle-end/79278
20053 * builtins.c (get_object_alignment_2): Use min_align_of_type
20054 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
20055 and ADJUST_FIELD_ALIGN.
20057 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
20059 * doc/tm.texi: Regenerate.
20060 * stor-layout.c (layout_decl): Adjust.
20061 (update_alignment_for_field): Likewise.
20062 (place_field): Likewise.
20063 (min_align_of_type): Likewise.
20064 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
20065 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
20066 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
20067 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
20068 * config/frv/frv.c (frv_adjust_field_align): Likewise.
20069 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
20070 * config/i386/i386.c (x86_field_alignment): Likewise.
20071 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
20072 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
20073 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
20074 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
20075 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
20076 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
20080 2017-01-30 Richard Biener <rguenther@suse.de>
20082 PR tree-optimization/79256
20083 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
20084 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
20087 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
20089 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
20090 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
20091 builtins to SImode and emit a zero-extend, if necessary.
20093 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20095 * docs/invoke.texi (RISC-V Options): Alphabetize.
20097 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20099 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
20102 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20104 * config/riscv/riscv.c: New file.
20105 * common/config/riscv/riscv-common.c: Likewise.
20106 * config.gcc: Likewise.
20107 * config/riscv/constraints.md: Likewise.
20108 * config/riscv/elf.h: Likewise.
20109 * config/riscv/generic.md: Likewise.
20110 * config/riscv/linux.h: Likewise.
20111 * config/riscv/multilib-generator: Likewise.
20112 * config/riscv/peephole.md: Likewise.
20113 * config/riscv/pic.md: Likewise.
20114 * config/riscv/predicates.md: Likewise.
20115 * config/riscv/riscv-builtins.c: Likewise.
20116 * config/riscv/riscv-c.c: Likewise.
20117 * config/riscv/riscv-ftypes.def: Likewise.
20118 * config/riscv/riscv-modes.def: Likewise.
20119 * config/riscv/riscv-opts.h: Likewise.
20120 * config/riscv/riscv-protos.h: Likewise.
20121 * config/riscv/riscv.h: Likewise.
20122 * config/riscv/riscv.md: Likewise.
20123 * config/riscv/riscv.opt: Likewise.
20124 * config/riscv/sync.md: Likewise.
20125 * config/riscv/t-elf-multilib: Likewise.
20126 * config/riscv/t-linux: Likewise.
20127 * config/riscv/t-linux-multilib: Likewise.
20128 * config/riscv/t-riscv: Likewise.
20129 * configure.ac: Likewise.
20130 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
20131 Waterman as RISC-V maintainers.
20132 * doc/install.texi: Add RISC-V entries.
20133 * doc/invoke.texi: Add RISC-V options section.
20134 * doc/md.texi: Add RISC-V constraints section.
20135 * configure: Regenerated.
20137 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
20140 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
20141 false values to be constant vectors with all 0 or all 1 bits set.
20142 (vcondu<mode><mode>): Likewise.
20143 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
20145 (fpmask_comparison_operator): Update comment.
20146 (vecint_comparison_operator): New predicate.
20147 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
20148 vector conditionals when the true and false values are constant
20149 vectors with all 0 bits or all 1 bits set.
20151 2017-02-06 Martin Sebor <msebor@redhat.com>
20153 PR tree-optimization/79376
20154 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
20156 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
20158 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
20159 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
20160 to simplify split condition.
20162 2017-02-06 Jakub Jelinek <jakub@redhat.com>
20164 * omp-expand.c (oxpand_omp_atomic_fetch_op,
20165 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
20168 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
20170 PR rtl-optimization/68664
20171 * target.def (can_speculate_insn): New hook.
20172 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
20173 * doc/tm.texi: Regenerate.
20174 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
20175 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
20176 (rs6000_sched_can_speculate_insn): New function.
20178 2017-02-06 Jakub Jelinek <jakub@redhat.com>
20180 PR tree-optimization/79284
20181 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
20182 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
20183 vectorizable_mask_load_store, vectorizable_operation,
20184 vect_is_simple_cond, get_same_sized_vectype): Use it instead
20185 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
20186 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
20187 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
20188 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
20189 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
20190 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
20191 is_gimple_assign (stmt). Replace another such test with
20192 is_gimple_assign (stmt).
20194 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
20197 * config/avr/avr.c (rtl-iter.h): Include it.
20198 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
20199 (avr_legitimate_combined_insn): ...and implementation.
20201 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
20203 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
20204 * config/s390/s390.c (s390_const_operand_ok)
20205 (s390_canonicalize_comparison, s390_extract_part)
20206 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
20207 (s390_contiguous_bitmask_p, s390_rtx_costs)
20208 (legitimize_pic_address): Likewise.
20209 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
20210 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
20211 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
20212 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
20213 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
20215 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
20217 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
20218 REGNO($0) == REGNO($1).
20220 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20222 * config/s390/linux.h(SIZE_TYPE): Add comment.
20224 2017-02-06 Julian Brown <julian@codesourcery.com>
20225 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20226 Virendra Pathak <virendra.pathak@broadcom.com>
20228 * config/aarch64/aarch64-cores.def: Change the scheduler
20230 * config/aarch64/aarch64.md: Include thunderx2t99.md.
20231 * config/aarch64/thunderx2t99.md: New file.
20233 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20235 * doc/standards.texi (Go Language): Update link to language
20238 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20240 * tree-eh.c (lower_resx): Sanitize profile.
20241 (cleanup_empty_eh_move_lp): Likewise.
20243 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20246 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
20248 * cfgloopmanip.h (loop_version): Update prototype.
20249 * modulo-sched.c (sms_schedule): Update call of loop_version.
20250 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
20251 * tree-parloops.c (gen_parallel_loop): Likewise.
20252 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
20253 * tree-ssa-loop-split.c (split_loop): Likewise.
20254 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
20255 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
20257 2017-02-05 Martin Liska <mliska@suse.cz>
20260 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
20262 (print_operand_address): Initialize a struct to zero.
20264 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20266 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
20267 garbage collector only in textual form.
20269 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20271 * doc/extend.texi (x86 specific memory model extensions for
20272 transactional memory): Simplify a phrase.
20274 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
20277 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
20278 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
20279 (atomic_storedi_1): Likewise.
20281 2017-02-04 Jakub Jelinek <jakub@redhat.com>
20283 PR tree-optimization/79338
20284 * tree-parloops.c (gather_scalar_reductions): Don't call
20285 vect_analyze_loop_form for loop->inner before destroying loop's
20288 2017-02-03 Martin Sebor <msebor@redhat.com>
20290 PR tree-optimization/79327
20291 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
20292 when precision has resulted in leading zeros.
20293 (format_integer): Adjust the likely counter to assume an unknown
20294 argument that may be zero is non-zero.
20296 2017-02-03 Jason Merrill <jason@redhat.com>
20299 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
20300 avoid copying non-taken branch.
20302 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20304 PR tree-optimization/79340
20305 * tree-vect-loop.c (vectorizable_reduction): Release
20306 vec_defs elements after safe_splicing them into other vectors.
20309 PR tree-optimization/79327
20310 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
20311 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
20313 (format_integer): Use wide_int_to_tree instead of build_int_cst
20314 + to_?hwi. If argmin is NULL, just set argmin and argmax to
20315 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
20316 of shortest and longest sequence.
20318 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
20320 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
20321 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
20323 2017-02-03 Walter Lee <walt@tilera.com>
20326 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
20327 after initial stackframe link reg save.
20328 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
20330 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20333 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
20334 wu for stxssp alternative.
20336 2017-02-03 Martin Sebor <msebor@redhat.com>
20338 PR tree-optimization/79352
20339 * gimple-fold.c (get_range_strlen): Add argument.
20340 (get_range_strlen): Change return type to bool.
20341 (get_maxval_strlen): Pass in a dummy argument.
20342 * gimple-fold.h (get_range_strlen): Change return type to bool.
20343 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
20344 * tree.h (array_at_struct_end_p): Add argument.
20345 * tree.c (array_at_struct_end_p): Handle it.
20347 2017-02-03 Martin Liska <mliska@suse.cz>
20350 * multiple_target.c (create_dispatcher_calls): Redirect edge
20351 from a caller of a dispatcher.
20352 (expand_target_clones): Make the clones local.
20353 (ipa_target_clone): Do both target clones and resolvers.
20354 (ipa_dispatcher_calls): Remove the pass.
20355 (pass_dispatcher_calls::gate): Likewise.
20356 (make_pass_dispatcher_calls): Likewise.
20357 * passes.def (pass_target_clone): Put as very first IPA early
20360 2017-02-03 Martin Liska <mliska@suse.cz>
20362 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
20363 in case of a function with ifunc attribute.
20365 2017-02-03 Martin Liska <mliska@suse.cz>
20367 * cgraph.c (cgraph_node::dump): Dump function version info.
20368 * symtab.c (symtab_node::dump_base): Add missing new line.
20370 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20372 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
20373 (ifcombine_ifandif): Use it.
20375 2017-02-03 Martin Liska <mliska@suse.cz>
20377 * doc/invoke.texi: Document default value for
20378 use-after-scope-direct-emission-threshold.
20380 2017-02-03 Martin Liska <mliska@suse.cz>
20382 PR tree-optimization/79339
20383 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
20384 (format_floating): Likewise.
20386 2017-02-03 Martin Liska <mliska@suse.cz>
20389 * ipa-prop.c (ipa_node_params_t::insert): Remove current
20391 (ipa_node_params_t::remove): Likewise.
20392 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
20393 initialization from removed ipa_node_params_t::insert.
20394 (ipa_node_params::~ipa_node_params): Move from removed
20395 ipa_node_params_t::release.
20396 * symbol-summary.h (symbol_summary::m_released): New member.
20397 Do not release a summary twice. Do not allow to call finalizer
20398 for types of a summary that live in GGC memory.
20400 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20402 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
20405 2017-02-02 Martin Sebor <msebor@redhat.com>
20407 PR middle-end/79275
20408 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
20409 (format_string): Tighten up the range of output for non-constant
20410 strings and correct the expected range for wide non-constant strings.
20412 2017-02-02 Martin Sebor <msebor@redhat.com>
20414 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
20416 PR middle-end/32003
20417 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
20419 (-fdump-tree-@var): Add to index and document how to come up
20420 with pass-specific option and dump file names.
20421 (-fdump-passes): Clarify where to look for output.
20423 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20425 PR middle-end/77445
20426 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
20427 statistics of the analyzed path; allow threading for speed when
20428 any of BBs along the path are optimized for speed.
20430 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
20432 PR middle-end/78468
20433 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
20434 settings of the virtual registers.
20437 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
20439 * explow.c (get_dynamic_stack_size): Take known alignment of stack
20440 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
20443 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20445 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
20446 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
20448 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20450 * config/s390/s390.md: Add missing comments with the expanded
20452 * config/s390/vector.md: Likewise.
20453 * config/s390/vx-builtins.md: Likewise.
20455 2017-02-02 Jakub Jelinek <jakub@redhat.com>
20458 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
20459 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
20460 conditions on a single line.
20462 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20464 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
20465 __S390_VX__ to __VX__.
20467 2017-02-01 Andrew Pinski <apinski@cavium.com>
20469 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
20470 stmt_info to record_stmt_cost.
20471 (vect_get_known_peeling_cost): Pass stmt_info if known to
20473 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
20474 cpu_vector_cost field into
20475 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
20476 field into vec_int_stmt_cost and vec_fp_stmt_cost.
20477 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
20478 splitting of scalar_stmt_cost and vec_stmt_cost.
20479 (thunderx_vector_cost): Likewise.
20480 (cortexa57_vector_cost): LIkewise.
20481 (exynosm1_vector_cost): Likewise.
20482 (xgene1_vector_cost): Likewise.
20483 (thunderx2t99_vector_cost): Improve after the splitting of the two
20485 (aarch64_builtin_vectorization_cost): Update for the splitting of
20486 scalar_stmt_cost and vec_stmt_cost.
20488 2017-02-01 Torvald Riegel <triegel@redhat.com>
20489 Richard Henderson <rth@redhat.com>
20491 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
20492 conditional on existance of a fast atomic load.
20493 * optabs-query.c (can_atomic_load_p): New function.
20494 * optabs-query.h (can_atomic_load_p): Declare it.
20495 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
20496 no fast atomic load is available for the particular size of access.
20497 (expand_atomic_compare_and_swap): Likewise.
20498 (expand_atomic_load): Likewise.
20499 (expand_atomic_store): Likewise.
20500 (expand_atomic_fetch_op): Likewise.
20501 * testsuite/lib/target-supports.exp
20502 (check_effective_target_sync_int_128): Remove x86 because it provides
20503 no fast atomic load.
20504 (check_effective_target_sync_int_128_runtime): Likewise.
20506 2017-02-01 Richard Biener <rguenther@suse.de>
20508 * graphite.c: Include tree-vectorizer.h for find_loop_location.
20509 (graphite_transform_loops): Provide opt-info for optimized nests.
20510 * tree-parloop.c (parallelize_loops): Provide opt-info for
20511 parallelized loops.
20513 2017-02-01 Richard Biener <rguenther@suse.de>
20515 PR middle-end/79315
20516 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
20517 was not set before.
20519 2017-02-01 Richard Biener <rguenther@suse.de>
20521 PR tree-optimization/71824
20522 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
20523 Verify the loops are valid in the merged SESE region.
20524 (scop_detection::can_represent_loop_1): Check analyzing the
20525 evolution of the number of iterations in the region succeeds.
