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 * coretypes.h (scalar_mode): New class.
6 * machmode.h (scalar_mode): Likewise.
7 (scalar_mode::includes_p): New function.
8 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
9 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
12 * expmed.c (store_bit_field_1): Likewise.
13 (extract_bit_field_1): Likewise.
14 * expr.c (write_complex_part): Likewise.
15 (read_complex_part): Likewise.
16 (emit_move_complex_push): Likewise.
17 (expand_expr_real_2): Likewise.
18 * function.c (assign_parm_setup_reg): Likewise.
19 (assign_parms_unsplit_complex): Likewise.
20 * optabs.c (expand_binop): Likewise.
21 * rtlanal.c (subreg_get_info): Likewise.
22 * simplify-rtx.c (simplify_immed_subreg): Likewise.
23 * varasm.c (output_constant_pool_2): Likewise.
25 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
26 Alan Hayward <alan.hayward@arm.com>
27 David Sherwood <david.sherwood@arm.com>
29 * expmed.c (extract_high_half): Use scalar_int_mode and remove
31 (expmed_mult_highpart_optab): Likewise.
32 (expmed_mult_highpart): Likewise.
34 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
35 Alan Hayward <alan.hayward@arm.com>
36 David Sherwood <david.sherwood@arm.com>
38 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
39 instead of a machine_mode.
40 (builtin_memset_read_str): Likewise.
41 * builtins.c (c_readstr): Likewise.
42 (builtin_memcpy_read_str): Likewise.
43 (builtin_strncpy_read_str): Likewise.
44 (builtin_memset_read_str): Likewise.
45 (builtin_memset_gen_str): Likewise.
46 (expand_builtin_signbit): Use scalar_int_mode for local variables.
47 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
48 instead of a machine_mode.
49 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
51 (make_extraction): Likewise.
52 (try_widen_shift_mode): Take and return scalar_int_modes instead
54 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
55 a scalar_int_mode instead of a machine_mode.
56 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
57 (avr_addr_space_pointer_mode): Likewise.
58 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
59 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
60 (msp430_unwind_word_mode): Likewise.
61 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
62 (spu_addr_space_pointer_mode): Likewise.
63 (spu_addr_space_address_mode): Likewise.
64 (spu_libgcc_cmp_return_mode): Likewise.
65 (spu_libgcc_shift_count_mode): Likewise.
66 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
67 (rl78_addr_space_pointer_mode): Likewise.
68 (fl78_unwind_word_mode): Likewise.
69 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
71 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
72 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
73 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
74 (mips_valid_pointer_mode): Likewise.
75 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
76 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
77 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
79 (ft32_addr_space_address_mode): Likewise.
80 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
81 scalar_int_mode instead of a machine_mode.
82 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
84 (m32c_addr_space_address_mode): Likewise.
85 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
86 (rs6000_eh_return_filter_mode): Likewise.
87 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
88 (rs6000_eh_return_filter_mode): Likewise.
89 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
90 (s390_libgcc_shift_count_mode): Likewise.
91 (s390_unwind_word_mode): Likewise.
92 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
94 * target.def (mode_rep_extended): Likewise.
95 (valid_pointer_mode): Likewise.
96 (addr_space.valid_pointer_mode): Likewise.
97 (eh_return_filter_mode): Return a scalar_int_mode rather than
99 (libgcc_cmp_return_mode): Likewise.
100 (libgcc_shift_count_mode): Likewise.
101 (unwind_word_mode): Likewise.
102 (addr_space.pointer_mode): Likewise.
103 (addr_space.address_mode): Likewise.
104 * doc/tm.texi: Regenerate.
105 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
107 (do_jump): Use scalar_int_mode for local variables.
108 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
109 rather than a machine_mode.
110 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
111 (scompare_loc_descriptor_wide): Likewise.
112 (scompare_loc_descriptor_narrow): Likewise.
113 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
115 * except.c (sjlj_emit_dispatch_table): Likewise.
116 (expand_builtin_eh_copy_values): Likewise.
117 * explow.c (convert_memory_address_addr_space_1): Likewise.
118 Take a scalar_int_mode rather than a machine_mode.
119 (convert_memory_address_addr_space): Take a scalar_int_mode rather
121 (memory_address_addr_space): Use scalar_int_mode for local variables.
122 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
123 rather than a machine_mode.
124 * expmed.c (mask_rtx): Likewise.
125 (init_expmed_one_conv): Likewise.
126 (expand_mult_highpart_adjust): Likewise.
127 (extract_high_half): Likewise.
128 (expmed_mult_highpart_optab): Likewise.
129 (expmed_mult_highpart): Likewise.
130 (expand_smod_pow2): Likewise.
131 (expand_sdiv_pow2): Likewise.
132 (emit_store_flag_int): Likewise.
133 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
135 (extract_low_bits): Likewise.
136 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
138 * expr.c (pieces_addr::adjust): Likewise.
139 (can_store_by_pieces): Likewise.
140 (store_by_pieces): Likewise.
141 (clear_by_pieces_1): Likewise.
142 (expand_expr_addr_expr_1): Likewise.
143 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
144 (expand_expr_real_1): Likewise.
145 (try_casesi): Likewise.
146 * final.c (shorten_branches): Likewise.
147 * fold-const.c (fold_convert_const_int_from_fixed): Change the
148 type of "mode" to machine_mode.
149 * internal-fn.c (expand_arith_overflow_result_store): Take a
150 scalar_int_mode rather than a machine_mode.
151 (expand_mul_overflow): Use scalar_int_mode for local variables.
152 * loop-doloop.c (doloop_modify): Likewise.
153 (doloop_optimize): Likewise.
154 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
156 (expand_doubleword_shift_condmove): Likewise.
157 (expand_doubleword_shift): Likewise.
158 (expand_doubleword_clz): Likewise.
159 (expand_doubleword_popcount): Likewise.
160 (expand_doubleword_parity): Likewise.
161 (expand_absneg_bit): Use scalar_int_mode for local variables.
162 (prepare_float_lib_cmp): Likewise.
163 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
164 rather than a machine_mode.
165 (convert_memory_address_addr_space): Likewise.
166 (get_mode_bounds): Likewise.
167 (get_address_mode): Return a scalar_int_mode rather than a
169 * rtlanal.c (get_address_mode): Likewise.
170 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
172 * targhooks.c (default_mode_rep_extended): Likewise.
173 (default_valid_pointer_mode): Likewise.
174 (default_addr_space_valid_pointer_mode): Likewise.
175 (default_eh_return_filter_mode): Return a scalar_int_mode rather
177 (default_libgcc_cmp_return_mode): Likewise.
178 (default_libgcc_shift_count_mode): Likewise.
179 (default_unwind_word_mode): Likewise.
180 (default_addr_space_pointer_mode): Likewise.
181 (default_addr_space_address_mode): Likewise.
182 * targhooks.h (default_eh_return_filter_mode): Likewise.
183 (default_libgcc_cmp_return_mode): Likewise.
184 (default_libgcc_shift_count_mode): Likewise.
185 (default_unwind_word_mode): Likewise.
186 (default_addr_space_pointer_mode): Likewise.
187 (default_addr_space_address_mode): Likewise.
188 (default_mode_rep_extended): Take a scalar_int_mode rather than
190 (default_valid_pointer_mode): Likewise.
191 (default_addr_space_valid_pointer_mode): Likewise.
192 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
194 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
195 rather than a machine_mode.
196 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
198 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
199 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
202 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
203 Alan Hayward <alan.hayward@arm.com>
204 David Sherwood <david.sherwood@arm.com>
206 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
207 the mode argument to scalar_int_mode.
208 (do_jump_by_parts_zero_rtx): Likewise.
209 (do_jump_by_parts_equality_rtx): Likewise.
210 (do_jump_by_parts_greater): Take a mode argument.
211 (do_jump_by_parts_equality): Likewise.
212 (do_jump_1): Update calls accordingly.
214 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
215 Alan Hayward <alan.hayward@arm.com>
216 David Sherwood <david.sherwood@arm.com>
218 * is-a.h (safe_dyn_cast): New function.
219 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
220 (jump_table_for_label): Likewise.
221 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
222 instead of an rtx_insn *.
223 (shorten_branches): Use dyn_cast instead of LABEL_P and
224 JUMP_TABLE_DATA_P. Use jump_table_for_label and
225 rtx_jump_table_data::get_data_mode.
226 (final_scan_insn): Likewise.
228 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
229 Alan Hayward <alan.hayward@arm.com>
230 David Sherwood <david.sherwood@arm.com>
232 * combine.c (try_combine): Use is_a <scalar_int_mode> when
233 trying to combine a full-register integer set with a subreg
236 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
237 Alan Hayward <alan.hayward@arm.com>
238 David Sherwood <david.sherwood@arm.com>
240 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
241 that is always either address_mode or pointer_mode.
243 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
244 Alan Hayward <alan.hayward@arm.com>
245 David Sherwood <david.sherwood@arm.com>
247 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
248 when the two are known to be equal.
250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
251 Alan Hayward <alan.hayward@arm.com>
252 David Sherwood <david.sherwood@arm.com>
254 * simplify-rtx.c (simplify_const_unary_operation): Use
255 is_a <scalar_int_mode> instead of checking for a nonzero
256 precision. Forcibly convert op_mode to a scalar_int_mode
257 in that case. More clearly differentiate the operand and
258 result modes and use the former when deciding what the value
259 of a count-bits operation should be. Use is_int_mode instead
260 of checking for a MODE_INT. Remove redundant check for whether
261 this mode has a zero precision.
263 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
264 Alan Hayward <alan.hayward@arm.com>
265 David Sherwood <david.sherwood@arm.com>
267 * optabs.c (widen_leading): Change the type of the mode argument
268 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
269 (widen_bswap): Likewise.
270 (expand_parity): Likewise.
271 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
272 (expand_ffs): Likewise.
273 (epand_unop): Check for scalar integer modes before calling the
276 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
277 Alan Hayward <alan.hayward@arm.com>
278 David Sherwood <david.sherwood@arm.com>
280 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
282 (expand_expr_real_1): Update call accordingly.
284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
285 Alan Hayward <alan.hayward@arm.com>
286 David Sherwood <david.sherwood@arm.com>
288 * expmed.c (store_bit_field_using_insv): Add op0_mode and
289 value_mode arguments. Use scalar_int_mode internally.
290 (store_bit_field_1): Rename the new integer mode from imode
291 to op0_mode and use it instead of GET_MODE (op0). Update calls
292 to store_split_bit_field, store_bit_field_using_insv and
293 store_fixed_bit_field.
294 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
295 Use scalar_int_mode internally. Use a bit count rather than a mode
296 when calculating the largest bit size for get_best_mode.
297 Update calls to store_split_bit_field and store_fixed_bit_field_1.
298 (store_fixed_bit_field_1): Add mode and value_mode arguments.
299 Remove assertion that OP0 has a scalar integer mode.
300 (store_split_bit_field): Add op0_mode and value_mode arguments.
301 Update calls to extract_fixed_bit_field.
302 (extract_bit_field_using_extv): Add an op0_mode argument.
303 Use scalar_int_mode internally.
304 (extract_bit_field_1): Rename the new integer mode from imode to
305 op0_mode and use it instead of GET_MODE (op0). Update calls to
306 extract_split_bit_field, extract_bit_field_using_extv and
307 extract_fixed_bit_field.
308 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
309 to extract_split_bit_field and extract_fixed_bit_field_1.
310 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
311 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
313 (extract_split_bit_field): Add an op0_mode argument. Update call
314 to extract_fixed_bit_field.
316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
317 Alan Hayward <alan.hayward@arm.com>
318 David Sherwood <david.sherwood@arm.com>
320 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
321 * explow.c (hard_function_value): Likewise.
322 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
323 convert_to_mode call outside the loop.
324 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
325 for the mode iterator. Require the mode specified by max_pieces
327 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
329 (copy_blkmode_to_reg): Likewise.
330 (set_storage_via_setmem): Likewise.
331 * optabs.c (prepare_cmp_insn): Likewise.
332 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
333 * stor-layout.c (finish_bitfield_representative): Likewise.
335 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
336 Alan Hayward <alan.hayward@arm.com>
337 David Sherwood <david.sherwood@arm.com>
339 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
340 * expr.c (convert_move): Use them.
341 (convert_modes): Likewise.
342 (store_expr_with_bounds): Likewise.
344 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
345 Alan Hayward <alan.hayward@arm.com>
346 David Sherwood <david.sherwood@arm.com>
348 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
349 parameter for the mode of "x". Remove the "known_x", "known_mode"
350 and "known_ret" arguments. Change the type of the mode argument
352 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
353 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
354 (reg_num_sign_bit_copies_for_combine): Likewise.
355 * rtlanal.c (nonzero_bits1): Likewise.
356 (num_sign_bit_copies1): Likewise.
357 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
358 (reg_num_sign_bit_copies_general): Likewise.
359 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
360 (reg_nonzero_bits_general): Likewise.
362 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
363 Alan Hayward <alan.hayward@arm.com>
364 David Sherwood <david.sherwood@arm.com>
366 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
367 than in subroutines. Return 1 for non-integer modes.
368 (cached_num_sign_bit_copies): Change the type of the mode parameter
370 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
371 classes. Handle CONST_INT_P first and then check whether X also
372 has a scalar integer mode. Check the same thing for inner registers
373 of a SUBREG and for values that are being extended or truncated.
375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
376 Alan Hayward <alan.hayward@arm.com>
377 David Sherwood <david.sherwood@arm.com>
379 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
380 in subroutines. Return the mode mask for non-integer modes.
381 (cached_nonzero_bits): Change the type of the mode parameter
383 (nonzero_bits1): Likewise. Remove early exit for other mode
384 classes. Handle CONST_INT_P first and then check whether X
385 also has a scalar integer mode.
387 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
388 Alan Hayward <alan.hayward@arm.com>
389 David Sherwood <david.sherwood@arm.com>
391 * combine.c (extract_left_shift): Add a mode argument and update
393 (make_compound_operation_int): Change the type of the mode parameter
394 to scalar_int_mode and update the call to extract_left_shift.
396 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
397 Alan Hayward <alan.hayward@arm.com>
398 David Sherwood <david.sherwood@arm.com>
400 * combine.c (simplify_and_const_int): Change the type of the mode
401 parameter to scalar_int_mode.
402 (simplify_and_const_int_1): Likewise. Update recursive call.
404 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
405 Alan Hayward <alan.hayward@arm.com>
406 David Sherwood <david.sherwood@arm.com>
408 * combine.c (simplify_compare_const): Check that the mode is a
409 scalar_int_mode (rather than VOIDmode) before testing its
411 (simplify_comparison): Move COMPARISON_P handling out of the
412 loop and restrict the latter part of the loop to scalar_int_modes.
413 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
414 and when considering SUBREG_REGs. Use is_int_mode instead of
415 checking GET_MODE_CLASS against MODE_INT.
417 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
418 Alan Hayward <alan.hayward@arm.com>
419 David Sherwood <david.sherwood@arm.com>
421 * combine.c (try_widen_shift_mode): Move check for equal modes to...
422 (simplify_shift_const_1): ...here. Use scalar_int_mode for
423 shift_unit_mode and for modes involved in scalar shifts.
425 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
426 Alan Hayward <alan.hayward@arm.com>
427 David Sherwood <david.sherwood@arm.com>
429 * combine.c (force_int_to_mode): New function, split out from...
430 (force_to_mode): ...here. Keep xmode up-to-date and use it
431 instead of GET_MODE (x).
433 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
434 Alan Hayward <alan.hayward@arm.com>
435 David Sherwood <david.sherwood@arm.com>
437 * optabs-query.h (extraction_insn::struct_mode): Change type to
438 opt_scalar_int_mode and update comment.
439 (extraction_insn::field_mode): Change type to scalar_int_mode.
440 (extraction_insn::pos_mode): Likewise.
441 * combine.c (make_extraction): Update accordingly.
442 * optabs-query.c (get_traditional_extraction_insn): Likewise.
443 (get_optab_extraction_insn): Likewise.
444 * recog.c (simplify_while_replacing): Likewise.
445 * expmed.c (narrow_bit_field_mem): Change the type of the mode
446 parameter to opt_scalar_int_mode.
448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
449 Alan Hayward <alan.hayward@arm.com>
450 David Sherwood <david.sherwood@arm.com>
452 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
453 to a scalar_int_mode instead of a machine_mode.
454 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
455 (get_best_mode): Return a boolean and use a pointer argument to store
456 the selected mode. Replace the limit mode parameter with a bit limit.
457 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
458 for the values returned by bit_field_mode_iterator::next_mode.
459 (store_bit_field): Update call to get_best_mode.
460 (store_fixed_bit_field): Likewise.
461 (extract_fixed_bit_field): Likewise.
462 * expr.c (optimize_bitfield_assignment_op): Likewise.
463 * fold-const.c (optimize_bit_field_compare): Likewise.
464 (fold_truth_andor_1): Likewise.
465 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
466 Update for new type of m_mode.
467 (get_best_mode): As above.
469 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
470 Alan Hayward <alan.hayward@arm.com>
471 David Sherwood <david.sherwood@arm.com>
473 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
474 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
475 (store_bit_field): Check is_a <scalar_int_mode> before calling
476 strict_volatile_bitfield_p.
477 (extract_bit_field): Likewise.
479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
480 Alan Hayward <alan.hayward@arm.com>
481 David Sherwood <david.sherwood@arm.com>
483 * target.def (cstore_mode): Return a scalar_int_mode.
484 * doc/tm.texi: Regenerate.
485 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
486 * targhooks.h (default_cstore_mode): Likewise.
487 * targhooks.c (default_cstore_mode): Likewise, using a forced
489 * expmed.c (emit_cstore): Expect the target of the cstore to be
492 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
493 Alan Hayward <alan.hayward@arm.com>
494 David Sherwood <david.sherwood@arm.com>
496 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
497 * builtins.c (expand_builtin_signbit): Use it.
498 * cfgexpand.c (expand_debug_expr): Likewise.
499 * dojump.c (do_jump): Likewise.
500 (do_compare_and_jump): Likewise.
501 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
502 * expmed.c (make_tree): Likewise.
503 * expr.c (expand_expr_real_2): Likewise.
504 (expand_expr_real_1): Likewise.
505 (try_casesi): Likewise.
506 * fold-const-call.c (fold_const_call_ss): Likewise.
507 * fold-const.c (unextend): Likewise.
508 (extract_muldiv_1): Likewise.
509 (fold_single_bit_test): Likewise.
510 (native_encode_int): Likewise.
511 (native_encode_string): Likewise.
512 (native_interpret_int): Likewise.
513 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
514 * internal-fn.c (expand_addsub_overflow): Likewise.
515 (expand_neg_overflow): Likewise.
516 (expand_mul_overflow): Likewise.
517 (expand_arith_overflow): Likewise.
518 * match.pd: Likewise.
519 * stor-layout.c (layout_type): Likewise.
520 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
521 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
522 * tree-ssanames.c (get_range_info): Likewise.
523 * tree-switch-conversion.c (array_value_type) Likewise.
524 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
525 (vect_recog_divmod_pattern): Likewise.
526 (vect_recog_mixed_size_cond_pattern): Likewise.
527 * tree-vrp.c (extract_range_basic): Likewise.
528 (simplify_float_conversion_using_ranges): Likewise.
529 * tree.c (int_fits_type_p): Likewise.
530 * ubsan.c (instrument_bool_enum_load): Likewise.
531 * varasm.c (mergeable_string_section): Likewise.
532 (narrowing_initializer_constant_valid_p): Likewise.
533 (output_constant): Likewise.
535 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
536 Alan Hayward <alan.hayward@arm.com>
537 David Sherwood <david.sherwood@arm.com>
539 * machmode.h (NARROWEST_INT_MODE): New macro.
540 * expr.c (alignment_for_piecewise_move): Use it instead of
541 GET_CLASS_NARROWEST_MODE (MODE_INT).
542 (push_block): Likewise.
543 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
545 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
547 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
548 Alan Hayward <alan.hayward@arm.com>
549 David Sherwood <david.sherwood@arm.com>
551 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
553 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
554 * optabs.c (expand_unop): Likewise.
556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
557 Alan Hayward <alan.hayward@arm.com>
558 David Sherwood <david.sherwood@arm.com>
560 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
561 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
562 (popcount_loc_descriptor): Likewise.
563 (bswap_loc_descriptor): Likewise.
564 (rotate_loc_descriptor): Likewise.
565 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
566 calling the functions above.
568 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
569 Alan Hayward <alan.hayward@arm.com>
570 David Sherwood <david.sherwood@arm.com>
572 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
574 (try_combine): Likewise.
575 (simplify_if_then_else): Likewise.
576 * cse.c (cse_insn): Likewise.
577 * dwarf2out.c (mem_loc_descriptor): Likewise.
578 * emit-rtl.c (gen_lowpart_common): Likewise.
579 * simplify-rtx.c (simplify_truncation): Likewise.
580 (simplify_binary_operation_1): Likewise.
581 (simplify_const_relational_operation): Likewise.
582 (simplify_ternary_operation): Likewise.
583 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
585 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
586 Alan Hayward <alan.hayward@arm.com>
587 David Sherwood <david.sherwood@arm.com>
589 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
590 * reload.c (push_reload): Likewise.
591 (find_reloads): Likewise.
593 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
594 Alan Hayward <alan.hayward@arm.com>
595 David Sherwood <david.sherwood@arm.com>
597 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
598 (make_compound_operation_int): Likewise.
599 (change_zero_ext): Likewise.
600 * expr.c (convert_move): Likewise.
601 (convert_modes): Likewise.
602 * fwprop.c (forward_propagate_subreg): Likewise.
603 * loop-iv.c (get_biv_step_1): Likewise.
604 * optabs.c (widen_operand): Likewise.
605 * postreload.c (move2add_valid_value_p): Likewise.
606 * recog.c (simplify_while_replacing): Likewise.
607 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
608 (simplify_binary_operation_1): Likewise. Remove redundant
611 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
612 Alan Hayward <alan.hayward@arm.com>
613 David Sherwood <david.sherwood@arm.com>
615 * combine.c (combine_simplify_rtx): Add checks for
616 is_a <scalar_int_mode>.
617 (simplify_if_then_else): Likewise.
618 (make_field_assignment): Likewise.
619 (simplify_comparison): Likewise.
620 * ifcvt.c (noce_try_bitop): Likewise.
621 * loop-invariant.c (canonicalize_address_mult): Likewise.
622 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
624 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
625 Alan Hayward <alan.hayward@arm.com>
626 David Sherwood <david.sherwood@arm.com>
628 * simplify-rtx.c (simplify_binary_operation_1): Use
629 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
631 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
632 Alan Hayward <alan.hayward@arm.com>
633 David Sherwood <david.sherwood@arm.com>
635 * wide-int.h (int_traits<unsigned char>) New class.
636 (int_traits<unsigned short>) Likewise.
637 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
638 Use GET_MODE_UNIT_PRECISION and remove redundant test for
640 * combine.c (set_nonzero_bits_and_sign_copies): Use
641 is_a <scalar_int_mode>.
642 (find_split_point): Likewise.
643 (combine_simplify_rtx): Likewise.
644 (simplify_logical): Likewise.
645 (expand_compound_operation): Likewise.
646 (expand_field_assignment): Likewise.
647 (make_compound_operation): Likewise.
648 (extended_count): Likewise.
649 (change_zero_ext): Likewise.
650 (simplify_comparison): Likewise.
651 * dwarf2out.c (scompare_loc_descriptor): Likewise.
652 (ucompare_loc_descriptor): Likewise.
653 (minmax_loc_descriptor): Likewise.
654 (mem_loc_descriptor): Likewise.
655 (loc_descriptor): Likewise.
656 * expmed.c (init_expmed_one_mode): Likewise.
657 * lra-constraints.c (lra_constraint_offset): Likewise.
658 * optabs.c (prepare_libcall_arg): Likewise.
659 * postreload.c (move2add_note_store): Likewise.
660 * reload.c (operands_match_p): Likewise.
661 * rtl.h (load_extend_op): Likewise.
662 * rtlhooks.c (gen_lowpart_general): Likewise.
663 * simplify-rtx.c (simplify_truncation): Likewise.
664 (simplify_unary_operation_1): Likewise.
665 (simplify_binary_operation_1): Likewise.
666 (simplify_const_binary_operation): Likewise.
667 (simplify_const_relational_operation): Likewise.
668 (simplify_subreg): Likewise.
669 * stor-layout.c (bitwise_mode_for_mode): Likewise.
670 * var-tracking.c (adjust_mems): Likewise.
671 (prepare_call_arguments): Likewise.
673 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
674 Alan Hayward <alan.hayward@arm.com>
675 David Sherwood <david.sherwood@arm.com>
677 * machmode.h (is_int_mode): New fuction.
678 * combine.c (find_split_point): Use it.
679 (combine_simplify_rtx): Likewise.
680 (simplify_if_then_else): Likewise.
681 (simplify_set): Likewise.
682 (simplify_shift_const_1): Likewise.
683 (simplify_comparison): Likewise.
684 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
685 * cse.c (notreg_cost): Likewise.
686 (cse_insn): Likewise.
687 * cselib.c (cselib_lookup_1): Likewise.
688 * dojump.c (do_jump_1): Likewise.
689 (do_compare_rtx_and_jump): Likewise.
690 * dse.c (get_call_args): Likewise.
691 * dwarf2out.c (rtl_for_decl_init): Likewise.
692 (native_encode_initializer): Likewise.
693 * expmed.c (emit_store_flag_1): Likewise.
694 (emit_store_flag): Likewise.
695 * expr.c (convert_modes): Likewise.
696 (store_field): Likewise.
697 (expand_expr_real_1): Likewise.
698 * fold-const.c (fold_read_from_constant_string): Likewise.
699 * gimple-ssa-sprintf.c (get_format_string): Likewise.
700 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
701 * optabs.c (expand_binop): Likewise.
702 (expand_unop): Likewise.
703 (expand_abs_nojump): Likewise.
704 (expand_one_cmpl_abs_nojump): Likewise.
705 * simplify-rtx.c (mode_signbit_p): Likewise.
706 (val_signbit_p): Likewise.
707 (val_signbit_known_set_p): Likewise.
708 (val_signbit_known_clear_p): Likewise.
709 (simplify_relational_operation_1): Likewise.
710 * tree.c (vector_type_mode): Likewise.
712 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
713 Alan Hayward <alan.hayward@arm.com>
714 David Sherwood <david.sherwood@arm.com>
716 * machmode.h (smallest_mode_for_size): Fix formatting.
717 (smallest_int_mode_for_size): New function.
718 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
719 instead of smallest_mode_for_size.
720 * combine.c (make_extraction): Likewise.
721 * config/arc/arc.c (arc_expand_movmem): Likewise.
722 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
723 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
724 * config/s390/s390.c (s390_expand_insv): Likewise.
725 * config/sparc/sparc.c (assign_int_registers): Likewise.
726 * config/spu/spu.c (spu_function_value): Likewise.
727 (spu_function_arg): Likewise.
728 * coverage.c (get_gcov_type): Likewise.
729 (get_gcov_unsigned_t): Likewise.
730 * dse.c (find_shift_sequence): Likewise.
731 * expmed.c (store_bit_field_1): Likewise.
732 * expr.c (convert_move): Likewise.
733 (store_field): Likewise.
734 * internal-fn.c (expand_arith_overflow): Likewise.
735 * optabs-query.c (get_best_extraction_insn): Likewise.
736 * optabs.c (expand_twoval_binop_libfunc): Likewise.
737 * stor-layout.c (layout_type): Likewise.
738 (initialize_sizetypes): Likewise.
739 * targhooks.c (default_get_mask_mode): Likewise.
740 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
742 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
743 Alan Hayward <alan.hayward@arm.com>
744 David Sherwood <david.sherwood@arm.com>
746 * machmode.h (opt_mode::else_blk): New function.
747 (int_mode_for_mode): Declare.
748 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
749 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
751 * cfgexpand.c (expand_debug_expr): Likewise.
752 * combine.c (gen_lowpart_or_truncate): Likewise.
753 (gen_lowpart_for_combine): Likewise.
754 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
755 * config/avr/avr.c (avr_to_int_mode): Likewise.
756 (avr_out_plus_1): Likewise.
757 (avr_out_plus): Likewise.
758 (avr_out_round): Likewise.
759 * config/i386/i386.c (ix86_split_to_parts): Likewise.
760 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
761 (s390_expand_vcond): Likewise.
762 * config/spu/spu.c (spu_split_immediate): Likewise.
763 (spu_expand_mov): Likewise.
764 * dse.c (get_stored_val): Likewise.
765 * expmed.c (store_bit_field_1): Likewise.
766 (convert_extracted_bit_field): Use int_mode_for_mode instead of
768 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
769 (extract_low_bits): Likewise.
770 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
771 handling rather than repeating the check.
772 (emit_group_store): Likewise.
773 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
774 * optabs.c (expand_absneg_bit): Likewise.
775 (expand_copysign_absneg): Likewise.
776 (expand_copysign_bit): Likewise.
777 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
778 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
779 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
780 * var-tracking.c (prepare_call_arguments): Likewise.
781 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
782 int_mode_for_mode instead of mode_for_size.
783 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
786 Alan Hayward <alan.hayward@arm.com>
787 David Sherwood <david.sherwood@arm.com>
789 * machmode.h (int_mode_for_size): New function.
790 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
791 instead of mode_for_size.
792 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
794 * combine.c (expand_field_assignment): Use int_mode_for_size
795 instead of mode_for_size.
796 (make_extraction): Likewise.
797 (simplify_shift_const_1): Likewise.
798 (simplify_comparison): Likewise.
799 * dojump.c (do_jump): Likewise.
800 * dwarf2out.c (mem_loc_descriptor): Likewise.
801 * emit-rtl.c (init_derived_machine_modes): Likewise.
802 * expmed.c (flip_storage_order): Likewise.
803 (convert_extracted_bit_field): Likewise.
804 * expr.c (copy_blkmode_from_reg): Likewise.
805 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
806 * internal-fn.c (expand_mul_overflow): Likewise.
807 * lower-subreg.c (simple_move): Likewise.
808 * optabs-libfuncs.c (init_optabs): Likewise.
809 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
810 * tree.c (vector_type_mode): Likewise.
811 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
812 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
813 * tree-vect-generic.c (expand_vector_parallel): Likewise.
814 * tree-vect-stmts.c (vectorizable_load): Likewise.
815 (vectorizable_store): Likewise.
817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
818 Alan Hayward <alan.hayward@arm.com>
819 David Sherwood <david.sherwood@arm.com>
821 * coretypes.h (pod_mode): New type.
822 (scalar_int_mode_pod): New typedef.
823 * machmode.h (pod_mode): New class.
824 (int_n_data_t::m): Change type to scalar_int_mode_pod.
825 * genmodes.c (emit_mode_int_n): Update accordingly.
826 * lower-subreg.h (target_lower_subreg): Change type to
828 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
832 Alan Hayward <alan.hayward@arm.com>
833 David Sherwood <david.sherwood@arm.com>
835 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
836 machine_mode to scalar_int_mode.
837 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
838 (rs6000_option_override_internal): Remove cast to int.
839 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
840 machine_mode to scalar_int_mode.
841 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
842 (rs6000_option_override_internal): Remove cast to int.
843 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
844 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
846 * config/s390/s390.c (s390_expand_builtin): Likewise.
847 * coretypes.h (scalar_int_mode): New type.
848 (opt_scalar_int_mode): New typedef.
849 * machmode.h (scalar_int_mode): New class.
850 (scalar_int_mode::includes_p): New function.
851 (byte_mode): Change type to scalar_int_mode.
852 (word_mode): Likewise.
853 (ptr_mode): Likewise.
854 * emit-rtl.c (byte_mode): Likewise.
855 (word_mode): Likewise.
856 (ptr_mode): Likewise.
857 (init_derived_machine_modes): Update accordingly.
858 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
859 and MODE_PARTIAL_INT.
860 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
864 Alan Hayward <alan.hayward@arm.com>
865 David Sherwood <david.sherwood@arm.com>
867 * target.def (libgcc_floating_mode_supported_p): Take a
869 * doc/tm.texi: Regenerate.
870 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
872 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
873 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
876 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
877 Alan Hayward <alan.hayward@arm.com>
878 David Sherwood <david.sherwood@arm.com>
880 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
881 * doc/tm.texi: Regenerate.
882 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
883 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
884 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
885 * targhooks.h (default_floatn_mode): Likewise.
886 * targhooks.c (default_floatn_mode): Likewise.
887 * tree.c (build_common_tree_nodes): Update accordingly.
889 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
890 Alan Hayward <alan.hayward@arm.com>
891 David Sherwood <david.sherwood@arm.com>
893 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
894 (mode_iterator::iterate_p): Likewise.
895 (mode_iterator::get_wider): Likewise.
896 * expr.c (init_expr_target): Use opt_scalar_float_mode.
898 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
899 Alan Hayward <alan.hayward@arm.com>
900 David Sherwood <david.sherwood@arm.com>
902 * coretypes.h (opt_scalar_float_mode): New typedef.
903 * machmode.h (float_mode_for_size): New function.
904 * emit-rtl.c (double_mode): Delete.
905 (init_emit_once): Use float_mode_for_size.
906 * stor-layout.c (layout_type): Likewise.
907 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
909 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
910 Alan Hayward <alan.hayward@arm.com>
911 David Sherwood <david.sherwood@arm.com>
913 * output.h (assemble_real): Take a scalar_float_mode.
914 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
915 * config/arm/arm.md (consttable_4): Likewise.
916 (consttable_8): Likewise.
917 (consttable_16): Likewise.
918 * config/mips/mips.md (consttable_float): Likewise.
919 * config/s390/s390.c (s390_output_pool_entry): Likewise.
920 * varasm.c (assemble_real): Take a scalar_float_mode.
921 (output_constant_pool_2): Update accordingly.
922 (output_constant): Likewise.
924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
925 Alan Hayward <alan.hayward@arm.com>
926 David Sherwood <david.sherwood@arm.com>
928 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
929 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
930 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
931 (native_encode_real): Likewise.
932 (native_interpret_real): Likewise.
933 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
934 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
937 Alan Hayward <alan.hayward@arm.com>
938 David Sherwood <david.sherwood@arm.com>
940 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
941 <scalar_float_mode>. Simplify.
942 (gen_extend_conv_libfunc): Likewise.
944 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
945 Alan Hayward <alan.hayward@arm.com>
946 David Sherwood <david.sherwood@arm.com>
948 * coretypes.h (scalar_float_mode): New type.
949 * machmode.h (mode_traits::from_int): Use machine_mode if
950 USE_ENUM_MODES is defined.
951 (is_a): New function.
953 (dyn_cast): Likewise.
954 (scalar_float_mode): New class.
955 (scalar_float_mode::includes_p): New function.
956 (is_float_mode): Likewise.
957 * gdbhooks.py (MachineModePrinter): New class.
958 (build_pretty_printer): Use it for scalar_float_mode.
959 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
960 (format_helper::format_helper): Turn into a template.
961 * genmodes.c (get_mode_class): New function.
962 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
963 or machine_mode if none.
964 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
965 as_a <scalar_float_mode>.
966 * dwarf2out.c (mem_loc_descriptor): Likewise.
967 (insert_float): Likewise.
968 (add_const_value_attribute): Likewise.
969 * simplify-rtx.c (simplify_immed_subreg): Likewise.
970 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
971 (expand_unop): Update accordingly.
972 (expand_abs_nojump): Likewise.
973 (expand_copysign_absneg): Take a scalar_float_mode.
974 (expand_copysign_bit): Likewise.
975 (expand_copysign): Update accordingly.
977 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
978 Alan Hayward <alan.hayward@arm.com>
979 David Sherwood <david.sherwood@arm.com>
981 * coretypes.h (opt_mode): New class.
982 * machmode.h (opt_mode): Likewise.
983 (opt_mode::else_void): New function.
984 (opt_mode::require): Likewise.
985 (opt_mode::exists): Likewise.
986 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
987 (GET_MODE_2XWIDER_MODE): Likewise.
988 (mode_iterator::get_wider): Update accordingly.
989 (mode_iterator::get_2xwider): Likewise.
990 (mode_iterator::get_known_wider): Likewise, turning into a template.