20527 2017-01-31 Ian Lance Taylor <iant@golang.org>
20529 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
20530 REG_ARGS_SIZE note to 32-bit push insns and call insn.
20532 2017-01-31 David Malcolm <dmalcolm@redhat.com>
20534 PR preprocessor/79210
20535 * input.c (get_substring_ranges_for_loc): Replace line_width
20536 assertion with error-handling.
20538 2017-01-31 Richard Biener <rguenther@suse.de>
20540 PR tree-optimization/77318
20541 * graphite-sese-to-poly.c (extract_affine): Fix assert.
20542 (create_pw_aff_from_tree): Take loop parameter.
20543 (add_condition_to_pbb): Pass loop of the condition to
20544 create_pw_aff_from_tree.
20546 2017-01-31 Jakub Jelinek <jakub@redhat.com>
20548 * config/s390/s390.c (s390_asan_shadow_offset): New function.
20549 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
20551 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
20555 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
20557 (convert_int_to_float128): Likewise.
20558 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
20559 (convert_int_to_float128): Likewise.
20560 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
20561 (UNSPEC_IEEE128_CONVERT): Likewise.
20562 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
20563 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
20564 Use local variables for IBM extended format.
20565 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
20566 (fix_trunc<mode>si2_fprs): Likewise.
20567 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
20568 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
20569 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
20570 to know that we can now have integers of all sizes in vector
20572 (fix<uns>_<mode>di2_hw): Likewise.
20573 (float<uns>_<mode>si2_hw): Likewise.
20574 (fix_<mode>si2_hw): Likewise.
20575 (fixuns_<mode>si2_hw): Likewise.
20576 (float<uns>_<mode>di2_hw): Likewise.
20577 (float_<mode>di2_hw): Likewise.
20578 (float_<mode>si2_hw): Likewise.
20579 (floatuns_<mode>di2_hw): Likewise.
20580 (floatuns_<mode>si2_hw): Likewise.
20581 (xscvqp<su>wz_<mode>): Delete, no longer used.
20582 (xscvqp<su>dz_<mode>): Likewise.
20583 (xscv<su>dqp_<mode>): Likewise.
20584 (ieee128_mfvsrd_64bit): Likewise.
20585 (ieee128_mfvsrd_32bit): Likewise.
20586 (ieee128_mfvsrwz): Likewise.
20587 (ieee128_mtvsrw): Likewise.
20588 (ieee128_mtvsrd_64bit): Likewise.
20589 (ieee128_mtvsrd_32bit): Likewise.
20591 2017-01-31 Martin Liska <mliska@suse.cz>
20594 * ipa-prop.c (ipa_free_all_node_params): Call release method
20595 instead of ~sumbol_summary to not to trigger double times
20598 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
20600 PR tree-optimization/71691
20601 * bitmap.h (class auto_bitmap): New.
20602 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
20603 is_maybe_undefined instead of ssa_undefined_value_p.
20605 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20607 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
20608 __S390_ARCH_LEVEL__ to __ARCH__.
20610 2017-01-31 Jakub Jelinek <jakub@redhat.com>
20612 PR tree-optimization/79267
20613 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
20614 if should_remove_lhs_p is true.
20616 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
20619 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
20620 (add_alignment_attribute): New.
20621 (base_type_die): Add alignment attribute.
20622 (subrange_type_die): Likewise.
20623 (modified_type_die): Likewise.
20624 (gen_array_type_die): Likewise.
20625 (gen_descr_array_type_die: Likewise.
20626 (gen_enumeration_type_die): Likewise.
20627 (gen_subprogram_die): Likewise.
20628 (gen_variable_die): Likewise.
20629 (gen_field_die): Likewise.
20630 (gen_ptr_to_mbr_type_die): Likewise.
20631 (gen_struct_or_union_type_die): Likewise.
20632 (gen_subroutine_type_die): Likewise.
20633 (gen_typedef_die): Likewise.
20634 (base_type_cmp): Compare alignment attribute.
20636 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20639 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
20640 (setb_unsigned) New pattern for setb with CCUNS.
20641 * config/rs6000/rs6000.c (expand_block_compare): Use a different
20642 subfc./subfe sequence to avoid overflow problems. Generate a
20643 shorter sequence with cmpld/setb for power9.
20644 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
20645 for generating subfc. instruction.
20646 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
20647 now uses this instruction.
20649 2017-01-30 Ian Lance Taylor <iant@google.com>
20652 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
20653 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
20655 2017-01-30 Martin Sebor <msebor@redhat.com>
20657 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
20658 Move constant to the right of a relational operator.
20659 (get_mpfr_format_length, format_character, format_string): Ditto.
20660 (should_warn_p, maybe_warn): Same.
20662 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
20664 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
20667 * asan.c (get_translation_unit_decl): Remove function.
20668 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
20670 2017-01-30 Martin Liska <mliska@suse.cz>
20672 PR gcov-profile/79259
20673 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
20674 -fprofile-generate.
20676 2017-01-30 Martin Liska <mliska@suse.cz>
20679 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
20680 Initialize variables with NULL value.
20682 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
20685 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
20687 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
20689 2017-01-30 Richard Biener <rguenther@suse.de>
20691 PR tree-optimization/79276
20692 * tree-vrp.c (process_assert_insertions): Properly adjust common
20693 when removing a duplicate.
20695 * gcc.dg/torture/pr79276.c: New testcase.
20697 2017-01-30 Richard Biener <rguenther@suse.de>
20699 PR tree-optimization/79256
20700 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
20701 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
20703 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
20705 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
20708 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
20709 ("*r<noxa>sbg_<mode>_sll_bitmask")
20710 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
20711 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
20712 Use contiguous_bitmask_nowrap_operand.
20714 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20717 * config/rs6000/altivec.h (vec_xl): Revise #define.
20718 (vec_xst): Likewise.
20720 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
20722 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
20724 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
20726 PR rtl-optimization/79194
20727 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
20728 traps before call to bypass_conditional_jumps.
20730 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
20732 PR tree-optimization/71374
20733 * lra-constraints.c (check_conflict_input_operands): New.
20734 (match_reload): Use it.
20736 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
20739 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
20740 account to calculate conflict_set.
20742 2017-01-27 Bin Cheng <bin.cheng@arm.com>
20744 PR rtl-optimization/78559
20745 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
20746 other_insn in combine.
20748 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20750 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
20751 uint16_type_node for BT_UINT16.
20753 2017-01-27 David Malcolm <dmalcolm@redhat.com>
20755 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
20756 "RTL Tests" to menu.
20757 (GIMPLE Tests): New node.
20758 (RTL Tests): New node.
20760 2017-01-27 Richard Biener <rguenther@suse.de>
20762 PR tree-optimization/79245
20763 * tree-loop-distribution.c (distribute_loop): Apply cost
20764 modeling also to detected patterns.
20766 2017-01-27 Richard Biener <rguenther@suse.de>
20768 PR tree-optimization/71433
20769 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
20770 (compare_assert_loc): New function.
20771 (process_assert_insertions): Sort and optimize assert locations
20772 to remove duplicates and push down identical assertions on
20773 edges to their destination block.
20775 2017-01-27 Richard Biener <rguenther@suse.de>
20777 PR tree-optimization/79244
20778 * tree-vrp.c (remove_range_assertions): Forcefully propagate
20779 out SSA names even if abnormal.
20781 2017-01-27 Jakub Jelinek <jakub@redhat.com>
20783 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
20784 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
20785 instead of MPFR_RNDN.
20787 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
20790 * arm.c (arm_option_override): Don't call build_target_option_node
20791 until after doing all option overrides.
20792 (arm_valid_target_attribute_tree): Likewise.
20794 2017-01-27 Martin Liska <mliska@suse.cz>
20796 * doc/invoke.texi (-fprofile-arcs): Document profiling support
20797 for {cd}tors and C++ {cd}tors.
20799 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
20801 * config/s390/s390.md ("*setmem_long_and")
20802 ("*setmem_long_and_31z"): Use zero_extend instead of and.
20804 2017-01-26 Martin Sebor <msebor@redhat.com>
20806 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
20809 2017-01-26 Martin Sebor <msebor@redhat.com>
20811 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
20812 HAVE_DFmode before using XFmode or DFmode.
20813 (parse_directive): Avoid using the z length modifier to avoid
20814 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
20816 PR middle-end/78703
20817 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
20818 to accept adjustment as an array.
20819 (get_int_range): New function.
20820 (struct directive): Make width and prec arrays.
20821 (directive::set_width, directive::set_precision): Call get_int_range.
20822 (format_integer, format_floating): Handle width and precision ranges.
20823 (format_string, parse_directive): Same.
20825 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20828 * dwarf2out.c (generate_skeleton_bottom_up): For children with
20829 comdat_type_p set, just clone them, but keep the children in the
20833 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
20834 which have direct callers with -fvar-tracking-assignments enabled
20836 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
20837 inside of type units.
20839 2017-01-26 Martin Sebor <msebor@redhat.com>
20841 PR middle-end/78703
20842 * gimple-ssa-sprintf.c (struct result_range): Add likely and
20844 (struct format_result): Replace number_chars, number_chars_min,
20845 and number_chars_max with a single member of struct result_range.
20847 (format_result::operator+=): Adjust.
20848 (struct fmtresult): Remove bounded. Handle likely and unlikely
20850 (fmtresult::adjust_for_width_or_precision): New function.
20851 (fmtresult:type_max_digits): New function.
20852 (bytes_remaining): Handle likely and unlikely counters.
20853 (min_bytes_remaining): Remove.
20854 (format_percent): Simplify.
20855 (format_integer, format_floating): Set likely and unlikely counters.
20856 (get_string_length, format_character, format_string): Same.
20857 (format_plain, should_warn_p): New function.
20858 (maybe_warn): Call should_warn_p. Update diagnostic messages
20859 and handle those for all directives, including plain strings.
20860 (format_directive): Handle likely and unlikely counters.
20861 Remove unnecessary quoting from diagnostics. Add an informational
20863 (add_bytes): Remove.
20864 (pass_sprintf_length::compute_format_length): Simplify.
20865 (try_substitute_return_value): Handle likely and unlikely counters.
20867 2017-01-26 Carl Love <cel@us.ibm.com>
20869 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
20870 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
20872 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
20875 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
20876 endianess for subregs into account.
20877 * lra-constraints.c (lra_constraints): Do risky transformations
20878 always on the first iteration.
20879 * lra-lives.c (check_pseudos_live_through_calls): Add arg
20880 last_call_used_reg_set.
20881 (process_bb_lives): Define and use last_call_used_reg_set.
20882 * lra.c (lra): Always continue after lra_constraints on the first
20885 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
20887 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
20889 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
20891 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20893 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
20894 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
20895 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
20896 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
20897 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
20898 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
20899 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
20900 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
20901 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
20903 2017-01-26 Marek Polacek <polacek@redhat.com>
20906 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
20907 for the third operand.
20909 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20911 PR middle-end/79236
20912 * omp-low.c (struct omp_context): Add simt_stmt field.
20913 (scan_omp_for): Return omp_context *.
20914 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
20915 context to the _simt_ SIMD stmt.
20916 (lower_omp_for): For combined SIMD with sibling _simt_
20917 SIMD, make sure to use the same decls in _looptemp_
20918 clauses as in the sibling.
20920 2017-01-26 David Sherwood <david.sherwood@arm.com>
20922 PR middle-end/79212
20923 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
20926 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20929 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
20930 emit fld b; fld a; if possible.
20932 * brig-builtins.def: Update copyright years.
20933 * config/arm/arm_acle_builtins.def: Update copyright years.
20935 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20938 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
20939 constraint instead of o for the stxsd instruction.
20941 2017-01-25 Carl Love <cel@us.ibm.com>
20943 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
20944 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
20946 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
20948 * doc/invoke.texi (C++ Dialect Options): Fix typo.
20950 2017-01-25 Richard Biener <rguenther@suse.de>
20952 PR tree-optimization/69264
20953 * target.def (vector_alignment_reachable): Improve documentation.
20954 * doc/tm.texi: Regenerate.
20955 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
20957 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
20958 earlier changes with respect to TYPE_USER_ALIGN.
20959 (vector_alignment_reachable_p): Likewise. Improve dumping.
20961 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20964 * config/arm/arm.md (xordi3): Force constant operand into a register
20967 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20969 * doc/invoke.texi (-fstore-merging): Correct default optimization
20970 levels at which it is enabled.
20971 (-O): Move -fstore-merging from list to...
20974 2017-01-25 Richard Biener <rguenther@suse.de>
20977 * omp-expand.c: Include debug.h.
20978 (expand_omp_taskreg): Make sure to generate early debug before
20979 outlining anything from a function.
20980 (expand_omp_target): Likewise.
20981 (grid_expand_target_grid_body): Likewise.
20983 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
20986 * asan.c (get_translation_unit_decl): New function.
20987 (asan_add_global): Extract modules file name from globals
20988 TRANSLATION_UNIT_DECL name.
20990 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
20993 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
20994 for long calls with APCS frame and VFP.
20996 2017-01-24 David Malcolm <dmalcolm@redhat.com>
20998 * cfg.c (original_copy_tables_initialized_p): New function.
20999 * cfg.h (original_copy_tables_initialized_p): New decl.
21000 * cfgrtl.c (relink_block_chain): Guard the call to
21001 free_original_copy_tables with a call to
21002 original_copy_tables_initialized_p.
21003 * cgraph.h (symtab_node::native_rtl_p): New decl.
21004 * cgraphunit.c (symtab_node::native_rtl_p): New function.
21005 (symtab_node::needed_p): Don't assert for early assembly output
21006 for __RTL functions.
21007 (cgraph_node::finalize_function): Set "force_output" for __RTL
21009 (cgraph_node::analyze): Bail out early for __RTL functions.