991 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
992 forcing a wider mode to exist.
993 * config/cr16/cr16.h (LONG_REG_P): Likewise.
994 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
995 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
996 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
997 * lower-subreg.c (init_lower_subreg): Likewise.
998 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
999 on the final iteration.
1000 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
1001 a wider mode exists before asking for a move pattern.
1002 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
1003 forcing a wider mode to exist.
1004 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
1005 returning false if no such mode exists.
1006 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
1007 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
1008 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
1009 Avoid checking for a MODE_INT if we already know the mode is not a
1011 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
1012 forcing a wider mode to exist.
1013 (expmed_mult_highpart_optab): Likewise.
1014 (expmed_mult_highpart): Likewise.
1015 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
1017 * lto-streamer-in.c (lto_input_mode_table): Likewise.
1018 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
1019 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1020 * internal-fn.c (expand_mul_overflow): Update use of
1021 GET_MODE_2XWIDER_MODE.
1022 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1023 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
1024 GET_MODE_WIDER_MODE.
1025 (convert_plusminus_to_widen): Likewise.
1026 * tree-switch-conversion.c (array_value_type): Likewise.
1027 * var-tracking.c (emit_note_insn_var_location): Likewise.
1028 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1029 Return false inside rather than outside the loop if no wider mode
1031 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
1032 and GET_MODE_2XWIDER_MODE
1033 (can_compare_p): Use else_void.
1034 * gdbhooks.py (OptMachineModePrinter): New class.
1035 (build_pretty_printer): Use it for opt_mode.
1037 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1038 Alan Hayward <alan.hayward@arm.com>
1039 David Sherwood <david.sherwood@arm.com>
1041 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
1042 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
1044 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1045 Alan Hayward <alan.hayward@arm.com>
1046 David Sherwood <david.sherwood@arm.com>
1048 * machmode.h (mode_traits): New structure.
1049 (get_narrowest_mode): New function.
1050 (mode_iterator::start): Likewise.
1051 (mode_iterator::iterate_p): Likewise.
1052 (mode_iterator::get_wider): Likewise.
1053 (mode_iterator::get_known_wider): Likewise.
1054 (mode_iterator::get_2xwider): Likewise.
1055 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
1056 (FOR_EACH_MODE): Likewise.
1057 (FOR_EACH_MODE_FROM): Likewise.
1058 (FOR_EACH_MODE_UNTIL): Likewise.
1059 (FOR_EACH_WIDER_MODE): Likewise.
1060 (FOR_EACH_2XWIDER_MODE): Likewise.
1061 * builtins.c (expand_builtin_strlen): Use new mode iterators.
1062 * combine.c (simplify_comparison): Likewise
1063 * config/i386/i386.c (type_natural_mode): Likewise.
1064 * cse.c (cse_insn): Likewise.
1065 * dse.c (find_shift_sequence): Likewise.
1066 * emit-rtl.c (init_derived_machine_modes): Likewise.
1067 (init_emit_once): Likewise.
1068 * explow.c (hard_function_value): Likewise.
1069 * expmed.c (extract_fixed_bit_field_1): Likewise.
1070 (extract_bit_field_1): Likewise.
1071 (expand_divmod): Likewise.
1072 (emit_store_flag_1): Likewise.
1073 * expr.c (init_expr_target): Likewise.
1074 (convert_move): Likewise.
1075 (alignment_for_piecewise_move): Likewise.
1076 (widest_int_mode_for_size): Likewise.
1077 (emit_block_move_via_movmem): Likewise.
1078 (copy_blkmode_to_reg): Likewise.
1079 (set_storage_via_setmem): Likewise.
1080 (compress_float_constant): Likewise.
1081 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1082 * optabs-query.c (get_best_extraction_insn): Likewise.
1083 * optabs.c (expand_binop): Likewise.
1084 (expand_twoval_unop): Likewise.
1085 (expand_twoval_binop): Likewise.
1086 (widen_leading): Likewise.
1087 (widen_bswap): Likewise.
1088 (expand_parity): Likewise.
1089 (expand_unop): Likewise.
1090 (prepare_cmp_insn): Likewise.
1091 (prepare_float_lib_cmp): Likewise.
1092 (expand_float): Likewise.
1093 (expand_fix): Likewise.
1094 (expand_sfix_optab): Likewise.
1095 * postreload.c (move2add_use_add2_insn): Likewise.
1096 * reg-stack.c (reg_to_stack): Likewise.
1097 * reginfo.c (choose_hard_reg_mode): Likewise.
1098 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1099 * stor-layout.c (mode_for_size): Likewise.
1100 (smallest_mode_for_size): Likewise.
1101 (mode_for_vector): Likewise.
1102 (finish_bitfield_representative): Likewise.
1103 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
1104 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
1105 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1106 * var-tracking.c (prepare_call_arguments): Likewise.
1108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1109 Alan Hayward <alan.hayward@arm.com>
1110 David Sherwood <david.sherwood@arm.com>
1112 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
1113 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
1114 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
1115 * machmode.h (mode_size): Move earlier in file.
1116 (mode_precision): Likewise.
1117 (mode_inner): Likewise.
1118 (mode_nunits): Likewise.
1119 (mode_unit_size): Likewise.
1120 (unit_unit_precision): Likewise.
1121 (mode_wider): Likewise.
1122 (mode_2xwider): Likewise.
1123 (machine_mode): New class.
1124 (mode_to_bytes): New function.
1125 (mode_to_bits): Likewise.
1126 (mode_to_precision): Likewise.
1127 (mode_to_inner): Likewise.
1128 (mode_to_unit_size): Likewise.
1129 (mode_to_unit_precision): Likewise.
1130 (mode_to_nunits): Likewise.
1131 (GET_MODE_SIZE): Use mode_to_bytes.
1132 (GET_MODE_BITSIZE): Use mode_to_bits.
1133 (GET_MODE_PRECISION): Use mode_to_precision.
1134 (GET_MODE_INNER): Use mode_to_inner.
1135 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
1136 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
1137 (GET_MODE_NUNITS): Use mode_to_nunits.
1138 * system.h (ALWAYS_INLINE): New macro.
1139 * config/powerpcspe/powerpcspe-c.c
1140 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
1141 int for arg1_mode and arg2_mode.
1143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1144 Alan Hayward <alan.hayward@arm.com>
1145 David Sherwood <david.sherwood@arm.com>
1147 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
1148 Prefix mode names with E_ in case statements.
1149 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1150 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
1151 (aarch64_split_simd_move): Likewise.
1152 (aarch64_gen_storewb_pair): Likewise.
1153 (aarch64_gen_loadwb_pair): Likewise.
1154 (aarch64_gen_store_pair): Likewise.
1155 (aarch64_gen_load_pair): Likewise.
1156 (aarch64_get_condition_code_1): Likewise.
1157 (aarch64_constant_pool_reload_icode): Likewise.
1158 (get_rsqrte_type): Likewise.
1159 (get_rsqrts_type): Likewise.
1160 (get_recpe_type): Likewise.
1161 (get_recps_type): Likewise.
1162 (aarch64_gimplify_va_arg_expr): Likewise.
1163 (aarch64_simd_container_mode): Likewise.
1164 (aarch64_emit_load_exclusive): Likewise.
1165 (aarch64_emit_store_exclusive): Likewise.
1166 (aarch64_expand_compare_and_swap): Likewise.
1167 (aarch64_gen_atomic_cas): Likewise.
1168 (aarch64_emit_bic): Likewise.
1169 (aarch64_emit_atomic_swap): Likewise.
1170 (aarch64_emit_atomic_load_op): Likewise.
1171 (aarch64_evpc_trn): Likewise.
1172 (aarch64_evpc_uzp): Likewise.
1173 (aarch64_evpc_zip): Likewise.
1174 (aarch64_evpc_ext): Likewise.
1175 (aarch64_evpc_rev): Likewise.
1176 (aarch64_evpc_dup): Likewise.
1177 (aarch64_gen_ccmp_first): Likewise.
1178 (aarch64_gen_ccmp_next): Likewise.
1179 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1180 (alpha_emit_xfloating_libcall): Likewise.
1181 (emit_insxl): Likewise.
1182 (alpha_arg_type): Likewise.
1183 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
1184 (arc_preferred_simd_mode): Likewise.
1185 (arc_secondary_reload): Likewise.
1186 (get_arc_condition_code): Likewise.
1187 (arc_print_operand): Likewise.
1188 (arc_legitimate_constant_p): Likewise.
1189 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1190 * config/arc/arc.md (casesi_load): Likewise.
1191 (casesi_compact_jump): Likewise.
1192 * config/arc/predicates.md (proper_comparison_operator): Likewise.
1193 (cc_use_register): Likewise.
1194 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1195 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
1196 (arm_init_iwmmxt_builtins): Likewise.
1197 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
1198 (neon_expand_vector_init): Likewise.
1199 (arm_attr_length_move_neon): Likewise.
1200 (maybe_get_arm_condition_code): Likewise.
1201 (arm_emit_vector_const): Likewise.
1202 (arm_preferred_simd_mode): Likewise.
1203 (arm_output_iwmmxt_tinsr): Likewise.
1204 (thumb1_output_casesi): Likewise.
1205 (thumb2_output_casesi): Likewise.
1206 (arm_emit_load_exclusive): Likewise.
1207 (arm_emit_store_exclusive): Likewise.
1208 (arm_expand_compare_and_swap): Likewise.
1209 (arm_evpc_neon_vuzp): Likewise.
1210 (arm_evpc_neon_vzip): Likewise.
1211 (arm_evpc_neon_vrev): Likewise.
1212 (arm_evpc_neon_vtrn): Likewise.
1213 (arm_evpc_neon_vext): Likewise.
1214 (arm_validize_comparison): Likewise.
1215 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
1216 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
1217 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
1218 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
1219 (c6x_preferred_simd_mode): Likewise.
1220 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
1221 (epiphany_rtx_costs): Likewise.
1222 * config/epiphany/predicates.md (proper_comparison_operator):
1224 * config/frv/frv.c (condexec_memory_operand): Likewise.
1225 (frv_emit_move): Likewise.
1226 (output_move_single): Likewise.
1227 (output_condmove_single): Likewise.
1228 (frv_hard_regno_mode_ok): Likewise.
1229 (frv_matching_accg_mode): Likewise.
1230 * config/h8300/h8300.c (split_adds_subs): Likewise.
1231 (h8300_rtx_costs): Likewise.
1232 (h8300_print_operand): Likewise.
1233 (compute_mov_length): Likewise.
1234 (output_logical_op): Likewise.
1235 (compute_logical_op_length): Likewise.
1236 (compute_logical_op_cc): Likewise.
1237 (h8300_shift_needs_scratch_p): Likewise.
1238 (output_a_shift): Likewise.
1239 (compute_a_shift_length): Likewise.
1240 (compute_a_shift_cc): Likewise.
1241 (expand_a_rotate): Likewise.
1242 (output_a_rotate): Likewise.
1243 * config/i386/i386.c (classify_argument): Likewise.
1244 (function_arg_advance_32): Likewise.
1245 (function_arg_32): Likewise.
1246 (function_arg_64): Likewise.
1247 (function_value_64): Likewise.
1248 (ix86_gimplify_va_arg): Likewise.
1249 (ix86_legitimate_constant_p): Likewise.
1250 (put_condition_code): Likewise.
1251 (split_double_mode): Likewise.
1252 (ix86_avx256_split_vector_move_misalign): Likewise.
1253 (ix86_expand_vector_logical_operator): Likewise.
1254 (ix86_split_idivmod): Likewise.
1255 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
1256 (ix86_build_const_vector): Likewise.
1257 (ix86_build_signbit_mask): Likewise.
1258 (ix86_match_ccmode): Likewise.
1259 (ix86_cc_modes_compatible): Likewise.
1260 (ix86_expand_branch): Likewise.
1261 (ix86_expand_sse_cmp): Likewise.
1262 (ix86_expand_sse_movcc): Likewise.
1263 (ix86_expand_int_sse_cmp): Likewise.
1264 (ix86_expand_vec_perm_vpermi2): Likewise.
1265 (ix86_expand_vec_perm): Likewise.
1266 (ix86_expand_sse_unpack): Likewise.
1267 (ix86_expand_int_addcc): Likewise.
1268 (ix86_split_to_parts): Likewise.
1269 (ix86_vectorize_builtin_gather): Likewise.
1270 (ix86_vectorize_builtin_scatter): Likewise.
1271 (avx_vpermilp_parallel): Likewise.
1272 (inline_memory_move_cost): Likewise.
1273 (ix86_tieable_integer_mode_p): Likewise.
1274 (x86_maybe_negate_const_int): Likewise.
1275 (ix86_expand_vector_init_duplicate): Likewise.
1276 (ix86_expand_vector_init_one_nonzero): Likewise.
1277 (ix86_expand_vector_init_one_var): Likewise.
1278 (ix86_expand_vector_init_concat): Likewise.
1279 (ix86_expand_vector_init_interleave): Likewise.
1280 (ix86_expand_vector_init_general): Likewise.
1281 (ix86_expand_vector_set): Likewise.
1282 (ix86_expand_vector_extract): Likewise.
1283 (emit_reduc_half): Likewise.
1284 (ix86_emit_i387_round): Likewise.
1285 (ix86_mangle_type): Likewise.
1286 (ix86_expand_round_sse4): Likewise.
1287 (expand_vec_perm_blend): Likewise.
1288 (canonicalize_vector_int_perm): Likewise.
1289 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
1290 (expand_vec_perm_1): Likewise.
1291 (expand_vec_perm_interleave3): Likewise.
1292 (expand_vec_perm_even_odd_pack): Likewise.
1293 (expand_vec_perm_even_odd_1): Likewise.
1294 (expand_vec_perm_broadcast_1): Likewise.
1295 (ix86_vectorize_vec_perm_const_ok): Likewise.
1296 (ix86_expand_vecop_qihi): Likewise.
1297 (ix86_expand_mul_widen_hilo): Likewise.
1298 (ix86_expand_sse2_abs): Likewise.
1299 (ix86_expand_pextr): Likewise.
1300 (ix86_expand_pinsr): Likewise.
1301 (ix86_preferred_simd_mode): Likewise.
1302 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
1303 * config/i386/sse.md (*andnot<mode>3): Likewise.
1304 (<mask_codefor><code><mode>3<mask_name>): Likewise.
1305 (*<code><mode>3): Likewise.
1306 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
1307 (ia64_expand_atomic_op): Likewise.
1308 (ia64_arg_type): Likewise.
1309 (ia64_mode_to_int): Likewise.
1310 (ia64_scalar_mode_supported_p): Likewise.
1311 (ia64_vector_mode_supported_p): Likewise.
1312 (expand_vec_perm_broadcast): Likewise.
1313 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
1314 (iq2000_function_arg_advance): Likewise.
1315 (iq2000_function_arg): Likewise.
1316 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
1317 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
1318 (m68k_libcall_value): Likewise.
1319 (m68k_function_value): Likewise.
1320 (sched_attr_op_type): Likewise.
1321 * config/mcore/mcore.c (mcore_output_move): Likewise.
1322 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
1324 (microblaze_function_arg): Likewise.
1325 * config/mips/mips.c (mips16_build_call_stub): Likewise.
1326 (mips_print_operand): Likewise.
1327 (mips_mode_ok_for_mov_fmt_p): Likewise.
1328 (mips_vector_mode_supported_p): Likewise.
1329 (mips_preferred_simd_mode): Likewise.
1330 (mips_expand_vpc_loongson_even_odd): Likewise.
1331 (mips_expand_vec_unpack): Likewise.
1332 (mips_expand_vi_broadcast): Likewise.
1333 (mips_expand_vector_init): Likewise.
1334 (mips_expand_vec_reduc): Likewise.
1335 (mips_expand_msa_cmp): Likewise.
1336 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
1337 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
1338 (cc_flags_for_mode): Likewise.
1339 * config/msp430/msp430.c (msp430_print_operand): Likewise.
1340 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
1341 (nds32_output_casesi_pc_relative): Likewise.
1342 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1343 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
1344 (nvptx_gen_unpack): Likewise.
1345 (nvptx_gen_pack): Likewise.
1346 (nvptx_gen_shuffle): Likewise.
1347 (nvptx_gen_wcast): Likewise.
1348 (nvptx_preferred_simd_mode): Likewise.
1349 * config/pa/pa.c (pa_secondary_reload): Likewise.
1350 * config/pa/predicates.md (base14_operand): Likewise.
1351 * config/powerpcspe/powerpcspe-c.c
1352 (altivec_resolve_overloaded_builtin): Likewise.
1353 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
1355 (rs6000_preferred_simd_mode): Likewise.
1356 (output_vec_const_move): Likewise.
1357 (rs6000_expand_vector_extract): Likewise.
1358 (rs6000_split_vec_extract_var): Likewise.
1359 (reg_offset_addressing_ok_p): Likewise.
1360 (rs6000_legitimate_offset_address_p): Likewise.
1361 (rs6000_legitimize_address): Likewise.
1362 (rs6000_emit_set_const): Likewise.
1363 (rs6000_const_vec): Likewise.
1364 (rs6000_emit_move): Likewise.
1365 (spe_build_register_parallel): Likewise.
1366 (rs6000_darwin64_record_arg_recurse): Likewise.
1367 (swap_selector_for_mode): Likewise.
1368 (spe_init_builtins): Likewise.
1369 (paired_init_builtins): Likewise.
1370 (altivec_init_builtins): Likewise.
1371 (do_load_for_compare): Likewise.
1372 (rs6000_generate_compare): Likewise.
1373 (rs6000_expand_float128_convert): Likewise.
1374 (emit_load_locked): Likewise.
1375 (emit_store_conditional): Likewise.
1376 (rs6000_output_function_epilogue): Likewise.
1377 (rs6000_handle_altivec_attribute): Likewise.
1378 (rs6000_function_value): Likewise.
1379 (emit_fusion_gpr_load): Likewise.
1380 (emit_fusion_p9_load): Likewise.
1381 (emit_fusion_p9_store): Likewise.
1382 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
1383 (fusion_gpr_mem_load): Likewise.
1384 (fusion_addis_mem_combo_load): Likewise.
1385 (fusion_addis_mem_combo_store): Likewise.
1386 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
1387 (fusion_gpr_mem_load): Likewise.
1388 (fusion_addis_mem_combo_load): Likewise.
1389 (fusion_addis_mem_combo_store): Likewise.
1390 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1392 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
1393 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
1394 (rs6000_preferred_simd_mode): Likewise.
1395 (output_vec_const_move): Likewise.
1396 (rs6000_expand_vector_extract): Likewise.
1397 (rs6000_split_vec_extract_var): Likewise.
1398 (reg_offset_addressing_ok_p): Likewise.
1399 (rs6000_legitimate_offset_address_p): Likewise.
1400 (rs6000_legitimize_address): Likewise.
1401 (rs6000_emit_set_const): Likewise.
1402 (rs6000_const_vec): Likewise.
1403 (rs6000_emit_move): Likewise.
1404 (rs6000_darwin64_record_arg_recurse): Likewise.
1405 (swap_selector_for_mode): Likewise.
1406 (paired_init_builtins): Likewise.
1407 (altivec_init_builtins): Likewise.
1408 (rs6000_expand_float128_convert): Likewise.
1409 (emit_load_locked): Likewise.
1410 (emit_store_conditional): Likewise.
1411 (rs6000_output_function_epilogue): Likewise.
1412 (rs6000_handle_altivec_attribute): Likewise.
1413 (rs6000_function_value): Likewise.
1414 (emit_fusion_gpr_load): Likewise.
1415 (emit_fusion_p9_load): Likewise.
1416 (emit_fusion_p9_store): Likewise.
1417 * config/rx/rx.c (rx_gen_move_template): Likewise.
1418 (flags_from_mode): Likewise.
1419 * config/s390/predicates.md (s390_alc_comparison): Likewise.
1420 (s390_slb_comparison): Likewise.
1421 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
1422 (s390_vector_mode_supported_p): Likewise.
1423 (s390_cc_modes_compatible): Likewise.
1424 (s390_match_ccmode_set): Likewise.
1425 (s390_canonicalize_comparison): Likewise.
1426 (s390_emit_compare_and_swap): Likewise.
1427 (s390_branch_condition_mask): Likewise.
1428 (s390_rtx_costs): Likewise.
1429 (s390_secondary_reload): Likewise.
1430 (__SECONDARY_RELOAD_CASE): Likewise.
1431 (s390_expand_cs): Likewise.
1432 (s390_preferred_simd_mode): Likewise.
1433 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
1434 * config/sh/sh.c (sh_print_operand): Likewise.
1435 (dump_table): Likewise.
1436 (sh_secondary_reload): Likewise.
1437 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1438 * config/sh/sh.md (casesi_worker_1): Likewise.
1439 (casesi_worker_2): Likewise.
1440 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
1441 (fcc_comparison_operator): Likewise.
1442 * config/sparc/sparc.c (sparc_expand_move): Likewise.
1443 (emit_soft_tfmode_cvt): Likewise.
1444 (sparc_preferred_simd_mode): Likewise.
1445 (output_cbranch): Likewise.
1446 (sparc_print_operand): Likewise.
1447 (sparc_expand_vec_perm_bmask): Likewise.
1448 (vector_init_bshuffle): Likewise.
1449 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1450 (spu_vector_mode_supported_p): Likewise.
1451 (spu_expand_insv): Likewise.
1452 (spu_emit_branch_or_set): Likewise.
1453 (spu_handle_vector_attribute): Likewise.
1454 (spu_builtin_splats): Likewise.
1455 (spu_builtin_extract): Likewise.
1456 (spu_builtin_promote): Likewise.
1457 (spu_expand_sign_extend): Likewise.
1458 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1459 (tilegx_simd_int): Likewise.
1460 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
1461 (tilepro_simd_int): Likewise.
1462 * config/v850/v850.c (const_double_split): Likewise.
1463 (v850_print_operand): Likewise.
1464 (ep_memory_offset): Likewise.
1465 * config/vax/vax.c (vax_rtx_costs): Likewise.
1466 (vax_output_int_move): Likewise.
1467 (vax_output_int_add): Likewise.
1468 (vax_output_int_subtract): Likewise.
1469 * config/visium/predicates.md (visium_branch_operator): Likewise.
1470 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
1471 (visium_print_operand_address): Likewise.
1472 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1473 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
1474 (xtensa_expand_conditional_branch): Likewise.
1475 (xtensa_copy_incoming_a7): Likewise.
1476 (xtensa_output_literal): Likewise.
1477 * dfp.c (decimal_real_maxval): Likewise.
1478 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
1480 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1481 Alan Hayward <alan.hayward@arm.com>
1482 David Sherwood <david.sherwood@arm.com>
1484 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
1485 (mode_nunits_inline): Likewise.
1486 (mode_inner_inline): Likewise.
1487 (mode_unit_size_inline): Likewise.
1488 (mode_unit_precision_inline): Likewise.
1489 (emit_insn_modes_h): Likewise. Also emit a #define of the
1491 (emit_mode_wider): Add an E_ prefix to mode names.
1492 (emit_mode_complex): Likewise.
1493 (emit_mode_inner): Likewise.
1494 (emit_mode_adjustments): Likewise.
1495 (emit_mode_int_n): Likewise.
1496 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
1497 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
1498 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
1499 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
1500 (CRC32_BUILTIN, ENTRY): Likewise.
1501 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
1502 (aarch64_pop_regs): Likewise.
1503 (aarch64_process_components): Likewise.
1504 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
1505 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
1506 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
1507 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
1508 * config/arm/arm.c (arm_init_libfuncs): Likewise.
1509 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
1511 * config/i386/i386-builtin.def (pcmpestr): Likewise.
1512 (pcmpistr): Likewise.
1513 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
1514 * config/mmix/mmix.c (mmix_output_condition): Likewise.
1515 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
1517 * config/rl78/rl78.c (mduc_regs): Likewise.
1518 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
1519 (htm_expand_builtin): Likewise.
1520 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
1521 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
1522 * config/xtensa/xtensa.c (print_operand): Likewise.
1523 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
1524 (NUM_MODE_VECTOR_INT): Likewise.
1525 * genoutput.c (null_operand): Likewise.
1526 (output_operand_data): Likewise.
1527 * genrecog.c (print_parameter_value): Likewise.
1528 * lra.c (debug_operand_data): Likewise.
1530 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1532 * dwarf2out.c (scompare_loc_descriptor_wide)
1533 (scompare_loc_descriptor_narrow): New functions, split out from...
1534 (scompare_loc_descriptor): ...here.
1535 * expmed.c (emit_store_flag_int): New function, split out from...
1536 (emit_store_flag): ...here.
1538 2017-08-30 Richard Biener <rguenther@suse.de>
1540 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
1541 (dwarf2out_early_finish): Move setting of AT_pubnames from
1542 early debug output to early finish.
1544 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
1546 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
1547 and -mdata-region to the assembler.
1549 2017-08-30 Richard Biener <rguenther@suse.de>
1551 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
1553 (gen_subprogram_die): Add DW_AT_object_pointer only early.
1554 (dwarf2out_early_global_decl): Only generate a DIE for the
1555 abstract origin if it doesn't already exist or is a declaration DIE.
1556 (resolve_addr): Do not add the linkage name twice when
1557 generating a stub DIE for the DW_TAG_GNU_call_site target.
1559 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1561 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1562 Use machine_mode rather than int for arg1_mode.
1564 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
1567 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
1568 that the second argument of the built-in functions to unpack
1569 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
1570 switch statement instead a lot of if statements.
1571 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
1572 Allow 64-bit values to be in Altivec registers as well as
1573 traditional floating point registers.
1574 (pack<mode>, FMOVE128_VSX iterator): Likewise.
1576 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
1578 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
1579 MAX_REGS_PER_ADDRESS == 1.
1581 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
1583 * config/i386/i386.opt (flag_fentry): Do not init to -1.
1584 * config/i386/i386.c (ix86_option_override_internal): Simplify
1585 setting of opts->x_flag_entry.
1587 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1588 Jakub Jelinek <jakub@redhat.com>
1589 Richard Biener <rguenther@suse.de>
1591 PR tree-optimization/81503
1592 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
1593 folded constant fits in the target type; reorder tests for clarity.
1595 2017-08-29 Martin Liska <mliska@suse.cz>
1597 * passes.def: Include pass_lower_switch.
1598 * stmt.c (dump_case_nodes): Remove and move to
1599 tree-switch-conversion.
1600 (case_values_threshold): Likewise.
1601 (expand_switch_as_decision_tree_p): Likewise.
1602 (emit_case_decision_tree): Likewise.
1603 (expand_case): Likewise.
1604 (balance_case_nodes): Likewise.
1605 (node_has_low_bound): Likewise.
1606 (node_has_high_bound): Likewise.
1607 (node_is_bounded): Likewise.
1608 (emit_case_nodes): Likewise.
1609 (struct simple_case_node): New struct.
1610 (add_case_node): Remove.
1611 (emit_case_dispatch_table): Use vector instead of case_list.
1612 (reset_out_edges_aux): Remove.
1613 (compute_cases_per_edge): Likewise.
1614 (expand_case): Build list of simple_case_node.
1615 (expand_sjlj_dispatch_table): Use it.
1616 * tree-switch-conversion.c (struct case_node): Moved from
1617 stmt.c and adjusted.
1618 (emit_case_nodes): Likewise.
1619 (node_has_low_bound): Likewise.
1620 (node_has_high_bound): Likewise.
1621 (node_is_bounded): Likewise.
1622 (case_values_threshold): Likewise.
1623 (reset_out_edges_aux): Likewise.
1624 (compute_cases_per_edge): Likewise.
1625 (add_case_node): Likewise.
1626 (dump_case_nodes): Likewise.
1627 (balance_case_nodes): Likewise.
1628 (expand_switch_as_decision_tree_p): Likewise.
1629 (emit_jump): Likewise.
1630 (emit_case_decision_tree): Likewise.
1631 (try_switch_expansion): Likewise.
1632 (do_jump_if_equal): Likewise.
1633 (emit_cmp_and_jump_insns): Likewise.
1634 (fix_phi_operands_for_edge): New function.
1635 (record_phi_operand_mapping): Likewise.
1636 (class pass_lower_switch): New pass.
1637 (pass_lower_switch::execute): New function.
1638 (make_pass_lower_switch): Likewise.
1639 (conditional_probability):
1640 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
1641 * tree-pass.h: Add make_pass_lower_switch.
1643 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
1646 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
1649 2017-08-29 Richard Biener <rguenther@suse.de>
1651 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
1652 we do not add a DW_AT_inline attribute twice.
1653 (gen_subprogram_die): Remove code setting DW_AT_inline on
1654 DECL_ABSTRACT_P nodes.
1656 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
1658 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
1659 calls to internal functions.
1660 (gimplify_modify_expr): Likewise.
1661 * tree-call-cdce.c (use_internal_fn): Likewise.
1662 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1663 (convert_to_divmod): Set the nothrow flag.
1664 * tree-if-conv.c (predicate_mem_writes): Likewise.
1665 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
1666 (vectorizable_call): Likewise.
1667 (vectorizable_store): Likewise.
1668 (vectorizable_load): Likewise.
1669 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
1670 (vect_recog_mask_conversion_pattern): Likewise.
1672 2017-08-29 Martin Liska <mliska@suse.cz>
1675 * gcc.c (driver_handle_option): Add new argument.
1676 * opts-common.c (handle_option): Pass
1677 target_option_override_hook.
1678 * opts-global.c (lang_handle_option): Add new option.
1679 (set_default_handlers): Add new argument.
1680 (decode_options): Likewise.
1681 * opts.c (target_handle_option): Likewise.
1682 (common_handle_option): Call target_option_override_hook.
1683 * opts.h (struct cl_option_handler_func): Add hook for
1684 target option override.
1685 (struct cl_option_handlers): Likewise.
1686 (set_default_handlers): Add new argument.
1687 (decode_options): Likewise.
1688 (common_handle_option): Likewise.
1689 (target_handle_option): Likewise.
1690 * toplev.c (toplev::main): Pass targetm.target_option.override
1693 2017-08-29 Richard Biener <rguenther@suse.de>
1694 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1696 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
1697 life to the active subtree.
1699 2017-08-28 Jeff Law <law@redhat.com>
1701 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
1702 derive_equivalences.
1703 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
1705 (edge_info::derive_equivalences): New private member function
1707 * tree-ssa-dom.c (class edge_info): Changed from a struct
1708 to a class. Add ctor/dtor, methods and data members.
1709 (edge_info::edge_info): Renamed from allocate_edge_info.
1710 Initialize additional members.
1711 (edge_info::~edge_info): New.
1712 (free_dom_edge_info): Delete the edge info.
1713 (record_edge_info): Use new class & associated member functions.
1714 Tighten forms for testing for edge equivalences.
1715 (record_temporary_equivalences): Iterate over the simple
1716 equivalences rather than assuming there's only one per edge.
1717 (cprop_into_successor_phis): Iterate over the simple
1718 equivalences rather than assuming there's only one per edge.
1719 (optimize_stmt): Use operand_equal_p rather than pointer
1720 equality for mini-DSE code.
1722 2017-08-28 Nathan Sidwell <nathan@acm.org>
1724 * gcc.c (execute): Fold SIGPIPE handling into switch
1725 statement. Adjust internal error message.
1727 2017-08-28 Richard Biener <rguenther@suse.de>
1730 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
1731 Do nothing for removed DIEs.
1733 2017-08-28 Richard Biener <rguenther@suse.de>
1735 PR tree-optimization/81977
1736 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
1739 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
1742 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
1743 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
1744 using targetm.gen_mem_thread_fence.
1746 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
1749 * config/i386/i386.md (*<btsc><mode>): Change operand 2
1750 predicate to register_operand. Reorder operands.
1751 (*btr<mode>): Ditto.
1752 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
1753 (*btr<mode>_mask): Ditto.
1755 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
1757 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
1758 * config/rs6000/xmmintrin.h: New file.
1759 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
1761 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1764 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
1765 parameter and_insn and return it.
1766 (recombine_lvx_pattern): Insert a copy to ensure availability of
1767 the base register of the copied masking operation at the point of
1768 the instruction replacement.
1769 (recombine_stvx_pattern): Likewise.
1771 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
1773 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
1774 undocumented switches.
1775 (-mpower9-dform-vector): Likewise.
1776 (-mpower9-dform): Likewise.
1777 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
1778 comments to delete references to -mpower9-dform* switches.
1779 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
1780 Delete reference to -mpower9-dform* switches, test for
1781 -mpower9-vector instead.
1782 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
1783 (OTHER_P9_VECTOR_MASKS): Likewise.
1784 (POWERPC_MASKS): Likewise.
1785 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
1786 tests against -mpower9-dform* to -mpower9-vector. Delete code
1787 that checked for -mpower9-dform* consistancy with other options.
1788 Add test for -mpower9-misc to enable other power9 switches.
1789 (rs6000_init_hard_regno_mode_ok): Likewise.
1790 (rs6000_option_override_internal): Likewise.
1791 (rs6000_emit_prologue): Likewise.
1792 (rs6000_emit_epilogue): Likewise.
1793 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
1794 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
1795 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
1797 (emit_fusion_p9_store): Likewise.
1798 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
1799 resetting these macros if the assembler does not support ISA 3.0
1801 (TARGET_P9_DFORM_VECTOR): Likewise.
1802 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
1803 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
1805 2017-08-25 Alan Modra <amodra@gmail.com>
1807 PR rtl-optimization/81747
1808 * cse.c (cse_extended_basic_block): Don't attempt to record
1809 equivalences for degenerate conditional branches that branch
1810 to their fall-through.
1812 2017-08-24 Martin Sebor <msebor@redhat.com>
1815 * gimple-fold.c (size_must_be_zero_p): New function.
1816 (gimple_fold_builtin_memory_op): Call it.
1818 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
1820 * config/rs6000/mm_malloc.h: New file.
1822 2017-08-24 Bin Cheng <bin.cheng@arm.com>
1824 PR tree-optimization/81913
1825 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
1826 analysis when either IVs in condition can wrap.
1828 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
1830 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
1831 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
1833 2017-08-24 Richard Biener <rguenther@suse.de>
1836 * targhooks.c (default_target_can_inline_p): Properly
1837 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
1838 is present and always compare.
1839 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
1840 infer -mfpmath=sse from TARGET_SSE_P.
1841 (ix86_can_inline_p): Properly use target_option_default_node when
1842 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
1844 2017-08-24 Richard Biener <rguenther@suse.de>
1847 * dwarf2out.c (output_die): Handle flag_generate_offload like
1849 (output_comp_unit): Likewise.
1850 (gen_array_type_die): Likewise.
1851 (dwarf2out_early_finish): Likewise.
1852 (note_variable_value_in_expr): Likewise.
1853 (dwarf2out_finish): Likewise. Adjust assert.
1854 * cgraphunit.c (symbol_table::compile): Move setting of
1855 flag_generate_offload earlier ...
1856 (symbol_table::finalize_compilation_unit): ... here, before
1857 early debug finalization.
1859 2017-08-24 Richard Biener <rguenther@suse.de>
1861 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
1862 and ipa-fnsummary.h.
1863 (ix86_can_inline_p): When ix86_fpmath flags do not match
1864 check whether the callee uses FP math at all.
1866 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
1869 * tree-ssanames.c (get_nonzero_bits): Use element_precision
1870 instead of TYPE_PRECISION.
1872 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
1873 Alan Hayward <alan.hayward@arm.com>
1874 David Sherwood <david.sherwood@arm.com>
1876 * combine.c (make_extraction): Use subreg_offset_from_lsb.
1878 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
1880 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
1882 (ix86_frame::stack_realign_allocate): New field.
1883 (struct machine_frame_state): Modify comments.
1884 (machine_frame_state::sp_realigned_fp_end): New field.
1885 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
1887 (sp_valid_at): Add assertion to assure no attempt to access invalid
1888 offset of a realigned stack.
1889 (fp_valid_at): Likewise.
1890 (choose_baseaddr): Modify comments.
1891 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
1892 ix86_expand_prologue.
1893 (ix86_expand_prologue): Modify stack realignment and allocation.
1894 (ix86_expand_epilogue): Modify comments.
1895 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
1896 avx2_runtime, avx512f, and avx512f_runtime.