21010 (analyze_functions): Update assertion to support __RTL functions.
21011 (cgraph_node::expand): Bail out early for __RTL functions.
21012 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
21014 * function.h (struct function): Update comment for field
21016 * gimple-expr.c: Include "tree-pass.h".
21017 (gimple_has_body_p): Return false for __RTL functions.
21018 * Makefile.in (OBJS): Add run-rtl-passes.o.
21019 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
21021 (gcc::pass_manager::get_clean_slate): New accessor.
21022 * passes.c: Include "insn-addr.h".
21023 (should_skip_pass_p): Add logging. Update logic for running
21024 "expand" to be compatible with both __GIMPLE and __RTL. Guard
21025 property-provider override so it is only done for gimple passes.
21027 (skip_pass): New function.
21028 (execute_one_pass): Call skip_pass when skipping passes.
21029 * read-md.c (md_reader::read_char): Support filtering
21030 the input to a subset of line numbers.
21031 (md_reader::md_reader): Initialize fields
21032 m_first_line and m_last_line.
21033 (md_reader::read_file_fragment): New function.
21034 * read-md.h (md_reader::read_file_fragment): New decl.
21035 (md_reader::m_first_line): New field.
21036 (md_reader::m_last_line): New field.
21037 * read-rtl-function.c (function_reader::create_function): Only
21038 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
21039 curr_properties. Set DECL_INITIAL to a dummy block.
21040 (read_rtl_function_body_from_file_range): New function.
21041 * read-rtl-function.h (read_rtl_function_body_from_file_range):
21043 * run-rtl-passes.c: New file.
21044 * run-rtl-passes.h: New file.
21046 2017-01-24 Jeff Law <law@redhat.com>
21048 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
21051 2017-01-24 Bin Cheng <bin.cheng@arm.com>
21053 PR tree-optimization/79159
21054 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
21055 (record_nonwrapping_iv): Improve boundary using above function if no
21056 value range information.
21058 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
21059 Martin Jambor <mjambor@suse.cz>
21061 * brig-builtins.def: New file.
21062 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
21063 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
21064 (DEF_HSAIL_SAT_BUILTIN): Likewise.
21065 (DEF_HSAIL_INTR_BUILTIN): Likewise.
21066 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
21067 * builtin-types.def (BT_INT8): New.
21068 (BT_INT16): Likewise.
21069 (BT_UINT8): Likewise.
21070 (BT_UINT16): Likewise.
21071 (BT_FN_ULONG): Likewise.
21072 (BT_FN_UINT_INT): Likewise.
21073 (BT_FN_UINT_ULONG): Likewise.
21074 (BT_FN_UINT_LONG): Likewise.
21075 (BT_FN_UINT_PTR): Likewise.
21076 (BT_FN_ULONG_PTR): Likewise.
21077 (BT_FN_INT8_FLOAT): Likewise.
21078 (BT_FN_INT16_FLOAT): Likewise.
21079 (BT_FN_UINT32_FLOAT): Likewise.
21080 (BT_FN_UINT16_FLOAT): Likewise.
21081 (BT_FN_UINT8_FLOAT): Likewise.
21082 (BT_FN_UINT64_FLOAT): Likewise.
21083 (BT_FN_UINT16_UINT32): Likewise.
21084 (BT_FN_UINT32_UINT16): Likewise.
21085 (BT_FN_UINT16_UINT16_UINT16): Likewise.
21086 (BT_FN_INT_PTR_INT): Likewise.
21087 (BT_FN_UINT_PTR_UINT): Likewise.
21088 (BT_FN_LONG_PTR_LONG): Likewise.
21089 (BT_FN_ULONG_PTR_ULONG): Likewise.
21090 (BT_FN_VOID_UINT64_UINT64): Likewise.
21091 (BT_FN_UINT8_UINT8_UINT8): Likewise.
21092 (BT_FN_INT8_INT8_INT8): Likewise.
21093 (BT_FN_INT16_INT16_INT16): Likewise.
21094 (BT_FN_INT_INT_INT): Likewise.
21095 (BT_FN_UINT_FLOAT_UINT): Likewise.
21096 (BT_FN_FLOAT_UINT_UINT): Likewise.
21097 (BT_FN_ULONG_UINT_UINT): Likewise.
21098 (BT_FN_ULONG_UINT_PTR): Likewise.
21099 (BT_FN_ULONG_ULONG_ULONG): Likewise.
21100 (BT_FN_UINT_UINT_UINT): Likewise.
21101 (BT_FN_VOID_UINT_PTR): Likewise.
21102 (BT_FN_UINT_UINT_PTR: Likewise.
21103 (BT_FN_UINT32_UINT64_PTR): Likewise.
21104 (BT_FN_INT_INT_UINT_UINT): Likewise.
21105 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
21106 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
21107 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
21108 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
21109 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
21110 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
21111 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
21112 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
21113 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
21114 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
21115 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
21116 * doc/frontends.texi: List BRIG FE.
21117 * doc/install.texi (Testing): Add BRIG tesring requirements.
21118 * doc/invoke.texi (Overall Options): Mention BRIG.
21119 * doc/standards.texi (Standards): Doucment BRIG HSA version.
21121 2017-01-24 Richard Biener <rguenther@suse.de>
21123 PR translation/79208
21124 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
21126 2017-01-24 Martin Jambor <mjambor@suse.cz>
21129 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
21130 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
21131 and known_contexts.
21133 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
21135 PR middle-end/79123
21136 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
21137 casts from signed to unsigned really don't have a range.
21139 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
21141 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
21142 GMP_RNDx for compatiblity.
21144 2017-01-24 Martin Liska <mliska@suse.cz>
21147 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
21148 that would prevent us to call alloca with -1 as argument.
21150 2017-01-24 Jakub Jelinek <jakub@redhat.com>
21152 * dwarf2out.c (output_compilation_unit_header, output_file_names):
21153 Avoid -Wformat-security warning.
21155 2017-01-23 Andrew Pinski <apinski@cavium.com>
21157 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
21160 2017-01-23 Martin Sebor <msebor@redhat.com>
21162 PR middle-end/78703
21163 * gimple-ssa-sprintf.c (warn_level): New global.
21164 (format_integer): Use it here and throughout the rest of the file.
21165 Use the same switch to compute sign as base.
21166 (maybe_warn): New function.
21167 (format_directive): Factor out warnings into maybe_warn.
21168 Add debugging output. Use warn_level.
21169 (add_bytes): Use warn_level.
21170 (pass_sprintf_length::compute_format_length): Add debugging output.
21171 (try_substitute_return_value): Same.
21172 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
21174 PR middle-end/78703
21175 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
21176 (struct fmtresult, format_integer, format_floating): Adjust.
21177 (fmtresult::fmtresult): Set max correctly in two argument ctor.
21178 (get_string_length, format_string,format_directive): Same.
21179 (pass_sprintf_length::compute_format_length): Same.
21180 (try_substitute_return_value): Simplify slightly.
21182 PR middle-end/78703
21183 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
21184 (fmtresult::operator+=): Outlined.
21185 (struct fmtresult): Add ctors.
21186 (struct conversion_spec): Rename...
21187 (struct directive): ...to this. Add and remove data members.
21188 (directive::set_width, directive::set_precision): New functions.
21189 (format_percent): Use fmtresult ctor.
21190 (get_width_and_precision): Remove.
21191 (format_integer): Make naming changes. Avoid computing width and
21193 (format_floating): Same. Adjust indentation.
21194 (format_character, format_none): New functions.
21195 (format_string): Moved character handling to format_character.
21196 (format_directive): Remove arguments, change return type.
21197 (parse_directive): New function.
21198 (pass_sprintf_length::compute_format_length): Move directive
21199 parsing to parse_directive.
21201 2017-01-23 Jakub Jelinek <jakub@redhat.com>
21203 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
21204 (assign_assembler_name_if_needed): ... this.
21205 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
21206 (assign_assembler_name_if_needed): ... this.
21207 (free_lang_data_in_cgraph): Adjust callers.
21208 * cgraphunit.c (cgraph_node::analyze): Likewise.
21209 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
21212 2017-01-23 Richard Biener <rguenther@suse.de>
21214 PR tree-optimization/79088
21215 PR tree-optimization/79188
21216 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
21217 resetting loop bounds after last path deletion. Reset loop
21218 bounds of the target loop, make code match the comments.
21219 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
21220 Make sure loops need no fixups.
21222 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
21224 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
21225 exponent support with double type for first argument.
21226 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
21227 type returned by __builtin_vec_extract_sig,
21228 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
21229 functions from "vector int" to "vector unsigned int" or from
21230 "vector long long int" to "vector unsigned long long int".
21231 Changed type returned by __builtin_vec_extract_exp,
21232 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
21233 functions from "vector int" to "vector unsigned int" or from
21234 "vector long long int" to "vector unsigned long long int".
21235 Changed return type of __builtin_vec_test_data_class,
21236 __builtin_vec_test_data_class_sp, and
21237 __builtin_vec_test_data_class_dp from "vector int" to
21238 "vector bool int" or from "vector long long int" to "vector bool
21239 long long int" and changed second argument type from "unsigned
21240 int" to "int". Added new overloaded function forms "vector float
21241 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
21242 "vector float __builtin_vec_insert_exp_sp (vector float, vector
21243 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
21244 double, vector unsigned long long int)" and "vector double
21245 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
21246 long int)". Changed return type of
21247 __builtin_scalar_test_data_class and
21248 __builtin_scalar_test_data_class_sp and
21249 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
21250 int" and changed second argument from "unsigned int" to "int".
21251 Changed type returned by __builtin_scalar_test_neg,
21252 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
21253 from "int" to "bool int". Added new overloaded function form
21254 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
21255 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
21256 exponent double-precision with floating point first argument.
21257 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
21258 documentation of scalar_test_data_class, scalar_test_neg,
21259 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
21260 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
21261 vec_test_data_class built-in functions to reflect refinements in
21262 their type signatures.
21264 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
21266 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
21268 (aarch64_elf_asm_destructor): Likewise.
21270 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
21272 PR rtl-optimization/78634
21273 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
21274 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
21275 * ifcvt.c (noce_try_cmove): Add missing cost check.
21277 PR rtl-optimization/71724
21278 * combine.c (if_then_else_cond): Look for situations where it is
21279 beneficial to undo the work of one of the recursive calls.
21281 2017-01-23 Bin Cheng <bin.cheng@arm.com>
21283 PR tree-optimization/70754
21284 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
21285 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
21286 combined stmt before it if not NULL.
21287 (combine_chains): Process refs reversely and compute dominance point
21290 2017-01-23 Martin Liska <mliska@suse.cz>
21292 PR tree-optimization/79196
21293 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
21294 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
21296 (strlen_optimize_stmt): Call the renamed function.
21298 2017-01-23 Michael Matz <matz@suse.de>
21300 PR tree-optimization/78384
21301 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
21303 2017-01-23 Richard Biener <rguenther@suse.de>
21305 PR tree-optimization/79186
21306 * tree-vrp.c (register_new_assert_for): Make sure we've seen
21307 both incoming edges before moving an assert.
21309 2017-01-23 Martin Jambor <mjambor@suse.cz>
21311 * ipa-prop.c (load_from_param_1): Removed.
21312 (load_from_unmodified_param): Bits from load_from_param_1 put back
21314 (load_from_param): Removed.
21315 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
21316 with stmt. Reverted back to use of load_from_unmodified_param.
21318 2017-01-23 Martin Jambor <mjambor@suse.cz>
21321 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
21322 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
21323 field a pointer to garbage collected vector, mark lattices and
21324 ipcp_orig_node with GTY((skip)).
21325 (ipa_get_param_count): Adjust to descriptors being a pointer.
21326 (ipa_get_param): Likewise.
21327 (ipa_get_type): Likewise.
21328 (ipa_get_param_move_cost): Likewise.
21329 (ipa_set_param_used): Likewise.
21330 (ipa_get_controlled_uses): Likewise.
21331 (ipa_set_controlled_uses): Likewise.
21332 (ipa_is_param_used): Likewise.
21333 (ipa_node_params_t): Move into garbage collector. New methods insert
21335 (ipa_node_params_sum): Annotate wth GTY(()).
21336 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
21338 (ipa_load_from_parm_agg): Adjust declaration.
21339 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
21340 * ipa-profile.c (ipa_profile): Likewise.
21341 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
21342 (ipa_populate_param_decls): Make descriptors parameter garbage
21344 (ipa_dump_param): Adjust to descriptors being a pointer.
21345 (ipa_alloc_node_params): Likewise.
21346 (ipa_initialize_node_params): Likewise.
21347 (load_from_param_1): Make descriptors parameter garbage collected.
21348 (load_from_unmodified_param): Likewise.
21349 (load_from_param): Likewise.
21350 (ipa_load_from_parm_agg): Likewise.
21351 (ipa_node_params::~ipa_node_params): Removed.
21352 (ipa_free_all_node_params): Remove call to delete operator.
21353 (ipa_node_params_t::insert): New.
21354 (ipa_node_params_t::remove): Likewise.
21355 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
21356 copy known_csts and known_contexts vectors.
21357 (ipa_read_node_info): Adjust to descriptors being a pointer.
21358 (ipcp_modif_dom_walker): Make m_descriptors field garbage
21360 (ipcp_transform_function): Make descriptors variable garbage
21363 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
21365 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
21366 * config/i386/avx512dqintrin.h: Ditto.
21367 * config/i386/avx512fintrin.h: Ditto.
21368 * config/i386/i386.c: Handle new builtins.
21369 * config/i386/i386-builtin.def: Add new builtins.
21370 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
21371 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
21373 2017-01-23 Jakub Jelinek <jakub@redhat.com>
21374 Martin Liska <mliska@suse.cz>
21376 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
21377 * asan.c (asan_expand_poison_ifn): Support stores and use
21378 appropriate ASAN report function.
21379 * internal-fn.c (expand_ASAN_POISON_USE): New function.
21380 * internal-fn.def (ASAN_POISON_USE): Declare.
21381 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
21382 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
21383 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
21384 ASAN_POISON calls w/o LHS.
21385 * tree-ssa.c (execute_update_addresses_taken): Create clobber
21386 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
21387 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
21388 * gimplify.c (asan_poison_variables): Add attribute
21389 use_after_scope_memory to variables that really needs to live
21391 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
21392 having the attribute.
21394 2017-01-23 Martin Liska <mliska@suse.cz>
21396 * asan.c (create_asan_shadow_var): New function.
21397 (asan_expand_poison_ifn): Likewise.
21398 * asan.h (asan_expand_poison_ifn): New declaration.
21399 * internal-fn.c (expand_ASAN_POISON): Likewise.
21400 * internal-fn.def (ASAN_POISON): New builtin.
21401 * sanopt.c (pass_sanopt::execute): Expand
21402 asan_expand_poison_ifn.
21403 * tree-inline.c (copy_decl_for_dup_finish): Make function
21405 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
21406 * tree-ssa.c (is_asan_mark_p): New function.
21407 (execute_update_addresses_taken): Rewrite local variables
21408 (identified just by use-after-scope as addressable) into SSA.
21410 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21412 * doc/install.texi (Specific): opensource.apple.com uses https
21413 now. Remove trailing slash.
21415 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21417 * README.Portability: Remove note on an Irix compatibility issue.
21419 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
21421 * gcov.c (INCLUDE_ALGORITHM): Define.
21422 (INCLUDE_VECTOR): Define.
21423 No longer include <vector> and <algorithm> directly.
21425 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
21427 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
21429 * doc/invoke.texi (Code Gen Options): Ditto.
21431 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
21434 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
21436 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
21438 rtl-optimization/79125
21439 * cprop.c (local_cprop_pass): Handle cases where we make an
21440 unconditional trap.
21442 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
21446 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
21447 read from, for big endian.
21449 2017-01-20 Jiong Wang <jiong.wang@arm.com>
21451 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
21452 register pauth builtins for LP64 only.
21454 2017-01-20 Marek Polacek <polacek@redhat.com>
21457 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
21460 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21462 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
21464 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
21465 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
21466 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
21468 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21471 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
21472 in big-endian mode when they are not a single duplicated value.
21474 2017-01-20 Richard Biener <rguenther@suse.de>
21476 * BASE-VER: Bump to 7.0.1.
21478 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21480 * omp-low.c (omplow_simd_context): New struct. Use it...
21481 (lower_rec_simd_input_clauses): ...here and...
21482 (lower_rec_input_clauses): ...here to hold common data. Adjust all
21483 references to idx, lane, max_vf, is_simt.
21485 2017-01-20 Graham Markall <graham.markall@embecosm.com>
21487 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
21490 2017-01-20 Martin Jambor <mjambor@suse.cz>
21492 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
21493 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
21495 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
21496 (GTFILES): Rename hsa.c to hsa-common.c.
21497 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
21498 * hsa-dump.c: Likewise.
21499 * hsa-gen.c: Likewise.
21500 * hsa-regalloc.c: Likewise.
21501 * ipa-hsa.c: Likewise.
21502 * omp-expand.c: Likewise.
21503 * omp-low.c: Likewise.
21504 * toplev.c: Likewise.
21506 2017-01-20 Marek Polacek <polacek@redhat.com>
21509 * doc/invoke.texi: Document -Wduplicated-branches.
21510 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
21511 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
21512 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
21513 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
21514 return 0 only when not OEP_LEXICOGRAPHIC.
21515 (fold_build_cleanup_point_expr): Use the expression
21516 location when building CLEANUP_POINT_EXPR.
21517 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
21518 * tree.c (add_expr): Handle error_mark_node.
21520 2017-01-20 Martin Liska <mliska@suse.cz>
21523 * tree-profile.c (init_ic_make_global_vars): Do not call
21525 (gimple_init_gcov_profiler): Likewise.
21527 2017-01-20 Martin Liska <mliska@suse.cz>
21530 * cgraph.h (maybe_create_reference): Remove argument and
21532 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
21534 * ipa-cp.c (create_specialized_node): Likewise.
21535 * symtab.c (symtab_node::maybe_create_reference): Handle
21536 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
21538 2017-01-20 Martin Liska <mliska@suse.cz>
21540 * read-rtl-function.c (function_reader::create_function): Use
21541 build_decl instread of build_decl_stat.
21543 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
21545 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
21546 * config/i386/avx512dqintrin.h: Ditto.
21547 * config/i386/avx512fintrin.h: Ditto.
21548 * config/i386/i386-builtin-types.def: Add new types.
21549 * config/i386/i386.c: Handle new types.
21550 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
21551 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
21552 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
21553 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
21554 (__builtin_ia32_kshiftridi): New.
21555 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
21557 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
21561 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
21562 define to rs6000_init_stack_protect_guard.
21563 (rs6000_init_stack_protect_guard): New function.
21565 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
21566 Yunqiang Su <yunqiang.su@imgtec.com>
21568 * config.gcc (supported_defaults): Add madd4.
21569 (with_madd4): Add validation.
21570 (all_defaults): Add madd4.
21571 * config/mips/mips.opt (mmadd4): New option.
21572 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
21574 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
21576 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
21577 (ISA_HAS_FUSED_MADD4): Likewise.
21578 * doc/invoke.texi (-mmadd4): Document the new option.
21579 * doc/install.texi (--with-madd4): Document the new option.
21581 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21583 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
21584 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
21585 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
21586 (aarch64_init_pauth_hint_builtins): New.
21587 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
21588 (aarch64_expand_builtin): Expand new builtins.
21590 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21592 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
21593 * combine-stack-adj.c (no_unhandled_cfa): Handle
21594 REG_CFA_TOGGLE_RA_MANGLE.
21595 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
21596 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
21597 info for return address signing.
21598 (aarch64_expand_epilogue): Likewise.
21600 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21602 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
21603 * config/aarch64/aarch64-protos.h
21604 (aarch64_return_address_signing_enabled): New declaration.
21605 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
21607 (aarch64_expand_prologue): Sign return address before it's pushed onto
21609 (aarch64_expand_epilogue): Authenticate return address fetched from
21611 (aarch64_override_options): Sanity check for ILP32 and ISA level.
21612 (aarch64_attributes): New function attributes for "sign-return-address".
21613 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
21614 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
21615 ("*do_return"): Generate combined instructions according to key index.
21616 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
21617 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
21619 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
21620 * config/aarch64/aarch64.opt (msign-return-address=): New.
21621 * doc/extend.texi (AArch64 Function Attributes): Documents
21622 "sign-return-address=".
21623 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
21625 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
21627 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
21628 overall option summary.
21630 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21632 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
21633 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
21634 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
21635 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
21637 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
21639 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
21640 -mpower9-minmax by default for -mcpu=power9.
21641 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
21642 128-bit floating point.
21644 2017-01-20 Alan Modra <amodra@gmail.com>
21646 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
21647 optimizing for size.
21649 2017-01-20 Alan Modra <amodra@gmail.com>
21652 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
21653 for strcmp and strncmp from corresponding builtin decl.
21655 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
21657 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
21658 instead of i386/rtems-64.h.
21659 * config/i386/rtems-64.h: Remove.
21661 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
21665 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
21667 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
21669 2017-01-19 Tamar Christina <tamar.christina@arm.com>
21671 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
21672 Change int to HOST_WIDE_INT.
21673 * config/aarch64/aarch64-protos.h
21674 (aarch64_simd_gen_const_vector_dup): Likewise.
21675 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
21677 2017-01-19 David Malcolm <dmalcolm@redhat.com>
21679 * langhooks-def.h (lhd_type_for_size): New decl.
21680 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
21681 * langhooks.c (lhd_type_for_size): New function, taken from
21684 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
21686 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
21687 define_bypass for CR latency.
21688 (power9-cracked-alu): Update bypass latency and remove power9-branch.
21689 (power9-alu2): Add define_bypass for CR latency.
21691 (power9-mul): Update insn latency.
21692 (power9-mul-compare): Update insn latency, bypass latency and remove
21695 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21697 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
21699 * config/aarch64/aarch64.md
21700 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
21701 aarch64_nopcrelative_literal_loads.
21702 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
21704 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
21706 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
21707 TARGET_LOONGSON_3A.
21708 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
21710 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
21713 * config.gcc (supported_defaults): Add lxc1-sxc1.
21714 (with_lxc1_sxc1): Add validation.
21715 (all_defaults): Add lxc1-sxc1.
21716 * config/mips/mips.opt (mlxc1-sxc1): New option.
21717 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
21719 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
21720 __mips_no_lxc1_sxc1.
21721 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
21722 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
21723 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
21725 2017-01-19 Richard Biener <rguenther@suse.de>
21727 PR tree-optimization/72488
21728 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
21729 sure to restore SSA info.
21730 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
21732 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
21734 PR rtl-optimization/79121
21735 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
21736 of the inner type when shifting an extended value.
21738 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
21741 * symtab.c (symtab_node::equal_address_to): Fix comparing of
21742 interposable aliases.
21744 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
21747 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
21748 Use the evmergelohi instruction.
21749 (mov_si<mode>_e500_subreg4_2_le): Likewise.
21750 (mov_sitf_e500_subreg8_2_be): Likewise.
21751 (mov_sitf_e500_subreg12_2_le): Likewise.
21752 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
21753 (mov_si<mode>_e500_subreg4_2_be): Likewise.
21754 (mov_sitf_e500_subreg8_2_le): Likewise.
21755 (mov_sitf_e500_subreg12_2_be): Likewise.
21757 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21759 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
21760 attribute from vecsimple to vecperm.
21761 (altivec_vbpermq2): Likewise.
21763 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21766 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
21768 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21769 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
21770 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
21771 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
21772 case where N arg is SIZE_MAX.
21773 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
21774 (cmpstrsi): Add pattern.
21776 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21778 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21779 __builtin_vec_revb builtins.
21780 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
21781 built-in functions to support generation of the ISA 3.0 XXBR<x>
21782 vector byte reverse instructions.
21783 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
21784 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
21785 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
21786 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
21787 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
21788 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
21789 (P9V_BUILTIN_VEC_REVB): Likewise.
21790 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
21791 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
21792 (p9_xxbrq_v16qi): Likewise.
21793 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
21794 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
21795 (p9_xxbrh_v8hi): Likewise.
21796 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
21797 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
21798 vec_revb built-in functions.
21800 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
21802 PR rtl-optimization/78952
21803 * config/i386/i386.md (any_extract): New code iterator.
21804 (*insvqi_2): Use any_extract for source operand.
21805 (*insvqi_3): Use any_shiftrt for source operand.
21807 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
21809 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
21811 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
21813 2017-01-18 Matthias Klose <doko@ubuntu.com>
21815 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
21817 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21819 * config/rs6000/altivec.h (vec_bperm): Change #define.
21820 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
21821 (altivec_vbpermq2): New define_insn.
21822 (altivec_vbpermd): Likewise.
21823 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
21824 function interface.
21825 (VBPERMD): Likewise.
21826 (VBPERM): New polymorphic function interface.
21827 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
21828 Add entries for P9V_BUILTIN_VEC_VBPERM.
21829 * doc/extend.texi: Add interfaces for vec_bperm.
21831 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21833 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
21834 first letter of error messages.
21835 (s390_resolve_overloaded_builtin): Likewise.
21836 * config/s390/s390.c (s390_expand_builtin): Likewise.
21837 (s390_invalid_arg_for_unprototyped_fn): Likewise.
21838 (s390_valid_target_attribute_inner_p): Likewise.
21839 * config/s390/s390.md ("tabort"): Likewise.
21841 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
21843 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
21844 (ISA_AVOID_DIV_HILO): New macro.
21845 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
21846 (ISA_HAS_DDIV): Likewise.
21848 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21850 * doc/invoke.texi (fabi-version): Correct number of occurrences.
21852 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21854 * doc/invoke.texi (fabi-version): Spelling fix.
21856 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21859 * doc/invoke.texi (fabi-version): Mention mangling fix for
21862 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21865 * doc/invoke.texi (fabi-version): Document discriminator mangling.
21867 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
21870 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
21871 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
21873 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
21875 (stack_protect_test): Ditto.
21876 * config/rs6000/rs6000.opt (mstack-protector-guard=,
21877 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
21879 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
21880 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
21881 -mstack-protector-guard-offset=.
21882 (RS/6000 and PowerPC Options): Ditto.
21884 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
21886 * config/i386/i386.h (MASK_CLASS_P): New define.
21887 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
21888 there are no registers from different register sets also when
21889 mask registers are used. Update function comment.
21890 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
21891 to (*k/*r) and (*k/*km) alternatives.
21893 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
21895 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
21896 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
21897 (EH_RETURN_HANDLER_RTX): New define.
21898 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
21899 Force frame pointer in EH return functions.
21900 (aarch64_expand_epilogue): Add barrier for eh_return.
21901 (aarch64_final_eh_return_addr): Remove.
21902 (aarch64_eh_return_handler_rtx): New function.
21903 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
21905 (aarch64_eh_return_handler_rtx): New prototype.
21907 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21909 * config/rs6000/altivec.h (vec_rlmi): New #define.
21910 (vec_vrlnm): Likewise.
21911 (vec_rlnm): Likewise.
21912 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
21913 (UNSPEC_VRLNM): Likewise.