1898 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
1900 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
1901 (mstackrealign): Do not init to -1.
1902 * config/i386/i386.c (ix86_option_override_internal):
1903 Check opts_set, not opts when setting default value of
1904 opts->x_ix86_force_align_arg_pointer.
1906 2017-08-23 Richard Biener <rguenther@suse.de>
1908 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
1909 lang_hooks.decl_printable_name.
1910 * print-rtl-function.c (print_rtx_function): Likewise.
1911 * tree-pretty-print.c (dump_function_header): Likewise.
1913 2017-08-23 Richard Biener <rguenther@suse.de>
1916 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
1917 -g0 at compile-time.
1919 2017-08-23 Tamar Christina <tamar.christina@arm.com>
1922 * doc/sourcebuild.texi (Other hardware attributes):
1925 2017-08-23 Tamar Christina <tamar.christina@arm.com>
1928 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
1929 Fix single-use check.
1931 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
1933 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
1935 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
1937 * doc/install.texi: Modify to add more details on running selected
1940 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
1942 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
1943 is combined with -mabi=ms.
1944 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
1947 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1949 PR tree-optimization/81488
1950 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
1951 and cached_basis fields.
1952 (MAX_SPREAD): New constant.
1953 (alloc_cand_and_find_basis): Initialize new fields.
1954 (clear_visited): New function.
1955 (create_phi_basis_1): Rename from create_phi_basis, set visited
1956 and cached_basis fields.
1957 (create_phi_basis): New wrapper function.
1958 (phi_add_costs_1): Rename from phi_add_costs, add spread
1959 parameter, set visited field, short-circuit when limits reached.
1960 (phi_add_costs): New wrapper function.
1961 (record_phi_increments_1): Rename from record_phi_increments, set
1963 (record_phi_increments): New wrapper function.
1964 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
1965 (phi_incr_cost): New wrapper function.
1966 (all_phi_incrs_profitable_1): Rename from
1967 all_phi_incrs_profitable, set visited field.
1968 (all_phi_incrs_profitable): New wrapper function.
1970 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
1971 Alan Hayward <alan.hayward@arm.com>
1972 David Sherwood <david.sherwood@arm.com>
1974 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
1975 that takes the outer and inner modes.
1976 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
1977 comparison as the canonical test for a paradoxical subreg.
1978 * combine.c (simplify_set): Use paradoxical_subreg_p.
1979 (make_extraction): Likewise.
1980 (force_to_mode): Likewise.
1981 (rtx_equal_for_field_assignment_p): Likewise.
1982 (gen_lowpart_for_combine): Likewise.
1983 (simplify_comparison): Likewise.
1984 * cse.c (equiv_constant): Likewise.
1985 * expmed.c (store_bit_field_1): Likewise.
1986 * final.c (alter_subreg): Likewise.
1987 * fwprop.c (propagate_rtx): Likewise.
1988 (forward_propagate_subreg): Likewise.
1989 * ira-conflicts.c (ira_build_conflicts): Likewise.
1990 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
1991 * lra-constraints.c (curr_insn_transform): Likewise.
1992 (split_reg): Likewise.
1993 * lra-eliminations.c (move_plus_up): Likewise.
1994 (lra_eliminate_regs_1): Likewise.
1995 * recog.c (general_operand): Likewise.
1996 * ree.c (combine_reaching_defs): Likewise.
1997 * reload.c (push_reload): Likewise.
1998 (find_reloads): Likewise.
1999 * reload1.c (elimination_effects): Likewise.
2000 (compute_reload_subreg_offset): Likewise.
2001 (choose_reload_regs): Likewise.
2002 * rtlanal.c (subreg_lsb_1): Likewise.
2003 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2004 (simplify_subreg): Likewise.
2005 * var-tracking.c (track_loc_p): Likewise.
2006 * emit-rtl.c (byte_lowpart_offset): Likewise.
2007 (paradoxical_subreg_p): Delete out-of-line definition.
2009 2017-08-22 Jeff Law <law@redhat.com>
2011 PR tree-optimization/81741
2012 PR tree-optimization/71947
2013 * tree-ssa-dom.c: Include tree-inline.h.
2014 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
2015 equivalences if one is more expensive to compute than the other.
2016 * tree-ssa-scopedtables.h (class const_or_copies): Make
2017 record_const_or_copy_raw method private.
2018 (class avail_exprs_stack): New method simplify_binary_operation.
2019 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
2020 avail_exprs_stack::simplify_binary_operation as needed.
2021 (avail_exprs_stack::simplify_binary_operation): New function.
2023 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
2025 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
2026 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
2027 (DOT_SYMBOLS): Likewise.
2028 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
2029 (RELOCATABLE_NEEDS_FIXUP): Likewise.
2030 (RS6000_ABI_NAME): Likewise.
2031 (TARGET_CMODEL): Likewise.
2032 (TOC_SECTION_ASM_OP): Likewise.
2033 (SET_CMODEL): New macro.
2034 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
2036 2017-08-22 Richard Biener <rguenther@suse.de>
2038 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
2039 to allow for free-lang-data replacements similar to verify_type_variant.
2041 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
2042 Alan Hayward <alan.hayward@arm.com>
2043 David Sherwood <david.sherwood@arm.com>
2045 * config/aarch64/aarch64.md (casesi): Use DImode rather than
2046 VOIDmode for the LABEL_REF.
2048 2017-08-22 Richard Biener <rguenther@suse.de>
2050 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
2052 2017-08-22 Richard Biener <rguenther@suse.de>
2054 * common.opt (feliminate-dwarf2-dups): Ignore.
2055 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
2056 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
2057 same_die_p_wrap, compute_section_prefix,
2058 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
2059 (comdat_symbol_id, comdat_symbol_number): Likewise.
2060 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
2062 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
2063 (output_die): Mark unreachable path unreachable.
2064 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
2065 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
2066 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
2067 (dwarf2out_early_finish): Likewise.
2069 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
2071 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
2073 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
2076 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
2077 not VAR_P. Filter attribute warnings with OPT_Wattributes.
2078 (avr_attribute_table) <io, io_low, address>: Initialize
2079 .decl_required with true.
2081 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
2083 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
2084 undocumented debugging options.
2085 (-mvsx-scalar-double): Likewise.
2086 (-mallow-df-permute): Likewise.
2087 (-mvectorize-builtins): Likewise.
2088 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2089 (rs6000_builtin_vectorized_function): Likewise.
2090 (rs6000_builtin_md_vectorized_function): Likewise.
2091 (rs6000_opt_vars): Likewise.
2093 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
2096 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
2097 (*btrq_imm): Rename from *btrq.
2098 (*btcq_imm): Rename from *btcq.
2099 (btsc): New code attribute.
2100 (*<btsc><mode>): New insn pattern.
2101 (*btr<mode>): Ditto.
2102 (*<btsc><mode>_mask): New insn_and_split pattern.
2103 (*btr<mode>_mask): Ditto.
2105 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2106 Alan Hayward <alan.hayward@arm.com>
2107 David Sherwood <david.sherwood@arm.com>
2109 * function.c (pad_below): Simplify padding calculation.
2111 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2112 Alan Hayward <alan.hayward@arm.com>
2113 David Sherwood <david.sherwood@arm.com>
2115 * target.def (function_prologue): Remove frame size argument.
2116 (function_epilogue): Likewise.
2117 * doc/tm.texi: Regenerate.
2118 * final.c (final_start_function): Update call to function_prologue.
2119 (final_end_function): Update call to function_epilogue.
2120 (default_function_pro_epilogue): Remove frame size argument.
2121 * output.h (default_function_pro_epilogue): Likewise.
2122 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
2123 (arm_output_function_prologue): Likewise.
2124 * config/frv/frv.c (frv_function_prologue): Likewise.
2125 (frv_function_epilogue): Likewise.
2126 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
2127 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
2128 (ia64_output_function_epilogue): Likewise.
2129 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
2130 (m32r_output_function_epilogue): Likewise.
2131 * config/microblaze/microblaze.c (microblaze_function_prologue)
2132 (microblaze_function_epilogue): Likewise.
2133 * config/mips/mips.c (mips_output_function_prologue): Likewise.
2134 (mips_output_function_epilogue): Likewise.
2135 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
2136 (mmix_target_asm_function_epilogue): Likewise.
2137 * config/msp430/msp430.c (msp430_start_function): Likewise.
2138 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
2139 (nds32_asm_function_epilogue): Likewise.
2140 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
2141 * config/pa/pa.c (pa_output_function_prologue): Likewise.
2142 (pa_output_function_epilogue): Likewise.
2143 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
2144 (rs6000_output_function_epilogue): Likewise.
2145 * config/rl78/rl78.c (rl78_start_function): Likewise.
2146 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
2147 (rs6000_output_function_epilogue): Likewise.
2148 * config/rx/rx.c (rx_output_function_prologue): Likewise.
2149 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
2150 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
2151 (sparc_asm_function_epilogue): Likewise.
2153 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2155 * tree.h (type_has_mode_precision_p): New function.
2156 * convert.c (convert_to_integer_1): Use it.
2157 * expr.c (expand_expr_real_2): Likewise.
2158 (expand_expr_real_1): Likewise.
2159 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
2160 * match.pd: Likewise.
2161 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
2162 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
2163 * tree-tailcall.c (process_assignment): Likewise.
2164 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2165 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
2166 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
2167 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2168 (vectorizable_assignment): Likewise.
2169 (vectorizable_shift): Likewise.
2170 (vectorizable_operation): Likewise.
2171 * tree-vrp.c (register_edge_assert_for_2): Likewise.
2173 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
2175 * match.pd: Add pow (C, x) simplification.
2177 2017-08-21 Richard Biener <rguenther@suse.de>
2179 PR tree-optimization/81900
2180 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
2181 for blocks with abnormal predecessors.
2182 (compute_antic): Do not set visited flag prematurely.
2184 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
2187 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
2189 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2191 * stor-layout.h (vector_type_mode): Move to...
2192 * tree.h (vector_type_mode): ...here.
2193 * stor-layout.c (vector_type_mode): Move to...
2194 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
2196 2017-08-21 Richard Biener <rguenther@suse.de>
2198 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
2199 register_external_die hooks.
2200 (debug_false_tree_charstarstar_uhwistar): Declare.
2201 (debug_nothing_tree_charstar_uhwi): Likewise.
2202 * debug.c (do_nothing_debug_hooks): Adjust.
2203 (debug_false_tree_charstarstar_uhwistar): New do nothing.
2204 (debug_nothing_tree_charstar_uhwi): Likewise.
2205 * dbxout.c (dbx_debug_hooks): Adjust.
2206 (xcoff_debug_hooks): Likewise.
2207 * sdbout.c (sdb_debug_hooks): Likewise.
2208 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2209 * dwarf2out.c (macinfo_label_base): New global.
2210 (dwarf2out_register_external_die): New function for the
2211 register_external_die hook.
2212 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
2213 (dwarf2_debug_hooks): Use them.
2214 (dwarf2_lineno_debug_hooks): Adjust.
2215 (struct die_struct): Add with_offset flag.
2216 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
2217 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
2218 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
2219 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
2220 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
2221 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
2222 defining section names for the early LTO debug variants.
2223 (reset_indirect_string): New helper.
2224 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
2225 (print_dw_val): Add support for offsetted symbol references.
2226 (get_ultimate_context): Split out from is_cxx.
2227 (is_cxx): Use get_ultimate_context.
2228 (is_fortran): Add decl overload.
2229 (compute_comp_unit_symbol): Split out worker from
2230 compute_section_prefix.
2231 (compute_section_prefix): Call compute_comp_unit_symbol and
2232 set comdat_type_p here.
2233 (output_die): Skip DIE symbol output for the LTO added one.
2234 Handle DIE symbol references with offset.
2235 (output_comp_unit): Guard section name mangling properly.
2236 For LTO debug sections emit a symbol at the section beginning
2237 which we use to refer to its DIEs.
2238 (add_abstract_origin_attribute): For DIEs registered via
2239 dwarf2out_register_external_die directly refer to the early
2240 DIE rather than indirectly through the shadow one we created.
2241 Remove obsolete call to dwarf2out_abstract_function for
2242 non-function/block origins.
2243 (gen_array_type_die): When generating early LTO debug do
2244 not emit DW_AT_string_length.
2245 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
2246 late when in LTO. As suggested place a gcc_unreachable for
2247 the DECL_ABSTRACT_P case.
2248 (gen_subprogram_die): Avoid another specification DIE
2249 for early built declarations/definitions for the late LTO case.
2250 (gen_variable_die): Add type references for late duplicated VLA dies
2252 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
2253 we have the abstract instance already.
2254 (process_scope_var): Adjust decl DIE contexts in LTO which
2255 first puts them in limbo.
2256 (gen_decl_die): Do not generate type DIEs late apart from
2257 types for VLAs or for decls we do not yet have a DIE. Do not
2258 call dwarf2out_abstract_function late.
2259 (dwarf2out_early_global_decl): Make sure to create DIEs
2260 for abstract instances of a decl first.
2261 (dwarf2out_late_global_decl): Adjust comment.
2262 (output_macinfo_op): With multiple macro sections use
2263 macinfo_label_base to distinguish labels.
2264 (output_macinfo): Likewise. Update macinfo_label_base.
2265 Pass in the line info label.
2266 (note_variable_value_in_expr): When generating LTO resolve
2267 all variable values here by generating DIEs as needed.
2268 (init_sections_and_labels): Add early LTO debug flag parameter
2269 and generate different sections and names if set. Add generation
2270 counter for the labels so we can have multiple of them.
2271 (reset_dies): Helper to allow DIEs to be output multiple times.
2272 (dwarf2out_finish): When outputting DIEs to the fat part of an
2273 LTO object first reset DIEs.
2274 (dwarf2out_early_finish): Output early DIEs when generating LTO.
2275 (modified_type_die): Check for decl_ultimate_origin being self
2277 (gen_type_die_with_usage): Likewise.
2278 (gen_typedef_die): Allow decl_ultimate_origin being self.
2279 (set_decl_abstract_flags): Remove.
2280 (set_block_abstract_flags): Likewise.
2281 (dwarf2out_abstract_function): Treat the early generated DIEs
2282 as the abstract copy and only add DW_AT_inline and
2283 DW_AT_artificial here and call set_decl_origin_self.
2284 If the DIE has an abstract origin don't do anything.
2285 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
2286 if we have none yet (Go fails to build one, PR78628).
2287 (variably_modified_type_p): Prevent endless recursion for Ada
2288 cyclic pointer types.
2289 * lto-streamer-in.c: Include debug.h.
2290 (dref_queue): New global.
2291 (lto_read_tree_1): Stream in DIE references.
2292 (lto_input_tree): Register DIE references.
2293 (input_function): Stream DECL_DEBUG_ARGS.
2294 * lto-streamer-out.c: Include debug.h.
2295 (lto_write_tree_1): Output DIE references.
2296 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
2297 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
2298 (output_function): Stream DECL_DEBUG_ARGS.
2299 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
2300 Stream DECL_ABSTRACT_ORIGIN.
2301 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
2302 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
2303 DECL_CONTEXT for file-scope decls.
2304 * lto-streamer.h (struct dref_entry): Declare.
2305 (dref_queue): Likewise.
2306 * cfgexpand.c (pass_expand::execute): Do not call the
2307 outlining_inline_function hook here.
2308 * lto-wrapper.c (debug_obj): New global.
2309 (tool_cleanup): Unlink it if required.
2310 (debug_objcopy): New function.
2311 (run_gcc): Handle early debug sections in the IL files by
2312 extracting them to separate files, partially linkin them and
2313 feeding the result back as result to the linker.
2314 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
2315 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
2316 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
2317 sections into a separate segment.
2318 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
2320 (darwin_asm_dwarf_section): Likewise.
2321 (darwin_asm_output_dwarf_offset): Likewise.
2322 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
2324 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2325 Alan Hayward <alan.hayward@arm.com>
2326 David Sherwood <david.sherwood@arm.com>
2328 * read-md.h (md_reader::record_potential_iterator_use): Replace
2329 pointer argument with an rtx and an index.
2330 * read-rtl.c (iterator_group::apply_iterator): Likewise.
2331 (apply_mode_iterator): Likewise.
2332 (apply_code_iterator): Likewise.
2333 (apply_int_iterator): Likewise.
2334 (apply_subst_iterator): Likewise.
2335 (record_iterator_use): Likewise.
2336 (record_attribute_use): Likewise.
2337 (md_reader::record_potential_iterator_use): Likewise. Update calls
2338 to record_iterator_use and apply_iterator.
2339 (iterator_use): Replace ptr with x and index.
2340 (attribute_use): Likewise.
2341 (apply_attribute_uses): Update calls to apply_iterator.
2342 (apply_iterators): Likewise. Update initialization of iterator_use.
2343 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
2344 and record_potential_iterator_use.
2345 (rtx_reader::read_rtx_operand): Likewise.
2347 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2348 Alan Hayward <alan.hayward@arm.com>
2349 David Sherwood <david.sherwood@arm.com>
2351 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
2354 2017-08-21 Richard Biener <rguenther@suse.de>
2357 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
2358 at struct end conservatively when comparing common bases.
2360 2017-08-21 Richard Biener <rguenther@suse.de>
2362 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
2363 (mem_ref_in_stmt): Remove.
2364 (determine_max_movement): Use ref index to get at the reference.
2365 (invariantness_dom_walker::before_dom_children): Deal with
2366 lim data already initialized.
2367 (gather_mem_refs_stmt): Initialize lim data and record ref index.
2369 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2371 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
2372 (TARGET_ISA_ROUND): Ditto.
2373 (TARGET_ROUND): Ditto.
2374 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
2375 * config/i386/i386.md: Ditto.
2376 * config/i386/sse.md: Ditto.
2377 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
2378 with OPTION_MASK_ISA_SSE4_1.
2380 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2383 * doc/extend.texi (x86 Built-in Functions): Correct the name of
2384 __builtin_ia32_lzcnt_u16.
2386 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2389 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
2390 (rs6000_set_current_function): Rewrite function to use it.
2392 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
2395 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
2396 and TYPE_WARN_IF_NOT_ALIGN.
2397 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
2398 (handle_warn_if_not_align): New.
2399 (place_union_field): Call handle_warn_if_not_align.
2400 (place_field): Call handle_warn_if_not_align.