21914 (VIlong): New mode iterator.
21915 (altivec_vrl<VI_char>mi): New define_insn.
21916 (altivec_vrl<VI_char>nm): Likewise.
21917 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
21919 (VRLDNM): Likewise.
21920 (RLNM): New polymorphic function entry.
21921 (VRLWMI): New monomorphic function entry.
21922 (VRLDMI): Likewise.
21923 (RLMI): New polymorphic function entry.
21924 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
21925 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
21926 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
21929 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21932 * dwarf2out.c (field_byte_offset): Restore the
21933 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
21934 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
21935 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
21938 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
21941 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
21943 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21946 * dwarf2out.c (add_data_member_location_attribute): For constant
21947 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
21948 instead of DW_AT_data_member_location, DW_AT_bit_offset and
21949 DW_AT_byte_size attributes.
21951 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
21953 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
21954 after forcing to constant memory when the code model is medium.
21956 2017-01-17 Julia Koval <julia.koval@intel.com>
21959 * config/i386/avx512fintrin.h
21960 (_mm512_i32gather_ps): Change __addr type to void const*.
21961 (_mm512_mask_i32gather_ps): Ditto.
21962 (_mm512_i32gather_pd): Ditto.
21963 (_mm512_mask_i32gather_pd): Ditto.
21964 (_mm512_i64gather_ps): Ditto.
21965 (_mm512_mask_i64gather_ps): Ditto.
21966 (_mm512_i64gather_pd): Ditto.
21967 (_mm512_mask_i64gather_pd): Ditto.
21968 (_mm512_i32gather_epi32): Ditto.
21969 (_mm512_mask_i32gather_epi32): Ditto.
21970 (_mm512_i32gather_epi64): Ditto.
21971 (_mm512_mask_i32gather_epi64): Ditto.
21972 (_mm512_i64gather_epi32): Ditto.
21973 (_mm512_mask_i64gather_epi32): Ditto.
21974 (_mm512_i64gather_epi64): Ditto.
21975 (_mm512_mask_i64gather_epi64): Ditto.
21976 (_mm512_i32scatter_ps): Change __addr type to void*.
21977 (_mm512_mask_i32scatter_ps): Ditto.
21978 (_mm512_i32scatter_pd): Ditto.
21979 (_mm512_mask_i32scatter_pd): Ditto.
21980 (_mm512_i64scatter_ps): Ditto.
21981 (_mm512_mask_i64scatter_ps): Ditto.
21982 (_mm512_i64scatter_pd): Ditto.
21983 (_mm512_mask_i64scatter_pd): Ditto.
21984 (_mm512_i32scatter_epi32): Ditto.
21985 (_mm512_mask_i32scatter_epi32): Ditto.
21986 (_mm512_i32scatter_epi64): Ditto.
21987 (_mm512_mask_i32scatter_epi64): Ditto.
21988 (_mm512_i64scatter_epi32): Ditto.
21989 (_mm512_mask_i64scatter_epi32): Ditto.
21990 (_mm512_i64scatter_epi64): Ditto.
21991 (_mm512_mask_i64scatter_epi64): Ditto.
21992 * config/i386/avx512pfintrin.h
21993 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
21994 (_mm512_mask_prefetch_i32gather_ps): Ditto.
21995 (_mm512_mask_prefetch_i64gather_pd): Ditto.
21996 (_mm512_mask_prefetch_i64gather_ps): Ditto.
21997 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
21998 (_mm512_prefetch_i32scatter_ps): Ditto.
21999 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
22000 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
22001 (_mm512_prefetch_i64scatter_pd): Ditto.
22002 (_mm512_prefetch_i64scatter_ps): Ditto.
22003 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
22004 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
22005 * config/i386/avx512vlintrin.h
22006 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
22007 (_mm_mmask_i32gather_ps): Ditto.
22008 (_mm256_mmask_i32gather_pd): Ditto.
22009 (_mm_mmask_i32gather_pd): Ditto.
22010 (_mm256_mmask_i64gather_ps): Ditto.
22011 (_mm_mmask_i64gather_ps): Ditto.
22012 (_mm256_mmask_i64gather_pd): Ditto.
22013 (_mm_mmask_i64gather_pd): Ditto.
22014 (_mm256_mmask_i32gather_epi32): Ditto.
22015 (_mm_mmask_i32gather_epi32): Ditto.
22016 (_mm256_mmask_i32gather_epi64): Ditto.
22017 (_mm_mmask_i32gather_epi64): Ditto.
22018 (_mm256_mmask_i64gather_epi32): Ditto.
22019 (_mm_mmask_i64gather_epi32): Ditto.
22020 (_mm256_mmask_i64gather_epi64): Ditto.
22021 (_mm_mmask_i64gather_epi64): Ditto.
22022 (_mm256_i32scatter_ps): Change __addr type to void*.
22023 (_mm256_mask_i32scatter_ps): Ditto.
22024 (_mm_i32scatter_ps): Ditto.
22025 (_mm_mask_i32scatter_ps): Ditto.
22026 (_mm256_i32scatter_pd): Ditto.
22027 (_mm256_mask_i32scatter_pd): Ditto.
22028 (_mm_i32scatter_pd): Ditto.
22029 (_mm_mask_i32scatter_pd): Ditto.
22030 (_mm256_i64scatter_ps): Ditto.
22031 (_mm256_mask_i64scatter_ps): Ditto.
22032 (_mm_i64scatter_ps): Ditto.
22033 (_mm_mask_i64scatter_ps): Ditto.
22034 (_mm256_i64scatter_pd): Ditto.
22035 (_mm256_mask_i64scatter_pd): Ditto.
22036 (_mm_i64scatter_pd): Ditto.
22037 (_mm_mask_i64scatter_pd): Ditto.
22038 (_mm256_i32scatter_epi32): Ditto.
22039 (_mm256_mask_i32scatter_epi32): Ditto.
22040 (_mm_i32scatter_epi32): Ditto.
22041 (_mm_mask_i32scatter_epi32): Ditto.
22042 (_mm256_i32scatter_epi64): Ditto.
22043 (_mm256_mask_i32scatter_epi64): Ditto.
22044 (_mm_i32scatter_epi64): Ditto.
22045 (_mm_mask_i32scatter_epi64): Ditto.
22046 (_mm256_i64scatter_epi32): Ditto.
22047 (_mm256_mask_i64scatter_epi32): Ditto.
22048 (_mm_i64scatter_epi32): Ditto.
22049 (_mm_mask_i64scatter_epi32): Ditto.
22050 (_mm256_i64scatter_epi64): Ditto.
22051 (_mm256_mask_i64scatter_epi64): Ditto.
22052 (_mm_i64scatter_epi64): Ditto.
22053 (_mm_mask_i64scatter_epi64): Ditto.
22054 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
22055 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
22056 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
22057 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
22058 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
22059 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
22060 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
22061 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
22062 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
22063 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
22064 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
22065 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
22066 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
22067 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
22068 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
22069 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
22070 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
22071 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
22072 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
22073 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
22074 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
22075 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
22076 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
22077 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
22078 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
22079 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
22080 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
22081 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
22082 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
22083 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
22084 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
22085 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
22086 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
22087 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
22088 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
22089 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
22090 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
22091 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
22092 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
22093 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
22094 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
22095 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
22096 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
22097 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
22098 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
22099 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
22100 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
22101 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
22102 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
22103 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
22104 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
22105 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
22106 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
22107 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
22108 definitions accordingly.
22110 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
22111 Kuan-Lin Chen <kuanlinchentw@gmail.com>
22114 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
22117 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
22120 * ira-conflicts.c (ira_build_conflicts): Update total conflict
22121 hard regs for inner regno.
22123 2017-01-17 Martin Liska <mliska@suse.cz>
22126 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
22127 assumption and add comment.
22129 2017-01-17 Nathan Sidwell <nathan@acm.org>
22131 * ipa-visibility.c (localize_node): New function, broken out of ...
22132 (function_and_variable_visibility): ... here. Call it.
22134 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
22136 PR middle-end/77445
22137 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
22138 correctly set frequency of oudgoing edge.
22139 (duplicate_thread_path): Fix profile updating.
22141 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22144 * configure.ac: Add GCC_BASE_VER.
22145 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
22146 version from BASE-VER file.
22147 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
22148 (gcc.o): Depend on $(BASEVER).
22149 * common.opt (dumpfullversion): New option.
22150 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
22151 * doc/invoke.texi: Document -dumpfullversion.
22152 * doc/install.texi: Document --with-gcc-major-version-only.
22153 * configure: Regenerated.
22155 2017-01-17 Richard Biener <rguenther@suse.de>
22157 PR tree-optimization/71433
22158 * tree-vrp.c (register_new_assert_for): Merge same asserts
22159 on all incoming edges.
22160 (process_assert_insertions_for): Handle insertions at the
22163 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
22165 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
22166 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
22168 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
22171 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
22174 2017-01-17 Alan Modra <amodra@gmail.com>
22177 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
22178 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
22179 symbolic stack limit when pic.
22181 2017-01-16 Martin Sebor <msebor@redhat.com>
22183 PR tree-optimization/78608
22184 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
22186 2017-01-16 Jeff Law <law@redhat.com>
22189 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
22190 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
22191 for several include directories that may be relative to sysroot.
22192 * config/i386/x-mingw32 (gplus_includedir): Define.
22193 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
22194 (native_system_includedir): Likewise.
22195 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
22196 override if TARGET_SYSTEM_ROOT is defined.
22197 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
22199 PR tree-optimization/79090
22200 PR tree-optimization/33562
22201 PR tree-optimization/61912
22202 PR tree-optimization/77485
22203 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
22204 and computed trims into the dump file.
22206 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
22208 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
22210 2017-01-16 Jakub Jelinek <jakub@redhat.com>
22213 * gimplify.c (gimplify_init_constructor): If want_value and
22214 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
22218 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
22219 sequence. Formatting fixes.
22220 (doloop_optimize): Formatting fixes.
22223 * gcc.c (debug_level_greater_than_spec_func): New function.
22224 (static_spec_functions): Add debug-level-gt spec function.
22225 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
22227 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22228 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22229 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
22230 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
22231 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
22232 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
22234 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
22236 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
22237 QImode fixups to general and mask registers only.
22239 2017-01-16 Carl Love <cel@us.ibm.com>
22241 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
22242 for built-in functions
22243 vector signed char vec_nabs (vector signed char)
22244 vector signed short vec_nabs (vector signed short)
22245 vector signed int vec_nabs (vector signed int)
22246 vector signed long long vec_nabs (vector signed long long)
22247 vector float vec_nabs (vector float)
22248 vector double vec_nabs (vector double)
22249 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
22251 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
22252 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
22253 * doc/extend.texi: Update the documentation file for the new built-in
22256 2017-01-16 Martin Sebor <msebor@redhat.com>
22258 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
22261 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22263 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
22264 UNSPEC_VSX__XXSPLTD to require special splat handling.
22266 2017-01-16 David Malcolm <dmalcolm@redhat.com>
22269 * system.h: Poison strndup.
22271 2017-01-16 Alan Modra <amodra@gmail.com>
22274 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
22277 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
22279 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
22281 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
22283 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
22284 call recog here. Assert that INSN_CODE (insn) is non-negative.
22286 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22289 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
22291 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
22292 in the currently scheduled RTL fragment.
22294 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22296 PR rtl-optimization/78751
22297 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
22300 2017-01-14 Jeff Law <law@redhat.com>
22302 PR tree-optimization/79090
22303 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
22304 variable length stores.
22305 (compute_trims): Delete dead assignment to *trim_tail.
22306 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
22309 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
22311 PR rtl-optimization/78626
22312 PR rtl-optimization/78727
22313 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
22314 of a block, and split such blocks after everything else is finished.
22316 2017-01-14 Alan Modra <amodra@gmail.com>
22319 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
22320 target legitimate_combined_insn.
22321 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
22322 (rs6000_legitimate_combined_insn): New function.
22323 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
22325 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
22326 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
22327 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
22329 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
22331 * doc/frontends.texi (G++ and GCC): Remove references to Java.
22333 2017-01-13 Jeff Law <law@redhat.com>
22335 PR tree-optimization/33562
22336 PR tree-optimization/61912
22337 PR tree-optimization/77485
22338 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
22340 (delete_dead_assignment): Likewise.
22341 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
22342 statement to delete_dead_call and delete_dead_assignment.
22344 2017-01-13 David Malcolm <dmalcolm@redhat.com>
22347 * substring-locations.c (format_warning_va): Strengthen case 1 so
22348 that both endpoints of the substring must be within the format
22349 range for just the substring to be printed.
22351 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
22353 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
22354 * config/i386/i386.c (ix86_target_string): Add missing options
22355 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
22356 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
22357 flags_other and ix86_target_other to flags2_other. Display unknown
22359 (ix86_valid_target_attribute_inner_p): Add missing options and
22360 reorder options by implied ISAs, as in ix86_target_string.
22362 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22364 * hash-table.h (hash_table::too_empty_p): New function.
22365 (hash_table::expand): Use it.
22366 (hash_table::traverse): Likewise.
22367 (hash_table::empty_slot): Use sizeof (value_type) instead of
22368 sizeof (PTR) to convert bytes to elements. Shrink the table
22369 if the current size is excessive for the current number of
22372 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22374 * ira-costs.c (record_reg_classes): Break from the inner loop
22375 early once alt_fail is known to be true. Update outer loop
22376 handling accordingly.
22378 2017-01-13 Jeff Law <law@redhat.com>
22380 * tree-ssa-dse.c (decrement_count): New function.
22381 (increment_start_addr, maybe_trim_memstar_call): Likewise.
22382 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
22383 when we know the partially dead statement is a mem* function.
22385 PR tree-optimization/61912
22386 PR tree-optimization/77485
22387 * tree-ssa-dse.c: Include expr.h.
22388 (maybe_trim_constructor_store): New function.
22389 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
22391 PR tree-optimization/33562
22392 PR tree-optimization/61912
22393 PR tree-optimization/77485
22394 * doc/invoke.texi: Document new dse-max-object-size param.
22395 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
22396 * tree-ssa-dse.c: Include params.h.
22397 (dse_store_status): New enum.
22398 (initialize_ao_ref_for_dse): New, partially extracted from
22400 (valid_ao_ref_for_dse, normalize_ref): New.
22401 (setup_live_bytes_from_ref, compute_trims): Likewise.
22402 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
22403 (maybe_trim_partially_dead_store): Likewise.
22404 (maybe_trim_complex_store): Likewise.
22405 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
22406 Track what bytes live from the original store. Return tri-state
22407 for dead, partially dead or live.
22408 (dse_dom_walker): Add constructor, destructor and new private members.
22409 (delete_dead_call, delete_dead_assignment): New extracted from
22411 (dse_optimize_stmt): Make a member of dse_dom_walker.
22412 Use initialize_ao_ref_for_dse.
22414 PR tree-optimization/33562
22415 PR tree-optimization/61912
22416 PR tree-optimization/77485
22417 * sbitmap.h (bitmap_count_bits): Prototype.
22418 (bitmap_clear_range, bitmap_set_range): Likewise.
22419 * sbitmap.c (bitmap_clear_range): New function.
22420 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
22422 2017-01-13 Martin Liska <mliska@suse.cz>
22425 * function.c (set_cfun): Add new argument force.
22426 * function.h (set_cfun): Likewise.
22427 * ipa-inline-transform.c (inline_call): Use the function when
22428 strict alising from is dropped for function we inline to.
22430 2017-01-13 Richard Biener <rguenther@suse.de>
22432 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
22433 for dumping GIMPLE INTEGER_CSTs.
22435 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22437 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
22438 to 201112L since C++17.
22440 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22443 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
22444 if -fsanitize=kernel-address is present.
22446 2017-01-13 Richard Biener <rguenther@suse.de>
22448 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
22449 as _Literal ( type ) number in case usual suffixes do not
22450 preserve all information.
22452 2017-01-13 Richard Biener <rguenther@suse.de>
22454 PR tree-optimization/77283
22455 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
22456 and ssa-iterators.h.
22457 (is_feasible_trace): Implement a cost model based on joiner
22460 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22463 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
22464 char or short to __float128/_Float128 directly.
22466 2017-01-12 Martin Sebor <msebor@redhat.com>
22468 to -Wformat-overflow.
22469 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
22470 (min_bytes_remaining): Same.
22471 (get_string_length): Same.
22472 (format_string): Same.
22473 (format_directive): Same.
22475 (pass_sprintf_length::handle_gimple_call): Same.
22477 2017-01-12 Jakub Jelinek <jakub@redhat.com>
22479 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
22480 info.nowrite calls with no lhs that can't throw. Return bool
22481 whether gsi_remove has been called or not.
22482 (pass_sprintf_length::handle_gimple_call): Return bool whether
22483 try_substitute_return_value called gsi_remove. Formatting fix.
22484 (pass_sprintf_length::execute): Don't use gsi_remove if
22485 handle_gimple_call returned true.
22488 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
22489 be removed due to side-effects, don't remove following barrier nor
22490 turn the successor edge into fallthru edge.
22492 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22495 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
22496 element-reversing loads and stores as not swappable.
22498 2017-01-12 Nathan Sidwell <nathan@acm.org>
22499 Nicolai Stange <nicstange@gmail.com>
22501 * combine.c (try_combine): Don't ignore result of overlap checking
22502 loop. Combine overlap & asm check into single loop.
22504 2017-01-12 Richard Biener <rguenther@suse.de>
22506 * tree-pretty-print.c (dump_generic_node): Provide -gimple
22507 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
22509 2017-01-12 Richard Biener <rguenther@suse.de>
22511 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
22512 and TS_TARGET_OPTION directly derive from TS_BASE.
22513 * tree-core.h (tree_optimization_option): Derive from tree_base.
22514 (tree_target_option): Likewise.
22516 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
22518 * config/i386/i386.c (memory_address_length): Increase len
22519 only when rip_relative_addr_p returns false.
22521 2017-01-11 Julia Koval <julia.koval@intel.com>
22523 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
22524 (OPTION_MASK_ISA_SGX_SET): New.
22525 (ix86_handle_option): Handle OPT_msgx.
22526 * config.gcc: Added sgxintrin.h.
22527 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
22528 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
22529 * config/i386/i386.c (ix86_target_string): Add -msgx.
22531 (ix86_option_override_internal): Handle new options.
22532 (ix86_valid_target_attribute_inner_p): Add sgx.
22533 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
22534 * config/i386/i386.opt: Add msgx.
22535 * config/i386/sgxintrin.h: New file.
22536 * config/i386/x86intrin.h: Add sgxintrin.h.
22538 2017-01-11 Jakub Jelinek <jakub@redhat.com>
22541 * fold-const.c (maybe_nonzero_address): Return 1 for function
22543 (tree_single_nonzero_warnv_p): Don't handle function local objects
22547 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
22550 2017-01-11 David Malcolm <dmalcolm@redhat.com>
22553 * opts.c: Include "spellcheck.h"
22554 (struct string_fragment): New struct.
22555 (struct edit_distance_traits<const string_fragment &>): New
22557 (get_closest_sanitizer_option): New function.
22558 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
22560 2017-01-11 Jakub Jelinek <jakub@redhat.com>
22562 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
22564 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
22565 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
22566 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
22567 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
22568 for initial die_offset if dwarf_split_debug_info.
22569 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
22570 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
22572 (output_skeleton_debug_sections): Formatting fix. Use
22573 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
22574 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
22576 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
22578 * config/arm/cortex-a53.md: Add bypasses for
22579 cortex_a53_r2f_cvt.
22580 (cortex_a53_r2f): Only use for transfers.
22581 (cortex_a53_f2r): Likewise.
22582 (cortex_a53_r2f_cvt): Add reservation for conversions.
22583 (cortex_a53_f2r_cvt): Likewise.
22585 2017-01-11 Tamar Christina <tamar.christina@arm.com>
22587 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
22588 to all inlined functions, change static to extern.
22590 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
22593 * config/arm/arm.c (legitimize_pic_address): Handle reference to
22595 (arm_assemble_integer): Likewise.
22597 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
22599 * config.gcc: Use new awk script to check CPU, FPU and architecture
22600 parameters for --with-... options.
22601 * config/arm/parsecpu.awk: New file
22602 * config/arm/arm-cpus.in: New file.
22603 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
22605 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
22607 * config/arm/t-arm: Update dependency rules.
22608 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
22609 of processing .def files.
22610 * config/arm/genopt.sh: Deleted.
22611 * config/arm/gentune.sh: Deleted.
22612 * config/arm/arm-cores.def: Deleted.
22613 * config/arm/arm-arches.def: Deleted.
22614 * config/arm/arm-fpus.def: Deleted.
22615 * config/arm/arm-tune.md: Regenerated.
22616 * config/arm/arm-tables.opt: Regenerated.
22617 * config/arm/arm-cpu.h: New generated file.
22618 * config/arm/arm-cpu-data.h: New generated file.
22619 * config/arm/arm-cpu-cdata.h: New generated file.
22621 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
22624 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
22626 (input_varpool_node): Unpack dynamically_initialized bit.
22628 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
22630 PR rtl-optimization/79032
22631 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
22632 the alignment of the adjusted memory reference against that of MODE,
22633 instead of the alignment of the original memory reference.
22635 2017-01-11 Martin Jambor <mjambor@suse.cz>
22637 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
22639 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
22640 decorated functions.
22642 2017-01-11 Richard Biener <rguenther@suse.de>
22644 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
22645 set range/nonnull info for PHI results. Do not set it on
22646 stmts marked for removal.
22648 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
22650 * expr.c (store_field): In the bitfield case, fetch the return value
22651 from the registers before applying a single big-endian adjustment.
22652 Always do a final load for a BLKmode value not larger than a word.
22654 2017-01-10 David Malcolm <dmalcolm@redhat.com>
22657 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
22658 that we correctly handle column numbers greater than
22659 LINE_MAP_MAX_COLUMN_NUMBER.
22661 2017-01-10 Martin Sebor <msebor@redhat.com>
22663 PR middle-end/78245
22664 * gimple-ssa-sprintf.c (get_destination_size): Call
22665 {init,fini}object_sizes.
22666 * tree-object-size.c (addr_object_size): Adjust.
22667 (pass_through_call): Adjust.
22668 (pass_object_sizes::execute): Adjust.
22669 * tree-object-size.h (fini_object_sizes): Declare.
22671 2017-01-10 Martin Sebor <msebor@redhat.com>
22673 PR tree-optimization/78775
22674 * builtins.c (get_size_range): Move...
22675 * calls.c: ...to here.
22676 (alloc_max_size): Accept zero argument.
22677 (operand_signed_p): Remove.
22678 (maybe_warn_alloc_args_overflow): Call get_size_range.
22679 * calls.h (get_size_range): Declare.
22681 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
22683 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
22684 from TI's devices.csv file as of September 2016.
22685 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
22687 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
22689 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
22690 * doc/invoke.texi: Likewise.
22691 * doc/md.texi: Likewise.
22692 * doc/objc.texi: Likewise.
22694 2017-01-10 Joshua Conner <joshconner@google.com>
22696 * config/arm/fuchsia-elf.h: New file.
22697 * config/fuchsia.h: New file.
22698 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
22699 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
22701 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
22703 2016-01-10 Richard Biener <rguenther@suse.de>
22705 PR tree-optimization/79034
22706 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
22707 Propagate out degenerate PHIs in the joiner.
22709 2017-01-10 Martin Liska <mliska@suse.cz>
22711 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
22712 (sort_congruence_classes_by_decl_uid): Likewise.
22713 (sort_congruence_class_groups_by_decl_uid): Likewise.
22714 (sem_item_optimizer::merge_classes): Sort class, groups in these
22715 classes and members in the groups by DECL_UID of declarations.
22716 This would make merge operations stable.
22718 2017-01-10 Martin Liska <mliska@suse.cz>
22720 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
22721 usage of m_classes_vec.
22722 (sem_item_optimizer::~sem_item_optimizer): Likewise.
22723 (sem_item_optimizer::get_group_by_hash): Likewise.
22724 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
22725 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
22726 (sem_item_optimizer::verify_classes): Likewise.
22727 (sem_item_optimizer::process_cong_reduction): Likewise.
22728 (sem_item_optimizer::dump_cong_classes): Likewise.
22729 (sem_item_optimizer::merge_classes): Likewise.
22730 * ipa-icf.h (congruence_class_hash): Rename from
22731 congruence_class_group_hash. Remove declaration of m_classes_vec.
22733 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
22735 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
22736 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
22737 * config.gcc: Add avx512vpopcntdqintrin.h.
22738 * config/i386/avx512vpopcntdqintrin.h: New.
22739 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
22740 * config/i386/i386-builtin-types.def: Add new types.
22741 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
22742 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
22743 __builtin_ia32_vpopcountq_v8di_mask): New.
22744 * config/i386/i386-c.c (ix86_target_macros_internal): Define
22745 __AVX512VPOPCNTDQ__.
22746 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
22747 (PTA_AVX512VPOPCNTDQ): Define.
22748 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
22749 TARGET_AVX512VPOPCNTDQ_P): Define.
22750 * config/i386/i386.opt: Add mavx512vpopcntdq.
22751 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
22752 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
22754 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22756 PR middle-end/77484
22757 * predict.def (PRED_CALL): Set to 67.
22759 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
22761 * expr.c (store_field): In the bitfield case, if the value comes from
22762 a function call and is of an aggregate type returned in registers, do
22763 not modify the field mode; extract the value in all cases if the mode
22764 is BLKmode and the size is not larger than a word.
22766 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
22769 * config/i386/cpuid.h: Fix undefined behavior.
22771 2017-01-04 Jeff Law <law@redhat.com>
22773 PR tree-optimization/79007
22774 PR tree-optimization/67955
22775 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
22776 conservative for pt.null when flag_non_call_exceptions is on.
22778 2017-01-09 Jakub Jelinek <jakub@redhat.com>
22780 PR translation/79019
22781 PR translation/79020
22782 * params.def (PARAM_INLINE_MIN_SPEEDUP,
22783 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
22784 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
22786 * config/avr/avr.opt (maccumulate-args): Likewise.
22787 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
22788 * common.opt (freport-bug): Likewise.
22789 * cif-code.def (CIF_FINAL_ERROR): Likewise.
22790 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
22791 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
22792 translatable string.
22793 * config/i386/i386.c (function_value_32): Likewise.
22794 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
22795 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
22797 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
22798 * common/config/msp430/msp430-common.c (msp430_handle_option):
22800 * symtab.c (symtab_node::verify_base): Likewise.
22801 * opts.c (set_debug_level): Likewise.
22802 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
22803 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
22804 missing whitespace to translatable strings.
22805 * config/avr/avr.md (bswapsi2): Fix typo in comment.
22806 * config/sh/superh.h: Likewise.
22807 * config/i386/xopintrin.h: Likewise.
22808 * config/i386/znver1.md: Likewise.
22809 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
22810 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
22811 * double-int.h (struct double_int): Likewise.
22812 * double-int.c (div_and_round_double): Likewise.
22813 * wide-int.cc: Likewise.