2401 Copy TYPE_WARN_IF_NOT_ALIGN.
2402 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
2403 (layout_type): Likewise.
2404 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
2406 (tree_decl_common): Add warn_if_not_align.
2407 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
2408 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
2409 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
2410 (DECL_WARN_IF_NOT_ALIGN): Likewise.
2411 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
2412 * doc/extend.texi: Document warn_if_not_aligned attribute.
2413 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
2415 2017-08-17 Martin Liska <mliska@suse.cz>
2418 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
2419 (get_data_dependence): Use it as pointer type.
2420 (distribute_loop): Likewise.
2422 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2424 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
2425 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
2426 (p8_vmrgow_<mode>_direct): New define_insn.
2427 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
2428 handle endianness for vmrgew and vmrgow permute patterns.
2430 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2432 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
2433 * config/rs6000/rs6000-cpus.def: Remove comment.
2434 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
2435 (POWERPC_MASKS): Likewise.
2436 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
2437 use of TARGET_VSX_TIMODE.
2438 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
2439 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
2440 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
2441 (rs6000_option_override_internal): Remove dead code.
2442 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
2443 (rs6000_legitimize_reload_address): Likewise.
2444 (rs6000_legitimate_address_p): Likewise.
2445 (rs6000_opt_masks): Delete "vsx-timode".
2446 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
2447 from function comment.
2448 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
2449 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
2450 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
2452 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
2453 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
2454 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
2457 (define_peephole2): Likewise.
2459 2017-08-17 Martin Sebor <msebor@redhat.com>
2462 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
2463 past the end of an array.
2464 (test_pp_format): Add test cases.
2466 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
2468 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
2469 missing ECF_NOTHROW flags.
2471 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2474 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
2475 operands residing in integer registers.
2476 (*vsx_le_perm_load_<mode>): Likewise.
2477 (*vsx_le_perm_store_<mode>): Likewise.
2478 (define_peephole2): Add peepholes to optimize the above.
2480 2017-08-17 Marek Polacek <polacek@redhat.com>
2483 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
2484 to mimic what shorten_compare did. Change the return type to bool.
2485 (fold_cond_expr_with_comparison): Update call to
2486 operand_equal_for_comparison_p.
2487 (fold_ternary_loc): Likewise.
2489 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
2491 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
2493 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
2495 2017-08-17 Richard Biener <rguenther@suse.de>
2497 * tree-ssa-structalias.c (solve_graph): When propagating
2498 to successors update the graphs succ edges and avoid duplicate work.
2500 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
2503 * config/i386/i386.c (ix86_option_override_internal): Save target
2504 specific options after ix86_stack_protector_guard_reg was changed.
2506 2017-08-17 Richard Biener <rguenther@suse.de>
2508 PR tree-optimization/81827
2509 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
2510 (new_var_info): Initialize it conservatively.
2511 (get_call_vi): Mark register vars.
2512 (new_scalar_tmp_constraint_exp): Likewise.
2513 (handle_rhs_call): Likewise.
2514 (handle_const_call): Likewise.
2515 (create_function_info_for): Likewise.
2516 (solve_constraints): Sort varinfos to separate register from
2517 non-register vars to pack points-to solution bitmaps during
2520 2017-08-17 Marek Polacek <polacek@redhat.com>
2522 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
2524 2017-08-17 Richard Biener <rguenther@suse.de>
2526 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
2527 to true when overflow is undefined and we saturated the result.
2529 2017-08-17 Alan Modra <amodra@gmail.com>
2532 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
2533 Don't use store multiple if only one reg needs saving.
2534 (interesting_frame_related_regno): New function.
2535 (rs6000_frame_related): Don't emit frame info for regs that
2537 (rs6000_emit_epilogue): Likewise.
2539 2017-08-16 Nathan Sidwell <nathan@acm.org>
2541 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
2542 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
2543 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
2544 * tree.c (free_lang_data_in_type): Use else-if chain. Always
2545 clear TYPE_LANG_1. Remove obsolete member-function stripping.
2546 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
2547 (verify_type): Adjust for TYPE_BINFO move.
2548 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
2549 process TYPE_BINFO directly.
2550 (hash_tree): Likewise.
2551 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
2553 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
2556 2017-08-16 David Malcolm <dmalcolm@redhat.com>
2558 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
2560 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
2563 * config/i386/i386.md (*anddi_1_btr): Change predicates of
2564 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
2565 Add ix86_binary_operator_ok to insn constraint.
2566 (*iordi_1_bts): Ditto.
2567 (*xordi_1_btc): Ditto.
2568 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
2569 Update corresponding peephole2 pattern.
2573 2017-08-16 Bin Cheng <bin.cheng@arm.com>
2575 PR tree-optimization/81832
2576 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
2577 copy loop header which has IFN_LOOP_DIST_ALIAS call.
2579 2017-08-16 Marek Polacek <polacek@redhat.com>
2582 * fold-const.c (fold_indirect_ref_1): Restore original behavior
2583 regarding size_zero_node.
2585 2017-08-16 Martin Liska <mliska@suse.cz>
2588 * config.gcc: Respect previously set extra_objs in case
2591 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
2593 PR tree-optimization/81835
2594 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
2595 the phi SSA_NAME. Check that the condition in a COND_EXPR does
2596 not depend on the phi.
2598 2017-08-16 Alan Modra <amodra@gmail.com>
2600 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
2603 2017-08-16 Alan Modra <amodra@gmail.com>
2605 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
2606 (save_reg_p): ..into this. Update all callers.
2607 (first_reg_to_save): Simplify.
2609 2017-08-16 Alan Modra <amodra@gmail.com>
2611 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
2614 2017-08-15 Joseph Myers <joseph@codesourcery.com>
2618 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
2619 constant count if that count is less than 32.
2621 2017-08-15 Nathan Sidwell <nathan@acm.org>
2623 * gcc.c (execute): Emit friendlier message if inferior is killed
2624 by an external cause.
2626 2017-08-15 Richard Biener <rguenther@suse.de>
2628 PR tree-optimization/81790
2629 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
2630 CONSTRUCTORs from simplifying and VN.
2632 2017-08-14 Martin Sebor <msebor@redhat.com>
2634 * builtin-attrs.def: Add comments.
2636 2017-08-14 Martin Sebor <msebor@redhat.com>
2639 * doc/extend.texi (attribute nonstring): Document new attribute.
2641 2017-08-14 Martin Sebor <msebor@redhat.com>
2644 * tree-diagnostic.c (default_tree_printer): Handle %G.
2645 * gimple-pretty-print.h (percent_G_format): Declare new function.
2646 * gimple-pretty-print.c (percent_G_format): Define.
2647 * tree-pretty-print.c (percent_K_format): Add argument.
2649 2017-08-14 Martin Sebor <msebor@redhat.com>
2651 PR translation/79998
2652 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
2653 Remove a stray space.
2655 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
2658 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
2659 (*iordi_1_bts): Ditto.
2660 (*xordi_1_btc): Ditto.
2662 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2665 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
2666 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2668 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
2669 quoted strings, and make more translator-friendly.
2670 (darwin_rs6000_override_options): Likewise.
2671 (rs6000_option_override_internal): Likewise.
2672 (rs6000_return_in_memory): Fix overlong line.
2673 (init_cmulative_args): Use quoted strings, and make more
2674 translator-friendly.
2675 (rs6000_pass_by_reference): Fix overlong line.
2676 (def_builtin): Use quoted strings.
2677 (altivec_expand_predicate_builtin): Use quoted strings, and make
2678 more translator-friendly.
2679 (htm_expand_builtin): Use quoted strings.
2680 (cpu_expand_builtin): Use quoted strings, and make more
2681 translator-friendly.
2682 (altivec_expand_builtin): Likewise.
2683 (paired_expand_predicate_builtin): Likewise.
2684 (rs6000_invalid_builtin): Likewise.
2685 (builtin_function_type): Use quoted strings.
2686 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
2687 more translator-friendly.
2688 (rs6000_trampoline_init): Likewise.
2689 (rs6000_handle_altivec_attribute): Likewise.
2690 (rs6000_inner_target_options): Use quoted strings.
2691 (rs6000_disable_incompatible_switches): Likewise.
2692 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
2693 strings, and make more translator-friendly.
2694 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
2696 2017-08-14 Bin Cheng <bin.cheng@arm.com>
2698 PR tree-optimization/81799
2699 * tree-loop-distribution.c (version_loop_by_alias_check): Force
2700 cond_expr to simple gimple operand.
2702 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
2705 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
2707 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
2711 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
2712 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
2713 TARGET_GASISR_PROLOGUES.
2714 * config/avr/avr.c (avr_option_override): Same.
2715 (avr_pass_pre_proep::execute): Same.
2717 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
2720 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
2721 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
2723 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
2725 * config/i386/i386.md (*load_tp_<mode>): Redefine as
2726 define_insn_and_split. Split to a memory load from 0 in
2727 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
2728 using PTR mode iterator.
2729 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
2730 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
2731 (*add_tp_<mode>): Redefine as define_insn_and_split.
2732 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
2733 address space. Merge with *add_tp_x32 using PTR mode iterator.
2734 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
2735 Split to an add with a memory load from 0 in
2736 DEFAULT_TLS_SEG_REG address space.
2738 2017-08-12 Andrew Pinski <apinski@cavium.com>
2740 * config/aarch64/aarch64-option-extensions.def (rdma):
2741 Fix feature string to what Linux prints out in /proc/cpuinfo.
2743 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
2746 * dwarf2out.c (modified_type_die): For C typedef types that have
2747 an ultimate origin, process the ultimate origin instead of the
2749 (gen_typedef_die): Assert that input DECLs have no ultimate
2751 (gen_type_die_with_usage): For typedef variants that have an
2752 ultimate origin, just call gen_decl_die on the original DECL.
2753 (process_scope_var): Avoid creating DIEs for local typedefs and
2754 concrete static variables.
2756 2017-08-12 Alan Modra <amodra@gmail.com>
2760 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
2761 match gnu-user.h startfile.
2762 (ENDFILE_LINUX_SPEC): Similarly.
2764 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
2767 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
2769 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
2771 2017-08-11 Tamar Christina <tamar.christina@arm.com>
2772 * config/aarch64/aarch64.md (mov<mode>): Change.
2773 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
2774 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
2775 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
2777 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
2779 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
2780 for storage order barriers.
2782 2017-08-11 Martin Liska <mliska@suse.cz>
2785 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
2786 variables of void type.
2788 2017-08-11 Martin Liska <mliska@suse.cz>
2790 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
2791 TARGET_SUPPORTS_ALIASES.
2792 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
2793 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
2794 (optimize_weakref): Likewise.
2795 * symtab.c (symtab_node::noninterposable_alias): Likewise.
2796 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
2797 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
2799 2017-08-11 Martin Liska <mliska@suse.cz>
2802 * config/i386/i386.c (make_resolver_func): Do complete
2803 refactoring of the function.
2805 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
2808 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
2809 * config/i386/i386.c (ix86_stack_protect_guard): Use
2810 ix86_stack_protect_guard_symbol_str to generate varible declaration.
2811 * doc/invoke.texi (x86 Options): Document
2812 -mstack-protector-guard-symbol= option.
2814 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
2816 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
2817 * config/i386/i386.c (ix86_split_stack_guard): New function.
2818 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
2819 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
2820 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
2821 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
2822 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
2823 (split_stack_space_check): Call ix86_split_stack_guard.
2825 2017-08-10 Martin Sebor <msebor@redhat.com>
2827 * print-tree.c (print_node): Print location using the established
2829 Replace spaces with colons.
2830 (debug_raw, debug): Ditto.
2832 2017-08-10 Martin Sebor <msebor@redhat.com>
2835 * pretty-print.c (pp_format): Correct the handling of %s precision.
2837 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
2840 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
2842 (ix86_finalize_stack_frame_flags): This. Also clear
2843 frame_pointer_needed if -fno-omit-frame-pointer is used without
2845 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
2846 with ix86_finalize_stack_frame_flags.
2847 (ix86_expand_epilogue): Likewise.
2848 (ix86_expand_split_stack_prologue): Likewise.
2849 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
2851 2017-08-10 Martin Liska <mliska@suse.cz>
2854 * c-attribs.c (handle_target_attribute):
2855 Report warning for an empty string argument of target attribute.
2857 2017-08-09 Jakub Jelinek <jakub@redhat.com>
2860 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
2862 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
2863 or DECL_NONLOCAL labels.
2864 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
2865 or DECL_NONLOCAL labels here.
2867 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
2869 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
2870 to indicate when early gimple folding has been disabled.
2871 (rs6000_gimple_fold_builtin): Add debug content.
2872 (rs6000_invalid_builtin): Fix whitespace.
2873 (rs6000_expand_builtin): Fix whitespace.
2874 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
2876 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
2879 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
2880 SAVE_MULTIPLE if not all the registers that saves, should be saved.
2882 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
2884 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
2885 (qdf24xx): Likewise.
2886 * config/aarch64/aarch64.md: Include falkor.md.
2887 * config/aarch64/falkor.md: New.
2889 2017-08-09 Marek Polacek <polacek@redhat.com>
2892 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
2893 * diagnostic.c (emit_diagnostic): Add a comment.
2894 (emit_diagnostic_valist): New function.
2896 2017-08-09 Marek Polacek <polacek@redhat.com>
2899 * input.c (make_location): New overload.
2900 * input.h (make_location): Declare.
2902 2017-08-08 Alan Modra <amodra@gmail.com>
2903 H.J. Lu <hongjiu.lu@intel.com>
2906 * gcc.c (NO_PIE_SPEC): Delete.
2907 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
2909 (LINK_PIE_SPEC): ..to here.
2910 (LINK_COMMAND_SPEC): Support -no-pie.
2911 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
2912 chain of crtbegin*.o selection, update for PIE_SPEC changes and
2914 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
2915 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
2916 (ENDFILE_CRTEND_SPEC): Similarly.
2918 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
2921 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
2922 (mstack-protector-guard-offset=): Ditto.
2923 * config/i386/i386.c (ix86_option_override): Handle
2924 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
2926 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
2927 ix86_stack_protect_guard_offset variables.
2928 (TARGET_STACK_PROTECT_GUARD): Always define.
2929 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
2930 and -mstack-protector-guard-offset= options.
2932 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2934 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
2935 boundary case for the last candidate.
2937 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2939 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
2940 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
2942 2017-08-08 Tamar Christina <tamar.christina@arm.com>
2945 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
2946 * config/aarch64/aarch64-builtins.c
2947 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
2948 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
2949 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
2951 2017-08-08 Tamar Christina <tamar.christina@arm.com>
2952 Andrew Pinski <pinskia@gmail.com>
2955 * internal-fn.def (XORSIGN): New.
2956 * optabs.def (xorsign_optab): New.
2957 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
2958 (convert_expand_mult_copysign): New.
2959 (pass_optimize_widening_mul::execute): Call
2960 convert_expand_mult_copysign.
2962 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2964 PR tree-optimization/81354
2965 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
2966 Insert on edges rather than explicitly creating landing pads.
2967 (analyze_candidates_and_replace): Commit edge inserts.
2969 2017-08-08 Richard Biener <rguenther@suse.de>
2972 * tree-ssa-loop-niter.c: Include tree-dfa.h.
2973 (expand_simple_operations): Also look through ADDR_EXPRs with
2974 MEM_REF bases treating them as POINTER_PLUS_EXPR.
2976 2017-08-08 Richard Biener <rguenther@suse.de>
2978 PR tree-optimization/81723
2979 * tree-vect-slp.c (struct bst_traits): New hash traits.
2980 (bst_fail): New global.
2981 (vect_build_slp_tree_2): New worker, split out from ...
2982 (vect_build_slp_tree): ... this now wrapping it with using
2983 bst_fail set to cache SLP tree build fails. Properly handle
2985 (vect_analyze_slp_instance): Allocate and free bst_fail.
2987 2017-08-08 Martin Liska <mliska@suse.cz>
2990 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
2991 LABEL_DECLs that can be from a different function.
2993 2017-08-08 Bin Cheng <bin.cheng@arm.com>
2995 PR tree-optimization/81744
2996 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
2997 loop's number of iterations.
2999 2017-08-08 Martin Liska <mliska@suse.cz>
3001 * asan.c: Include header files.
3002 * attribs.c (build_decl_attribute_variant): New function moved
3004 (build_type_attribute_qual_variant): Likewise.
3005 (cmp_attrib_identifiers): Likewise.
3006 (simple_cst_list_equal): Likewise.
3007 (omp_declare_simd_clauses_equal): Likewise.
3008 (attribute_value_equal): Likewise.
3009 (comp_type_attributes): Likewise.
3010 (build_type_attribute_variant): Likewise.
3011 (lookup_ident_attribute): Likewise.
3012 (remove_attribute): Likewise.
3013 (merge_attributes): Likewise.
3014 (merge_type_attributes): Likewise.
3015 (merge_decl_attributes): Likewise.
3016 (merge_dllimport_decl_attributes): Likewise.
3017 (handle_dll_attribute): Likewise.
3018 (attribute_list_equal): Likewise.
3019 (attribute_list_contained): Likewise.
3020 * attribs.h (lookup_attribute): New function moved from tree.[ch].
3021 (lookup_attribute_by_prefix): Likewise.
3022 * bb-reorder.c: Include header files.
3023 * builtins.c: Likewise.
3024 * calls.c: Likewise.
3025 * cfgexpand.c: Likewise.
3026 * cgraph.c: Likewise.
3027 * cgraphunit.c: Likewise.
3028 * convert.c: Likewise.
3029 * dwarf2out.c: Likewise.
3030 * final.c: Likewise.
3031 * fold-const.c: Likewise.
3032 * function.c: Likewise.
3033 * gimple-expr.c: Likewise.
3034 * gimple-fold.c: Likewise.
3035 * gimple-pretty-print.c: Likewise.
3036 * gimple.c: Likewise.
3037 * gimplify.c: Likewise.
3038 * hsa-common.c: Likewise.
3039 * hsa-gen.c: Likewise.
3040 * internal-fn.c: Likewise.
3041 * ipa-chkp.c: Likewise.
3042 * ipa-cp.c: Likewise.
3043 * ipa-devirt.c: Likewise.