22814 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
22815 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
22816 * cfgcleanup.c (crossjumps_occured): Renamed to ...
22817 (crossjumps_occurred): ... this.
22818 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
22821 PR tree-optimization/78899
22822 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
22823 returning bool return struct loop *, NULL for failure and the new
22825 (versionable_outer_loop_p): Don't version outer loop if it has
22826 dont_vectorized bit set.
22827 (tree_if_conversion): When versioning outer loop, ensure
22828 tree_if_conversion is performed also on the inner loop of the
22829 non-vectorizable outer loop copy.
22830 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
22831 LOOP_VECTORIZED in inner loop of the scalar outer loop and
22832 prevent vectorization of it.
22833 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
22834 the outer loop vectorization of the non-scalar version is attempted
22835 before vectorization of the inner loop in scalar version. If
22836 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
22837 vectorization of its inner loop.
22838 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
22839 has 2 inner loops, rename also on edges from bb whose single pred
22840 is outer_loop->header. Fix typo in function comment.
22842 2017-01-09 Martin Sebor <msebor@redhat.com>
22845 * asan.c (asan_emit_stack_protection): Increase local buffer size
22846 to avoid snprintf truncation warning.
22848 2017-01-09 Andrew Pinski <apinski@cavium.com>
22850 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
22851 to reference thunderx2t99 for the tuning structure
22852 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
22854 (thunderx2t99_extra_costs): This.
22855 * config/aarch64/aarch64-tune.md: Regenerate.
22856 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
22857 (vulcan_addrcost_table): This.
22858 (vulcan_regmove_cost): Rename to ...
22859 (thunderx2t99_regmove_cost): This.
22860 (vulcan_vector_cost): Rename to ...
22861 (thunderx2t99_vector_cost): this.
22862 (vulcan_branch_cost): Rename to ...
22863 (thunderx2t99_branch_cost): This.
22864 (vulcan_tunings): Rename to ...
22865 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
22866 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
22868 2017-01-09 Martin Jambor <mjambor@suse.cz>
22872 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
22873 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
22874 (propagate_vr_accross_jump_function): Use the above function for all
22875 value range computations for pass-through jump functions and type
22876 converasion from explicit value range values.
22877 (ipcp_propagate_stage): Do not attempt to deduce types of formal
22878 parameters from TYPE_ARG_TYPES.
22879 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
22880 (ipa_write_node_info): Stream type of the actual argument.
22881 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
22883 2017-01-09 Martin Liska <mliska@suse.cz>
22886 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
22887 (lookup_compiler): Do not show error message with have_E.
22889 2017-01-09 Jakub Jelinek <jakub@redhat.com>
22891 PR tree-optimization/78938
22892 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
22893 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
22894 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
22895 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
22898 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22900 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
22903 2017-01-09 Richard Biener <rguenther@suse.de>
22905 PR tree-optimization/78997
22906 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
22907 name condition properly.
22909 2017-01-09 Richard Biener <rguenther@suse.de>
22912 * dwarf2out.c (is_cxx): New overload with context.
22913 (is_naming_typedef_decl): Use it.
22915 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22917 * invoke.texi (Option Summary): Correct spacing in option lists
22918 and add line breaks to fix over-long lines.
22920 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22922 PR middle-end/17660
22924 * extend.texi (Common Variable Attributes): Add xref to GCC
22925 Internals manual to explain mode attribute keywords.
22927 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22930 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
22931 and Preprocessor Options.
22932 (Options for Linking): Document -pthread here....
22933 (RS/6000 and PowerPC Options): ...not here.
22934 (Solaris 2 Options): ...or here.
22935 * doc/cppopts.texi: Document -pthread.
22937 2017-01-08 Martin Sebor <msebor@redhat.com>
22939 PR middle-end/77708
22940 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
22941 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
22942 New member functions.
22943 (format_directive): Used them.
22945 (pass_sprintf_length::handle_gimple_call): Same.
22946 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
22947 to avoid truncation for any argument.
22948 (extract_affine_mul): Same.
22949 * tree.c (get_file_function_name): Same.
22951 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22953 PR middle-end/77484
22954 * predict.def (PRED_INDIR_CALL): Set to 86.
22956 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22958 PR preprocessor/54124
22959 * doc/cppopts.texi: Reformat -d subtable to list the full name
22960 of the options. Add cross-reference to the docs for the general
22961 compiler -d options.
22962 * doc/invoke.texi (Developer Options): Add cross-reference to the
22963 preprocessor-specific -d option documentation.
22965 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22967 PR preprocessor/13498
22968 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
22969 redudant material, and reflect new command-line options.
22970 (System Headers): Likewise.
22972 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22974 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
22975 -isystem, and -idirafter. Copy-edit.
22976 * doc/cppopts.texi: Copy-edit. Remove contradiction about
22977 default for -ftrack-macro-expansion. Delete obsolete and
22978 badly-formatted implementation details about -fdebug-cpp output.
22979 * doc/cppwarnopts.texi: Copy-edit.
22981 2017-01-07 David Malcolm <dmalcolm@redhat.com>
22984 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
22985 that the transition from a max line width >= 1<<10 to narrower
22986 lines works correctly.
22988 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
22990 * doc/options.texi (PerFunction): New.
22991 * opt-functions.awk (switch_flags): Map both Optimization and
22992 PerFunction to CL_OPTIMIZATION.
22993 * opth-gen.awk: Test for PerFunction flag along with
22995 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
22996 it only when the latter is present. Skip those that don't in
22997 the hash function generator.
22998 * common.opt (fvar-tracking): Mark as PerFunction instead of
23000 (fvar-tracking-assignments): Likewise.
23001 (fvar-tracking-assignments-toggle): Likewise.
23002 (fvar-tracking-uninit): Likewise.
23004 2017-01-07 Jakub Jelinek <jakub@redhat.com>
23006 PR translation/79018
23007 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
23010 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
23013 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
23014 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
23015 TARGET_LONG_JUMP_TABLE_OFFSETS.
23016 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
23017 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
23018 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
23019 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
23020 * config/m68k/m68k.md (tablejump expander): Likewise.
23021 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
23022 TARGET_LONG_JUMP_TABLE_OFFSETS.
23023 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
23024 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
23026 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
23027 David Holsgrove <david.holsgrove@xilinx.com>
23029 * common/config/microblaze/microblaze-common.c
23030 (TARGET_EXCEPT_UNWIND_INFO): Remove.
23031 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
23033 * config/microblaze/microblaze.c (microblaze_must_save_register)
23034 (microblaze_expand_epilogue, microblaze_return_addr): Handle
23036 (microblaze_eh_return): New function.
23037 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
23038 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
23039 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
23040 * config/microblaze/microblaze.md (eh_return): New pattern.
23042 2017-01-06 Jakub Jelinek <jakub@redhat.com>
23044 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
23045 GCC_DIAGNOSTIC_STRINGIFY): Define.
23047 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
23049 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23051 * config/arm/arm.md (<mcrr>): New.
23053 * config/arm/arm.c (arm_arch5te): New.
23054 (arm_option_override): Set arm_arch5te.
23055 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
23057 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
23058 (arm_mcrr_qualifiers): ... this. New.
23059 (MRRC_QUALIFIERS): Define to...
23060 (arm_mrrc_qualifiers): ... this. New.
23061 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
23063 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
23064 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
23065 (MRRCI, mrrc, MRRC): New.
23066 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
23067 VUNSPEC_MRRC2): New.
23069 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23071 * config/arm/arm.md (<mcr>): New.
23073 * config/arm/arm.c (arm_coproc_builtin_available): Add
23074 support for mcr, mrc, mcr2 and mrc2.
23075 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
23076 (arm_mcr_qualifiers): ... this. New.
23077 (MRC_QUALIFIERS): Define to ...
23078 (arm_mrc_qualifiers): ... this. New.
23079 (MCR_QUALIFIERS): Define to ...
23080 (arm_mcr_qualifiers): ... this. New.
23081 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
23083 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
23084 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
23085 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
23086 VUNSPEC_MRC2): New.
23088 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23090 * config/arm/arm.md (*ldc): New.
23094 * config/arm/arm.c (arm_coproc_builtin_available): Add
23095 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
23096 (arm_coproc_ldc_stc_legitimate_address): New.
23097 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
23098 'qualifier_const_pointer'.
23099 (LDC_QUALIFIERS): Define to...
23100 (arm_ldc_qualifiers): ... this. New.
23101 (STC_QUALIFIERS): Define to...
23102 (arm_stc_qualifiers): ... this. New.
23103 * config/arm/arm-protos.h
23104 (arm_coproc_ldc_stc_legitimate_address): New.
23105 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
23106 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
23107 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
23108 stc2, stcl, stc2l): New.
23109 * config/arm/constraints.md (Uz): New.
23110 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
23111 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
23112 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
23113 VUNSPEC_STC2L): New.
23115 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23117 * config/arm/arm.md (<cdp>): New.
23118 * config/arm/arm.c (neon_const_bounds): Rename this ...
23119 (arm_const_bounds): ... this.
23120 (arm_coproc_builtin_available): New.
23121 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
23122 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
23123 (CDP_QUALIFIERS): Define to...
23124 (arm_cdp_qualifiers): ... this. New.
23126 (arm_expand_builtin_args): Add case for 6 arguments.
23127 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
23128 (arm_const_bounds): ... this.
23129 (arm_coproc_builtin_available): New.
23130 * config/arm/arm_acle.h (__arm_cdp): New.
23132 * config/arm/arm_acle_builtins.def (cdp): New.
23134 * config/arm/iterators.md (CDPI,CDP,cdp): New.
23135 * config/arm/neon.md: Rename all 'neon_const_bounds' to
23136 'arm_const_bounds'.
23137 * config/arm/types.md (coproc): New.
23138 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
23139 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
23140 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
23141 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
23143 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23145 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
23146 (UBINOP_QUALIFIERS): New.
23148 (acle_builtin_data): New. Change comment.
23149 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
23150 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
23151 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
23152 arm_acle_builtins.def.
23153 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
23154 (arm_init_acle_builtins): New.
23155 (CRC32_BUILTIN): Remove.
23156 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
23157 crc32cb, crc32ch and crc32cw.
23158 (arm_init_crc32_builtins): Remove.
23159 (arm_init_builtins): Use arm_init_acle_builtins rather
23160 than arm_init_crc32_builtins.
23161 (arm_expand_acle_builtin): New.
23162 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
23163 * config/arm/arm_acle_builtins.def: New.
23165 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23167 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
23168 (arm_builtin_datum): ... this.
23169 (arm_init_neon_builtin): Rename to ...
23170 (arm_init_builtin): ... this. Add a new parameters PREFIX
23171 and USE_SIG_IN_NAME.
23172 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
23173 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
23174 'arm_builtin_datum'.
23175 (arm_init_vfp_builtins): Likewise.
23176 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
23177 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
23178 (arm_expand_neon_args): Rename to ...
23179 (arm_expand_builtin_args): ... this. Rename builtin_arg
23180 enum values and differentiate between ARG_BUILTIN_MEMORY
23181 and ARG_BUILTIN_NEON_MEMORY.
23182 (arm_expand_neon_builtin_1): Rename to ...
23183 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
23184 values, arm_expand_builtin_args and add bool parameter NEON.
23185 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
23186 (arm_expand_vfp_builtin): Likewise.
23187 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
23189 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23191 PR middle-end/77484
23192 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
23193 * predict.c (tree_estimate_probability_bb): Reverse direction of
23194 polymorphic call predictor.
23196 2017-01-06 David Malcolm <dmalcolm@redhat.com>
23198 * passes.c (execute_one_pass): Split out pass-skipping logic into...
23199 (determine_pass_name_match): ...this new function and...
23200 (should_skip_pass_p): ...this new function.
23202 2017-01-06 Nathan Sidwell <nathan@acm.org>
23204 * ipa-visibility.c (function_and_variable_visibility): Reformat
23205 comments and long lines. Remove extrneous if.
23206 * symtab.c (symtab_node::make_decl_local): Fix code format.
23207 (symtab_node::set_section_for_node): Fix comment typo.
23209 2017-01-06 Martin Liska <mliska@suse.cz>
23212 * lra-constraints.c: Rename invariant to lra_invariant.
23213 * predict.c (set_even_probabilities): Initialize e to NULL.
23215 2017-01-05 Martin Sebor <msebor@redhat.com>
23217 PR tree-optimization/78910
23218 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
23219 (format_integer): Correct off-by-one error in the handling
23220 of precision with negative numbers in signed conversions..
23222 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
23224 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
23226 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23228 PR tree-optimization/71016
23229 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
23230 factor_out_conditional_conversion. Formatting fix.
23231 (factor_out_conditional_conversion): Add cond_stmt argument.
23232 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
23233 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
23236 2017-01-05 David Malcolm <dmalcolm@redhat.com>
23238 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
23239 read-rtl-function.o, and selftest-rtl.o.
23240 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
23241 (selftest::aarch64_test_loading_full_dump): New function.
23242 (selftest::aarch64_run_selftests): New function.
23243 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
23244 selftest::aarch64_run_selftests.
23245 * config/i386/i386.c
23246 (selftest::ix86_test_loading_dump_fragment_1): New function.
23247 (selftest::ix86_test_loading_call_insn): New function.
23248 (selftest::ix86_test_loading_full_dump): New function.
23249 (selftest::ix86_test_loading_unspec): New function.
23250 (selftest::ix86_run_selftests): Call the new functions.
23251 * emit-rtl.c (maybe_set_max_label_num): New function.
23252 * emit-rtl.h (maybe_set_max_label_num): New decl.
23253 * function.c (instantiate_decls): Guard call to
23254 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
23255 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
23257 * gensupport.c (gen_reader::gen_reader): Pass "false"
23258 for new "compact" param of rtx_reader.
23259 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
23260 rather than an empty string for NULL strings.
23261 * read-md.c: Potentially include config.h rather than bconfig.h.
23262 Wrap include of errors.h with #ifdef GENERATOR_FILE.
23263 (have_error): New global, copied from errors.c.
23264 (md_reader::read_name): Rename to...
23265 (md_reader::read_name_1): ...this, adding "out_loc" param,
23266 and converting "missing name or number" to returning false, rather
23268 (md_reader::read_name): Reimplement in terms of read_name_1.
23269 (md_reader::read_name_or_nil): New function.
23270 (md_reader::read_string): Handle "(nil)" by returning NULL.
23271 (md_reader::md_reader): Add new param "compact".
23272 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
23273 (md_reader::read_file): New method.
23274 * read-md.h (md_reader::md_reader): Add new param "compact".
23275 (md_reader::read_file): New method.
23276 (md_reader::is_compact): New accessor.
23277 (md_reader::read_name): Convert return type from void to file_location.
23278 (md_reader::read_name_or_nil): New decl.
23279 (md_reader::read_name_1): New decl.
23280 (md_reader::m_compact): New field.
23281 (noop_reader::noop_reader): Pass "false" for new "compact" param
23283 (rtx_reader::rtx_reader): Add new "compact" param.
23284 (rtx_reader::read_rtx_operand): Make virtual and convert return
23285 type from void to rtx.
23286 (rtx_reader::read_until): New decl.
23287 (rtx_reader::handle_any_trailing_information): New virtual function.
23288 (rtx_reader::postprocess): New virtual function.
23289 (rtx_reader::finalize_string): New virtual function.
23290 (rtx_reader::m_in_call_function_usage): New field.
23291 (rtx_reader::m_reuse_rtx_by_id): New field.
23292 * read-rtl-function.c: New file.
23293 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
23294 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
23295 (selftest::verify_three_block_rtl_cfg): New decl.
23296 * read-rtl-function.h: New file.
23297 * read-rtl.c: Potentially include config.h rather than bconfig.h.
23298 For host, include function.h, memmodel.h, and emit-rtl.h.
23299 (one_time_initialization): New function.
23300 (struct compact_insn_name): New struct.
23301 (compact_insn_names): New array.
23302 (find_code): Handle insn codes in compact dumps.
23303 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
23304 (bind_subst_iter_and_attr): Likewise.
23305 (add_condition_to_string): Likewise.
23306 (add_condition_to_rtx): Likewise.
23307 (apply_attribute_uses): Likewise.
23308 (add_current_iterators): Likewise.
23309 (apply_iterators): Likewise.
23310 (initialize_iterators): Guard usage of apply_subst_iterator with
23311 #ifdef GENERATOR_FILE.
23312 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
23313 (md_reader::read_mapping): Likewise.
23314 (add_define_attr_for_define_subst): Likewise.
23315 (add_define_subst_attr): Likewise.
23316 (read_subst_mapping): Likewise.
23317 (check_code_iterator): Likewise.
23318 (rtx_reader::read_rtx): Likewise. Move one-time initialization
23320 (one_time_initialization): New function.
23321 (rtx_reader::read_until): New method.
23322 (read_flags): New function.
23323 (parse_reg_note_name): New function.
23324 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
23325 Handle reuse_rtx ids.
23326 Wrap iterator lookup within #ifdef GENERATOR_FILE.
23327 Add parsing support for RTL dumps, mirroring the special-cases in
23328 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
23329 values, and calling handle_any_trailing_information.
23330 (rtx_reader::read_rtx_operand): Convert return type from void
23331 to rtx, returning return_rtx. Handle case 'e'. Call
23332 finalize_string on XSTR and XTMPL fields.
23333 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
23334 "(nil)" values were omitted. Call the postprocess vfunc on the
23336 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
23337 class ctor. Initialize m_in_call_function_usage. Call
23338 one_time_initialization.
23339 * rtl-tests.c (selftest::test_uncond_jump): Call
23340 set_new_first_and_last_insn.
23341 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
23342 * selftest-rtl.c: New file.
23343 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
23344 (selftest::get_insn_by_uid): New decl.
23345 * selftest-run-tests.c (selftest::run_tests): Call
23346 read_rtl_function_c_tests.
23347 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
23348 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
23351 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
23353 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
23354 operands in a special way. Assert that pos+len <= mode precision.
23356 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23358 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
23359 3 argument Alias with unlimited for the negative form.
23360 (fno-vect-cost-model): Removed.
23362 2017-01-05 Martin Liska <mliska@suse.cz>
23364 * hsa-gen.c (gen_hsa_divmod): New function.
23365 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
23367 2017-01-05 Martin Liska <mliska@suse.cz>
23370 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
23373 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23375 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
23376 small constant length operands.
23378 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23380 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
23381 between loop iterations.
23383 2017-01-05 Martin Liska <mliska@suse.cz>
23386 * gimplify.c (gimplify_decl_expr): Compare to
23387 asan_poisoned_variables instread of checking flags.
23388 (gimplify_target_expr): Likewise.
23389 (gimplify_expr): Likewise.
23390 (gimplify_function_tree): Conditionally initialize
23391 asan_poisoned_variables.
23393 2017-01-04 Jeff Law <law@redhat.com>
23395 PR tree-optimizatin/78812
23396 * rtl.h (contains_mem_rtx_p): Prototype.
23397 * ifcvt.c (containts_mem_rtx_p): Move from here to...
23398 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
23399 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
23400 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
23401 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
23403 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23405 * input.c (assert_char_at_range): Default-initialize actual_range.
23407 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23409 * df-scan.c (df_ref_create_structure): Make regno unsigned,
23410 to match the caller.
23412 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23414 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
23415 insns after final jump in test to emit dummy move.
23417 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23419 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
23420 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
23422 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23424 * multiple_target.c (create_dispatcher_calls): Init e_next.
23425 * tree-ssa-loop-split.c (split_loop): Init border.
23426 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
23429 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
23434 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
23435 (altivec_register_operand): Do not return true if the operand
23436 contains a SUBREG mixing SImode and SFmode.
23437 (vsx_register_operand): Likewise.
23438 (vsx_reg_sfsubreg_ok): New predicate.
23439 (vfloat_operand): Do not return true if the operand contains a
23440 SUBREG mixing SImode and SFmode.
23441 (vint_operand): Likewise.
23442 (vlogical_operand): Likewise.
23443 (gpc_reg_operand): Likewise.
23444 (int_reg_operand): Likewise.
23445 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
23446 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
23447 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
23449 (rs6000_emit_move_si_sf_subreg): New helper function.
23450 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
23451 fixup SUBREGs involving SImode and SFmode.
23452 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
23453 numbers for the new peephole2 optimization.
23454 (peephole2 for SFmode unions): New peephole2 to optimize cases in
23455 the GLIBC math library that do AND/IOR/XOR operations on single
23456 precision floating point.
23457 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
23458 target macros to say whether we need to avoid SUBREGs mixing
23460 (TARGET_ALLOW_SF_SUBREG): Likewise.
23461 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
23462 (UNSPEC_SI_FROM_SF): Likewise.
23463 (iorxor): Change spacing.
23464 (and_ior_xor): New iterator for AND, IOR, and XOR.
23465 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
23466 (movdi_from_sf_zero_ext): Likewise.
23467 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
23468 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
23469 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
23470 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
23471 (fms<mode>4): Likewise.
23472 (fnma<mode>4): Likewise.
23473 (fnms<mode>4): Likewise.
23474 (nfma<mode>4): Likewise.
23475 (nfms<mode>4): Likewise.
23477 2017-01-04 Marek Polacek <polacek@redhat.com>
23480 * doc/invoke.texi: Document -Wpointer-compare.
23482 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23484 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
23487 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
23488 descriptions for -gdwarf-5 and emit them as uleb128 instead of
23491 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
23494 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
23495 documentation of the powerpc_popcntb_ok attribute.
23496 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23497 code to issue warning messages if a requested CPU configuration is
23498 not supported by the binary (assembler and loader) toolchain.
23499 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
23500 made to define a built-in function that has been disabled.
23501 (paired_init_builtins): Add assertion to prevent ICE if attempt is
23502 made to define a built-in function that has been disabled.
23503 (altivec_init_builtins): Add comment explaining why definition
23504 of the DST built-in functions is not preceded by an assertion
23505 check. Add assertions to prevent ICE if attempts are made to
23506 define an altivec predicate or an abs* built-in function that has
23508 (htm_init_builtins): Add comment explaining why definition of the
23509 htm built-in functions is not preceded by an assertion check.
23511 2017-01-04 Jeff Law <law@redhat.com>
23513 PR tree-optimizatin/67955
23514 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
23515 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
23516 the points-to solution does not include pt_null. Use DECL_PT_UID
23519 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
23521 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
23522 Use gen_int_mode instead of gen_lopwart for const_int operands.
23524 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23526 PR tree-optimization/71563
23527 * match.pd: Simplify X << Y into X if Y is known to be 0 or
23528 out of range value - has low bits known to be zero.
23530 2017-01-04 Alan Modra <amodra@gmail.com>
23532 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
23533 * configure: Regenerate.
23534 * config.in: Regenerate.
23536 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23539 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
23540 a substring of the message, but strcmp with the whole message. Ifdef
23541 ENABLE_NLS, translate the message first using dgettext.
23543 2017-01-03 Jeff Law <law@redhat.com>
23545 PR tree-optimizatin/78856
23546 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
23547 (mark_threaded_blocks): Remove code to truncate thread paths that
23548 cross multiple loop headers. Instead invalidate the cached loop
23549 iteration information and handle case of a thread path walking
23550 into an irreducible region.
23552 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
23555 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
23556 assertions. Add support for doing the signbit if the IEEE 128-bit
23557 floating point value is in a GPR.
23558 * config/rs6000/rs6000.md (Fsignbit): Delete.
23559 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
23560 Update the length attribute if the value is in a GPR.
23561 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
23562 the sign or zero extension instruction, since the value is always 0/1.
23563 (signbit<mode>2_dm2): Delete using <Fsignbit>.
23566 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
23567 extracting SImode to a GPR register so that we can generate a
23568 store, limit the vector to be in a traditional Altivec register
23569 for the vextuwrx instruction.
23571 2017-01-03 Ian Lance Taylor <iant@google.com>
23573 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
23575 2017-01-03 Martin Sebor <msebor@redhat.com>
23577 PR tree-optimization/78696
23578 * gimple-ssa-sprintf.c (format_floating): Correct handling of
23579 precision. Use MPFR for %f for greater fidelity. Correct handling
23581 (pass_sprintf_length::compute_format_length): Set width and precision
23582 specified by asrerisk to void_node for vararg functions.
23583 (try_substitute_return_value): Adjust dump output.
23585 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
23587 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
23589 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
23591 * doc/invoke.texi (SPARC options): Document -mlra as the default.
23592 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
23593 -mlra/-mno-lra was passed to the compiler.
23595 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
23597 PR rtl-optimization/65618
23598 * emit-rtl.c (try_split): Move initialization of "before" and
23599 "after" to just before the call to emit_insn_after_setloc.
23601 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
23603 * doc/md.texi (Standard Names): Remove reference to Java frontend.
23605 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
23607 * dwarf2out.c (gen_enumeration_type_die): When
23608 -gno-strict-dwarf, add a DW_AT_encoding attribute.
23610 2017-01-03 Jakub Jelinek <jakub@redhat.com>
23612 PR tree-optimization/78965
23613 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
23614 Change first argument from const call_info & to call_info &. For %n
23615 set info.nowrite to false.
23617 PR middle-end/78901
23618 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
23619 possibly throwing calls.
23621 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
23622 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
23623 and fns handling, rather than in a separate case SSA_NAME.
23625 2017-01-02 Jeff Law <law@redhat.com>
23627 * config/darwin-driver.c (darwin_driver_init): Const-correctness
23628 fixes for first_period and second_period variables.
23630 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
23633 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
23634 (*insvqi_1): New insn pattern.
23635 (*insvqi_1_mem_rex64): Ditto.
23636 (*insvqi_2): Ditto.
23637 (*insvqi_3): Rename from *insvqi.
23639 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
23641 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
23643 * doc/cfg.texi (Edges): Remove reference to Java.
23644 (Maintaining the CFG): Ditto.
23646 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23648 PR middle-end/77674
23649 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
23650 transparent aliases.
23652 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23654 PR middle-end/77484
23655 * predict.def (PRED_CALL): Update hitrate.
23656 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
23657 * predict.c (tree_estimate_probability_bb): Split CALL predictor
23658 into direct/indirect/polymorphic variants.
23660 2017-01-01 Jakub Jelinek <jakub@redhat.com>
23662 Update copyright years.
23664 * gcc.c (process_command): Update copyright notice dates.
23665 * gcov-dump.c (print_version): Ditto.
23666 * gcov.c (print_version): Ditto.
23667 * gcov-tool.c (print_version): Ditto.
23668 * gengtype.c (create_file): Ditto.
23669 * doc/cpp.texi: Bump @copying's copyright year.
23670 * doc/cppinternals.texi: Ditto.
23671 * doc/gcc.texi: Ditto.
23672 * doc/gccint.texi: Ditto.
23673 * doc/gcov.texi: Ditto.
23674 * doc/install.texi: Ditto.
23675 * doc/invoke.texi: Ditto.
23677 Copyright (C) 2017 Free Software Foundation, Inc.
23679 Copying and distribution of this file, with or without modification,
23680 are permitted in any medium without royalty provided the copyright
23681 notice and this notice are preserved.