3044 * ipa-fnsummary.c: Likewise.
3045 * ipa-inline.c: Likewise.
3046 * ipa-visibility.c: Likewise.
3048 * lto-cgraph.c: Likewise.
3049 * omp-expand.c: Likewise.
3050 * omp-general.c: Likewise.
3051 * omp-low.c: Likewise.
3052 * omp-offload.c: Likewise.
3053 * omp-simd-clone.c: Likewise.
3054 * opts-global.c: Likewise.
3055 * passes.c: Likewise.
3056 * predict.c: Likewise.
3057 * sancov.c: Likewise.
3058 * sanopt.c: Likewise.
3059 * symtab.c: Likewise.
3060 * toplev.c: Likewise.
3061 * trans-mem.c: Likewise.
3062 * tree-chkp.c: Likewise.
3063 * tree-eh.c: Likewise.
3064 * tree-into-ssa.c: Likewise.
3065 * tree-object-size.c: Likewise.
3066 * tree-parloops.c: Likewise.
3067 * tree-profile.c: Likewise.
3068 * tree-ssa-ccp.c: Likewise.
3069 * tree-ssa-live.c: Likewise.
3070 * tree-ssa-loop.c: Likewise.
3071 * tree-ssa-sccvn.c: Likewise.
3072 * tree-ssa-structalias.c: Likewise.
3073 * tree-ssa.c: Likewise.
3074 * tree-streamer-in.c: Likewise.
3075 * tree-vectorizer.c: Likewise.
3076 * tree-vrp.c: Likewise.
3078 * ubsan.c: Likewise.
3079 * varasm.c: Likewise.
3080 * varpool.c: Likewise.
3081 * tree.c: Remove functions moved to attribs.[ch].
3083 * config/aarch64/aarch64.c: Add attrs.h header file.
3084 * config/alpha/alpha.c: Likewise.
3085 * config/arc/arc.c: Likewise.
3086 * config/arm/arm.c: Likewise.
3087 * config/avr/avr.c: Likewise.
3088 * config/bfin/bfin.c: Likewise.
3089 * config/c6x/c6x.c: Likewise.
3090 * config/cr16/cr16.c: Likewise.
3091 * config/cris/cris.c: Likewise.
3092 * config/darwin.c: Likewise.
3093 * config/epiphany/epiphany.c: Likewise.
3094 * config/fr30/fr30.c: Likewise.
3095 * config/frv/frv.c: Likewise.
3096 * config/ft32/ft32.c: Likewise.
3097 * config/h8300/h8300.c: Likewise.
3098 * config/i386/winnt.c: Likewise.
3099 * config/ia64/ia64.c: Likewise.
3100 * config/iq2000/iq2000.c: Likewise.
3101 * config/lm32/lm32.c: Likewise.
3102 * config/m32c/m32c.c: Likewise.
3103 * config/m32r/m32r.c: Likewise.
3104 * config/m68k/m68k.c: Likewise.
3105 * config/mcore/mcore.c: Likewise.
3106 * config/microblaze/microblaze.c: Likewise.
3107 * config/mips/mips.c: Likewise.
3108 * config/mmix/mmix.c: Likewise.
3109 * config/mn10300/mn10300.c: Likewise.
3110 * config/moxie/moxie.c: Likewise.
3111 * config/msp430/msp430.c: Likewise.
3112 * config/nds32/nds32-isr.c: Likewise.
3113 * config/nds32/nds32.c: Likewise.
3114 * config/nios2/nios2.c: Likewise.
3115 * config/nvptx/nvptx.c: Likewise.
3116 * config/pa/pa.c: Likewise.
3117 * config/pdp11/pdp11.c: Likewise.
3118 * config/powerpcspe/powerpcspe.c: Likewise.
3119 * config/riscv/riscv.c: Likewise.
3120 * config/rl78/rl78.c: Likewise.
3121 * config/rx/rx.c: Likewise.
3122 * config/s390/s390.c: Likewise.
3123 * config/sh/sh.c: Likewise.
3124 * config/sol2.c: Likewise.
3125 * config/sparc/sparc.c: Likewise.
3126 * config/spu/spu.c: Likewise.
3127 * config/stormy16/stormy16.c: Likewise.
3128 * config/tilegx/tilegx.c: Likewise.
3129 * config/tilepro/tilepro.c: Likewise.
3130 * config/v850/v850.c: Likewise.
3131 * config/vax/vax.c: Likewise.
3132 * config/visium/visium.c: Likewise.
3133 * config/xtensa/xtensa.c: Likewise.
3135 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3138 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
3139 constraints since the -mupper-regs-* switches have been
3141 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
3142 into a vector from a double word element that was extracted from
3143 another vector, and eliminate extra XXPERMDI instructions.
3144 (vsx_concat_<mode>_2): Likewise.
3145 (vsx_concat_<mode>_3): Likewise.
3146 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
3147 concat to allow optimizing inserts from previous extracts.
3149 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
3151 * config/i386/i386.c (ix86_stack_protect_guard): Generate
3152 memory reference to a SSP offset in TLS address space.
3153 (ix86_print_operand) <case '@'>: Remove.
3154 (ix86_print_operand_punct_valid_p): Remove '@' code.
3155 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
3157 (stack_tls_protect_set_<mode>): Remove.
3158 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
3159 (stack_tls_protect_test_<mode>): Remove.
3160 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
3162 2017-08-07 Olivier Hainque <hainque@adacore.com>
3165 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
3167 2017-08-07 Douglas Rupp <rupp@adacore.com>
3169 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
3170 variable was referenced as multidir in command.
3172 2017-08-07 Jakub Jelinek <jakub@redhat.com>
3175 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
3178 2017-08-07 Martin Liska <mliska@suse.cz>
3180 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
3181 * config/rl78/rl78.c: Add include of attribs.h.
3182 * config/sh/sh.c: Likewise.
3183 * config/v850/v850.c: Likewise.
3185 2017-08-07 Tom de Vries <tom@codesourcery.com>
3188 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
3190 2017-08-07 Martin Liska <mliska@suse.cz>
3192 * config/mips/mips.c: Include attribs.h.
3194 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
3197 * doc/invoke.texi: Document change in behvaior for -Ofast for
3200 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
3202 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
3204 (aarch64_pop_regs): Likewise.
3205 (aarch64_gen_load_pair): Likewise.
3206 (aarch64_save_callee_saves): Likewise.
3207 (aarch64_restore_callee_saves): Likewise.
3209 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3211 * config/i386/i386.c: Revert the last change.
3213 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3216 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
3218 (ix86_finalize_stack_frame_flags): This. Also clear
3219 frame_pointer_needed if -fno-omit-frame-pointer is used without
3221 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
3222 with ix86_finalize_stack_frame_flags.
3223 (ix86_expand_epilogue): Likewise.
3224 (ix86_expand_split_stack_prologue): Likewise.
3226 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3229 * config/i386/i386.c (get_builtin_code_for_version): Set priority
3230 to P_AES for Westmere.
3232 2017-08-07 Jonathan Yong <10walls@gmail.com>
3234 * config/i386/mingw.opt (fset-stack-executable): Removed.
3235 * config/i386/cygming.opt (fset-stack-executable): Moved
3237 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
3239 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
3241 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
3243 2017-08-07 Marek Polacek <polacek@redhat.com>
3246 * fold-const.c (fold_indirect_ref_1): Check type_domain.
3248 2017-08-07 Martin Liska <mliska@suse.cz>
3250 * attribs.h (canonicalize_attr_name): New function.
3251 (cmp_attribs): Move from c-format.c and adjusted.
3252 (is_attribute_p): Moved from tree.h.
3253 * tree-inline.c: Add new includes.
3254 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
3255 (private_is_attribute_p): Remove.
3256 (private_lookup_attribute): Likewise.
3257 (private_lookup_attribute_by_prefix): Simplify.
3258 (remove_attribute): Use is_attribute_p.
3259 * tree.h: Remove removed declarations.
3261 2017-08-07 Jakub Jelinek <jakub@redhat.com>
3264 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
3265 instead of computing it in the function. Formatting fix.
3266 (expand_case): Don't rely on default_edge being the first edge,
3267 clear it if removing it, pass default_edge to
3268 emit_case_dispatch_table.
3269 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
3272 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
3274 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
3275 insn in the function, emit NOP after the insn.
3277 2017-08-06 Tom de Vries <tom@codesourcery.com>
3279 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
3282 2017-08-06 Tom de Vries <tom@codesourcery.com>
3284 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
3287 2017-08-04 Yury Gribov <tetra2005@gmail.com>
3289 PR tree-optimization/57371
3290 * match.pd: New pattern.
3292 2017-08-04 Marek Polacek <polacek@redhat.com>
3295 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
3296 perform the computation in offset_int.
3298 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3300 PR tree-optimization/81136
3301 * tree-vectorizer.h: Include tree-hash-traits.h.
3302 (vec_base_alignments): New typedef.
3303 (vec_info): Add a base_alignments field.
3304 (vect_record_base_alignments): Declare.
3305 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
3307 (DR_IS_CONDITIONAL_IN_STMT): New macro.
3308 (create_data_ref): Add an is_conditional_in_stmt argument.
3309 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
3310 the is_conditional_in_stmt field.
3311 (data_ref_loc): Add an is_conditional_in_stmt field.
3312 (get_references_in_stmt): Set the is_conditional_in_stmt field.
3313 (find_data_references_in_stmt): Update call to create_data_ref.
3314 (graphite_find_data_references_in_stmt): Likewise.
3315 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
3316 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
3317 (vect_record_base_alignment): New function.
3318 (vect_record_base_alignments): Likewise.
3319 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
3320 for nested statements even if we fail to compute a misalignment.
3321 Use pooled base alignments for unconditional references.
3322 (vect_find_same_alignment_drs): Compare base addresses instead
3324 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
3325 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
3327 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3329 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
3330 Add an explicit name for the enum. Use auto_vec for slp_instances
3332 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
3334 (_bb_vec_info): Add a constructor and destructor.
3335 (vinfo_for_stmt): Return NULL for uids of -1 as well.
3336 (destroy_loop_vec_info): Delete.
3337 (vect_destroy_datarefs): Likewise.
3338 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
3339 (vec_info::vec_info): New function.
3340 (vec_info::~vec_info): Likewise.
3341 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
3342 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
3343 destroy_loop_vec_info.
3344 * tree-vect-loop.c (new_loop_vec_info): Replace with...
3345 (_loop_vec_info::_loop_vec_info): ...this.
3346 (destroy_loop_vec_info): Replace with...
3347 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
3348 the stmt_vec_infos. Leave handling of vec_info information to its
3349 destructor. Remove explicit vector releases.
3350 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
3351 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
3352 * tree-vect-slp.c (new_bb_vec_info): Replace with...
3353 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
3354 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
3355 (destroy_bb_vec_info): Replace with...
3356 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
3357 information to its destructor.
3358 (vect_slp_analyze_bb_1): Use new and delete instead of
3359 new_bb_vec_info and destroy_bb_vec_info.
3360 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
3363 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3365 * tree-data-ref.h (subscript): Add access_fn field.
3366 (data_dependence_relation): Add could_be_independent_p.
3367 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
3368 (same_access_functions): Move to tree-data-ref.c.
3369 * tree-data-ref.c (ref_contains_union_access_p): New function.
3370 (access_fn_component_p): Likewise.
3371 (access_fn_components_comparable_p): Likewise.
3372 (dr_analyze_indices): Add a reference to access_fn_component_p.
3373 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
3375 (constant_access_functions): Likewise.
3376 (add_other_self_distances): Likewise.
3377 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
3378 (initialize_data_dependence_relation): Use XCNEW and remove
3379 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
3380 of access functions that have the same type. Allow the
3381 subsequence to end with different bases in some circumstances.
3382 Record the chosen access functions in SUB_ACCESS_FN.
3383 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
3384 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
3385 (subscript_dependence_tester_1): Likewise dra and drb.
3386 (build_classic_dist_vector): Update calls accordingly.
3387 (subscript_dependence_tester): Likewise.
3388 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
3389 DDR_COULD_BE_INDEPENDENT_P.
3390 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
3391 comp_alias_ddrs instead of may_alias_ddrs.
3392 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
3394 (vect_analyze_data_ref_dependence): Use it if
3395 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
3396 distance vectors if that fails.
3397 (dependence_distance_ge_vf): New function.
3398 (vect_prune_runtime_alias_test_list): Use it. Don't clear
3399 LOOP_VINFO_MAY_ALIAS_DDRS.
3401 2017-08-04 Richard Biener <rguenther@suse.de>
3404 * fold-const.c (fold_binary_loc): Properly restrict
3405 minus_var0 && minus_var1 case when associating undefined overflow
3408 2017-08-04 Tom de Vries <tom@codesourcery.com>
3410 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
3412 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3414 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3415 Don't start diagnostic messages with a capital letter.
3416 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3418 (rs6000_invalid_builtin): Likewise.
3419 (rs6000_trampoline_init): Likewise.
3421 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3424 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
3425 after setting changeable df flags.
3427 2017-08-03 Richard Biener <rguenther@suse.de>
3429 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
3430 up if the use is in USE - X.
3432 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
3434 * toplev.c (dumpfile.h): New include.
3435 (internal_error_reentered): New static function. Use it...
3436 (internal_error_function): ...here to handle reentered internal_error.
3438 2017-08-03 Richard Biener <rguenther@suse.de>
3441 * fold-const.c (split_tree): Add minus_var and minus_con
3442 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
3443 here but always use minus_*.
3444 (associate_trees): Assert we never associate with MINUS_EXPR
3445 and NULL first operand. Do not recurse for PLUS_EXPR operands
3446 when associating as MINUS_EXPR either.
3447 (fold_binary_loc): Track minus_var and minus_con.
3449 2017-08-03 Tom de Vries <tom@codesourcery.com>
3452 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
3453 ACCEL_COMPILER, apply finish_options on
3454 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
3456 2017-08-03 Tom de Vries <tom@codesourcery.com>
3459 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
3460 function_entry_patch_area_size > 0.
3462 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3465 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
3466 instead of 10??LU, perform unit multiplication in wide_int,
3467 don't change alloc_object_size_limit if the limit is larger
3470 PR tree-optimization/81655
3471 PR tree-optimization/81588
3472 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
3473 the case when ranges[i].low and high are 1 for unsigned type with
3477 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
3478 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
3480 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3482 * tree-vrp.h: Add include guard.
3484 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
3487 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
3488 (ud2): New insn pattern.
3489 * config/i386/i386.c (ix86_expand_epilogue):
3490 For naked functions, generate ud2 instead of trap insn.
3492 2017-08-02 Marek Polacek <polacek@redhat.com>
3495 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
3497 2017-08-02 Tom de Vries <tom@codesourcery.com>
3498 Cesar Philippidis <cesar@codesourcery.com>
3500 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
3501 Add missing edge probabilities.
3503 2017-08-02 Tamar Christina <tamar.christina@arm.com>
3505 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
3508 2017-08-02 Jakub Jelinek <jakub@redhat.com>
3511 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
3512 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
3513 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
3514 prologue_seq sequences - if any.
3516 2017-08-02 Richard Biener <rguenther@suse.de>
3518 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
3519 via vectors if supported, integer extracts via punning if supported
3520 or otherwise vector extracts.
3522 2017-08-02 Richard Biener <rguenther@suse.de>
3524 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
3526 (bitmap_insert_into_set): ... this.
3528 2017-08-02 Richard Biener <rguenther@suse.de>
3530 PR tree-optimization/81633
3532 2015-08-17 Alan Hayward <alan.hayward@arm.com>
3534 PR tree-optimization/71752
3535 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
3537 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
3539 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
3540 (machine_function::call_ms2sysv_pad_out): Remove field.
3541 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
3542 (ix86_compute_frame_layout): Likewise.
3544 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
3547 * config/i386/i386.c (ix86_set_func_type): Disallow naked
3548 attribute with interrupt attribute.
3550 2017-08-01 Andrew Pinski <apinski@cavium.com>
3552 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
3553 BIT_INSERT_EXPR's operand 1
3554 to see if the types precision matches.
3556 2017-08-01 Martin Liska <mliska@suse.cz>
3559 * builtins.c (expand_builtin_memcpy_args): Remove.
3560 (expand_builtin_memcpy): Call newly added function
3561 expand_builtin_memory_copy_args.
3562 (expand_builtin_memcpy_with_bounds): Likewise.
3563 (expand_builtin_mempcpy): Remove last argument.
3564 (expand_builtin_mempcpy_with_bounds): Likewise.
3565 (expand_builtin_memory_copy_args): New function created from
3566 expand_builtin_mempcpy_args with small modifications.
3567 (expand_builtin_mempcpy_args): Remove.
3568 (expand_builtin_stpcpy): Remove unused argument.
3569 (expand_builtin): Likewise.
3570 (expand_builtin_with_bounds): Likewise.
3572 2017-08-01 Martin Liska <mliska@suse.cz>
3575 Make mempcpy more optimal (PR middle-end/70140).
3577 2017-08-01 Jakub Jelinek <jakub@redhat.com>
3580 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
3581 __builtin_vec_cmpne verify both arguments are compatible vectors
3582 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
3583 verify arg1_type is a pointer or array type. For __builtin_vec_st,
3584 move computation of aligned to after checking the argument types.
3588 * config/rs6000/vsx.md (vextract_fp_from_shorth,
3589 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
3592 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
3593 Doug Rupp <rupp@adacore.com>
3594 Olivier Hainque <hainque@adacore.com>
3596 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
3597 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
3599 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
3600 for TARGET_OS_CPP_BUILTIN.
3601 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
3602 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
3604 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
3605 passing required abi options to the assembler for EABI configurations.
3606 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
3607 of .text.hot and .text.unlikely sections for kernel modules when
3608 using ARM style exceptions.
3609 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
3610 options. Add EXTRA_CC1_SPEC.
3611 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
3613 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
3615 (ARM_TARGET2_DWARF_FORMAT): Define.
3616 * config/arm/t-vxworks: Adjust multilib control to removal of the
3617 Diab command line options.
3619 2017-08-01 Martin Liska <mliska@suse.cz>
3621 PR gcov-profile/81561
3622 * gcov.c (unblock): Make unblocking safe as we need to preserve
3623 index correspondence of blocks and block_lists.
3625 2017-08-01 Richard Biener <rguenther@suse.de>
3627 PR tree-optimization/81181
3628 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3629 (compute_antic): ... end of iteration here.
3631 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
3633 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
3634 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
3635 (ftree-slp-vectorize): Likewise.
3636 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
3637 can no longer be set independent of flag_tree_loop_vectorize.
3638 * omp-general.c (emp_max_vf): Likewise.
3639 * opts.c (enable_fdo_optimizations): Remove references to
3640 flag_tree_vectorize, these are now implicit.
3641 (common_handle_option): Remove handling for OPT_ftree_vectorize,
3642 and leave it for the options machinery.
3644 2017-08-01 Martin Liska <mliska@suse.cz>
3647 * builtins.c (expand_builtin_memcpy_args): Remove.
3648 (expand_builtin_memcpy): Call newly added function
3649 expand_builtin_memory_copy_args.
3650 (expand_builtin_memcpy_with_bounds): Likewise.
3651 (expand_builtin_mempcpy): Remove last argument.
3652 (expand_builtin_mempcpy_with_bounds): Likewise.
3653 (expand_builtin_memory_copy_args): New function created from
3654 expand_builtin_mempcpy_args with small modifications.
3655 (expand_builtin_mempcpy_args): Remove.
3656 (expand_builtin_stpcpy): Remove unused argument.
3657 (expand_builtin): Likewise.
3658 (expand_builtin_with_bounds): Likewise.
3660 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
3663 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
3664 print "ds:" only for immediates in generic address space.
3666 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
3669 * config/i386/i386.c (ix86_funciton_naked): New prototype.
3670 (ix86_function_ok_for_sibcall): Return false for naked functions.
3672 2017-08-01 Richard Biener <rguenther@suse.de>
3674 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
3675 (compute_antic): Seed worklist with exit block predecessors.
3676 * cfganal.c (dfs_find_deadend): For a cycle return the source
3677 of the edge closing it.
3679 2017-08-01 Tamar Christina <tamar.christina@arm.com>
3681 * config/aarch64/aarch64.c
3682 (aarch64_can_const_movi_rtx_p): Move 0 check.
3684 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3686 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
3687 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
3689 * match.pd: Ditto in address comparison pattern.
3691 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3693 PR tree-optimization/81627
3694 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
3695 closed ssa form for store-store chain.
3697 2017-08-01 Bin Cheng <bin.cheng@arm.com>
3699 PR tree-optimization/81620
3700 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
3701 for store-store chain.
3703 2017-08-01 Jakub Jelinek <jakub@redhat.com>
3705 PR tree-optimization/81588
3706 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
3707 ranges[i].in_p, invert comparison code ccode. For >/>=,
3708 swap rhs1 and rhs2 and comparison code unconditionally,
3709 for </<= don't do that. Don't swap rhs1/rhs2 again if
3710 ranges[i].in_p, instead invert comparison code ccode if
3711 opcode or oe->rank is BIT_IOR_EXPR.
3714 * optabs.def (vec_extract_optab, vec_init_optab): Change from
3715 a direct optab to conversion optab.
3716 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
3717 with GET_MODE_INNER as last argument instead of optab_handler.
3718 * expmed.c (extract_bit_field_1): Likewise. Use vector from
3719 vector extraction if possible and optab is available.
3720 * expr.c (store_constructor): Use convert_optab_handler instead
3721 of optab_handler. Use vector initialization from smaller
3722 vectors if possible and optab is available.
3723 * tree-vect-stmts.c (vectorizable_load): Likewise.
3724 * doc/md.texi (vec_extract, vec_init): Document that the optabs
3726 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
3727 of vec_init from half-sized vectors with the same element mode.
3728 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
3729 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
3730 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
3731 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
3732 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
3733 after mode in gen_vec_extract* calls.
3734 (vec_extract<mode>): Renamed to ...
3735 (vec_extract<mode><ssescalarmodelower>): ... this.
3736 (vec_extract<mode><ssehalfvecmodelower>): New expander.
3737 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
3738 element mode after mode in gen_vec_init* calls.
3739 (VEC_INIT_HALF_MODE): New mode iterator.
3740 (vec_init<mode>): Renamed to ...
3741 (vec_init<mode><ssescalarmodelower>): ... this.
3742 (vec_init<mode><ssehalfvecmodelower>): New expander.
3743 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
3744 (vec_extractv2sfsf): ... this.
3745 (vec_initv2sf): Renamed to ...
3746 (vec_initv2sfsf): ... this.
3747 (vec_extractv2si): Renamed to ...
3748 (vec_extractv2sisi): ... this.
3749 (vec_initv2si): Renamed to ...
3750 (vec_initv2sisi): ... this.
3751 (vec_extractv4hi): Renamed to ...
3752 (vec_extractv4hihi): ... this.
3753 (vec_initv4hi): Renamed to ...
3754 (vec_initv4hihi): ... this.
3755 (vec_extractv8qi): Renamed to ...
3756 (vec_extractv8qiqi): ... this.
3757 (vec_initv8qi): Renamed to ...
3758 (vec_initv8qiqi): ... this.
3759 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
3760 (vec_init<mode>): Renamed to ...
3761 (vec_init<mode><VEC_base_l>): ... this.
3762 (vec_extract<mode>): Renamed to ...
3763 (vec_extract<mode><VEC_base_l>): ... this.
3764 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
3765 (vec_initv2sfsf): ... this.
3766 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
3767 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
3768 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
3769 element mode after mode in gen_vec_init* calls.
3770 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
3771 (vec_init<mode><Vel>): ... this.
3772 (vec_extract<mode>): Renamed to ...
3773 (vec_extract<mode><Vel>): ... this.
3774 * config/aarch64/iterators.md (Vel): New mode attribute.
3775 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
3776 Add element mode after mode in gen_vec_extract* calls.
3777 * config/s390/vector.md (non_vec_l): New mode attribute.
3778 (vec_extract<mode>): Renamed to ...
3779 (vec_extract<mode><non_vec_l>): ... this.
3780 (vec_init<mode>): Renamed to ...
3781 (vec_init<mode><non_vec_l>): ... this.
3782 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
3783 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
3785 * config/arm/iterators.md (V_elem_l): New mode attribute.
3786 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
3787 (vec_extract<mode><V_elem_l>): ... this.
3788 (vec_extractv2di): Renamed to ...
3789 (vec_extractv2didi): ... this.
3790 (vec_init<mode>): Renamed to ...
3791 (vec_init<mode><V_elem_l>): ... this.
3792 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
3793 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
3794 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
3795 Add element mode after gen_vec_extract* calls.
3796 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
3797 (vec_init<mode><unitmode>): ... this.
3798 (vec_extract<mode>): Renamed to ...
3799 (vec_extract<mode><unitmode>): ... this.
3800 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
3801 (vec_init<mode><unitmode>): ... this.
3802 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
3803 (vec_initv2sfsf): ... this.
3804 (vec_extractv2sf): Renamed to ...
3805 (vec_extractv2sfsf): ... this.
3806 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
3807 Add element mode after gen_vec_extract* calls.
3808 * config/mips/mips.md (unitmode): New mode iterator.
3809 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
3810 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
3811 * config/spu/spu.md (inner_l): New mode attribute.
3812 (vec_init<mode>): Renamed to ...
3813 (vec_init<mode><inner_l>): ... this.
3814 (vec_extract<mode>): Renamed to ...
3815 (vec_extract<mode><inner_l>): ... this.
3816 * config/sparc/sparc.md (veltmode): New mode iterator.
3817 (vec_init<VMALL:mode>): Renamed to ...
3818 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
3819 * config/ia64/vect.md (vec_initv2si): Renamed to ...
3820 (vec_initv2sisi): ... this.
3821 (vec_initv2sf): Renamed to ...
3822 (vec_initv2sfsf): ... this.
3823 (vec_extractv2sf): Renamed to ...
3824 (vec_extractv2sfsf): ... this.
3825 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
3826 (vec_init<mode>): Renamed to ...
3827 (vec_init<mode><VEC_base_l>): ... this.
3828 (vec_extract<mode>): Renamed to ...
3829 (vec_extract<mode><VEC_base_l>): ... this.
3830 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
3831 (vec_initv2sfsf): ... this.
3832 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
3833 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
3834 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
3835 gen_vec_init* calls.
3837 2017-08-01 Richard Biener <rguenther@suse.de>
3839 PR tree-optimization/81297
3840 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
3841 TREE_OVERFLOW from INTEGER_CSTs.
3843 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
3845 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
3847 2017-07-31 Carl Love <cel@us.ibm.com>
3849 * config/rs6000/rs6000-c: Add support for built-in functions
3850 vector signed char vec_xl_be (signed long long, signed char *);
3851 vector unsigned char vec_xl_be (signed long long, unsigned char *);
3852 vector signed int vec_xl_be (signed long long, signed int *);
3853 vector unsigned int vec_xl_be (signed long long, unsigned int *);
3854 vector signed long long vec_xl_be (signed long long, signed long long *);
3855 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
3856 vector signed short vec_xl_be (signed long long, signed short *);
3857 vector unsigned short vec_xl_be (signed long long, unsigned short *);
3858 vector double vec_xl_be (signed long long, double *);
3859 vector float vec_xl_be (signed long long, float *);
3860 * config/rs6000/altivec.h (vec_xl_be): Add #define.
3861 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
3862 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
3864 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
3865 (altivec_expand_builtin): Add switch statement to call
3866 altivec_expand_xl_be for each builtin.
3867 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
3868 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
3869 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
3870 __builtin_vsx_le_be_v16qi.
3871 * doc/extend.texi: Update the built-in documentation file for the
3872 new built-in functions.
3874 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
3877 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
3879 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
3881 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3883 * config.gcc: Add z14.
3884 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
3885 CPU model numbers for z13s and z14.
3886 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
3888 * config/s390/s390-opts.h (enum processor_type): Rename
3889 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
3890 * config/s390/s390.c (processor_table): Add field for CPU name to
3891 be passed to Binutils.
3892 (s390_asm_output_machine_for_arch): Use the new field in
3893 processor_table for Binutils.
3894 (s390_expand_builtin): Replace arch12 with z14.
3895 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
3896 (s390_get_sched_attrmask): Likewise.
3897 (s390_get_unit_mask): Likewise.
3898 * config/s390/s390.opt: Add z14 to processor_type enum.
3900 2017-07-31 Martin Jambor <mjambor@suse.cz>
3903 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
3904 regardless of optimization level.
3906 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
3907 Martin Liska <mliska@suse.cz>
3909 * predict.def: Remove old comment and adjust probability.
3910 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
3913 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
3916 * config/i386/i386.c (ix86_function_naked): New function.
3917 (ix86_can_use_return_insn_p): Return false for naked functions.
3918 (ix86_expand_prologue): Skip prologue for naked functions.
3919 (ix86_expand_epilogue): Skip epilogue for naked functions
3920 and emit trap instruction.
3921 (ix86_warn_func_return): New function.
3922 (ix86_attribute_table): Add "naked" attribute specification.
3923 (TARGET_WARN_FUNC_RETURN): Define.
3924 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
3926 2017-07-31 Martin Liska <mliska@suse.cz>
3928 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
3929 (dump_gimple_bb_header): Always dump BB info.
3930 (pp_cfg_jump): Do not append info about BB when dumping a jump.
3932 2017-07-31 Martin Liska <mliska@suse.cz>
3935 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
3936 also with current_function_decl non-null equality.
3938 2017-07-31 Jakub Jelinek <jakub@redhat.com>
3941 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
3942 change type to the element type, instead add eltype variable and
3943 use it where we are interested in the element type.
3945 PR tree-optimization/81603
3946 * ipa-polymorphic-call.c
3947 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
3948 offset arithmetic in offset_int, bail out if the resulting bit offset
3949 doesn't fit into shwi.
3951 2017-07-31 Martin Liska <mliska@suse.cz>
3953 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
3954 (gimplify_save_expr): Fix comment.
3956 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
3959 * config/i386/i386.c (ix86_function_arg): Update arguments for
3961 (ix86_compute_frame_layout): Set the initial stack offset to
3962 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
3963 INCOMING_FRAME_SP_OFFSET.
3964 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
3965 stack before exception handler returns.
3966 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
3967 the 'ERROR_CODE' for exception handler.
3969 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
3971 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
3972 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
3973 (ASM_OUTPUT_REG_POP): Ditto.
3974 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
3975 instead of asm_fprintf to output pure string.
3977 2017-07-29 Jakub Jelinek <jakub@redhat.com>
3979 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
3980 to imported_module_or_decl hook.
3981 (debug_nothing_tree_tree_tree_bool): Remove.
3982 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
3983 * debug.c (do_nothing_debug_hooks): Use
3984 debug_nothing_tree_tree_tree_bool_bool instead of
3985 debug_nothing_tree_tree_tree_bool.
3986 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
3987 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
3988 * sdbout.c (sdb_debug_hooks): Likewise.
3989 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
3990 (gen_namespace_die): Add DW_AT_export_symbols attribute if
3992 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
3993 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
3994 attribute, don't add anything.
3996 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
3998 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
3999 (fold_build2_stat_loc): Likewise.
4000 (fold_build3_stat_loc): Likewise.
4001 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
4002 (fold_build1_loc): Remove macro.
4003 (fold_build2_loc): Likewise.
4004 (fold_build3_loc): Likewise.
4006 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4008 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
4009 (gimple_build_debug_bind_source_stat): Likewise.
4010 * gimple.h (gimple_build_debug_bind): Remove macro.
4011 (gimple_build_debug_bind_source): Likewise.
4013 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4015 * bitmap.c (bitmap_alloc): Adjust.
4016 (bitmap_gc_alloc): Likewise.
4017 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
4019 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4021 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
4022 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
4023 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
4024 (bitmap_gc_alloc_stat): Likewise.
4025 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
4027 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4029 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
4030 * rtl.h (shallow_copy_rtx): Remove macro.
4032 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4034 * emit-rtl.c (gen_raw_REG): Adjust.
4035 * gengenrtl.c (gendef): Likewise.
4036 * rtl.c (rtx_alloc_stat): Remove _stat from name.
4037 * rtl.h (rtx_alloc): Remove macro.
4039 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4041 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
4042 (build_tree_list_stat): Likewise.
4043 * tree.h (build_tree_list): Remove macro.
4044 (build_tree_list_vec): Likewise.
4046 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4048 * tree.c (make_vector_stat): Remove _stat from name.
4049 (build_vector_stat): Likewise.
4050 * tree.h (make_vector_stat): Remove macro.
4051 (build_vector_stat): Likewise.
4053 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4055 * tree.h (build_var_debug_value): Remove prototype.
4057 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4059 * tree.c (tree_cons_stat): Remove _stat from name.
4060 * tree.h (tree_cons): Remove macro.
4062 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4064 * tree.c (build_vl_exp_stat): Remove _stat from name.
4065 * tree.h (build_vl_exp): Remove macro.
4067 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4069 * tree.c (build_decl_stat): Remove _stat from name.
4070 * tree.h (build_decl): Remove macro.
4072 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4074 * gimple.c (gimple_build_with_ops_stat): Adjust.
4075 (gimple_alloc_stat): Remove _stat from name.
4076 * gimple.h (gimple_alloc): Remove macro.
4078 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4080 * tree.c (make_tree_vec_stat): Remove _stat from name.
4081 (grow_tree_vec_stat): Likewise.
4082 * tree.h (make_tree_vec_stat): Adjust prototype.
4083 (grow_tree_vec_stat): Likewise.
4084 (make_tree_vec): Remove macro.
4085 (grow_tree_vec): Likewise.
4087 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4089 * fold-const.c (fold_build1_stat_loc): Adjust.
4090 (fold_build2_stat_loc): Likewise.
4091 (fold_build3_stat_loc): Likewise.
4092 * tree.c (build0_stat): Remove _stat from name.
4093 (build1_stat): Likewise.
4094 (build2_stat): Likewise.
4095 (build3_stat): Likewise.
4096 (build4_stat): Likewise.
4097 (build5_stat): Likewise.
4098 * tree.h (build1_loc): Remove macro, and rename _stat function
4100 (build2_loc): Likewise.
4101 (build3_loc): Likewise.
4102 (build4_loc): Likewise.
4103 (build5_loc): Likewise.
4105 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4107 * tree.c (make_int_cst_stat): Remove _stat from name.
4108 * tree.h (make_int_cst_stat): Adjust prototype.
4109 (make_int_cst): Remove macro.
4111 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4113 * tree.c (make_tre_binfo_stat): Remove _stat from name.
4114 * tree.h (make_tree_binfo_stat): Adjust prototype.
4115 (make_tree_binfo): Remove.
4117 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4119 * tree.c (copy_node_stat): Rename to copy_node.
4120 (build_distinct_type_copy): Adjust.
4121 * tree.h (copy_node_stat): Adjust prototype.
4122 (copy_node): Remove macro.
4124 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4126 * tree.c (make_node_stat): rename to make_node.
4127 (build_tree_list_stat): Adjust.
4128 (build0_stat): Likewise.
4129 (build2_stat): Likewise.
4130 (build3_stat): Likewise.
4131 (build4_stat): Likewise.
4132 (build5_stat): Likewise.
4133 (build_decl_stat): Likewise.
4134 * tree.h (make_node_stat): Adjust prototype.
4135 (make_node): remove macro.
4137 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
4139 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
4140 (PPC_FEATURE2_SCV): Likewise.
4141 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
4143 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4145 * config/aarch64/aarch64.c
4146 (aarch64_internal_mov_immediate): Add new special pattern.
4147 * config/aarch64/aarch64.md (*movdi_aarch64):
4148 Add reg/32bit const mov case.
4150 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4151 Richard Sandiford <richard.sandiford@linaro.org>
4153 * config/aarch64/aarch64.md (mov<mode>): Generalize.
4154 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4155 Add integer and movi cases.
4156 (movi-split-hf-df-sf split, fp16): New.
4157 (enabled): Added TARGET_FP_F16INST.
4158 * config/aarch64/iterators.md (GPF_HF): New.
4159 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
4161 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4163 * config/aarch64/aarch64.c
4164 (aarch64_simd_container_mode): Add prototype.
4165 (aarch64_expand_mov_immediate): Add HI support.
4166 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
4167 (aarch64_can_const_movi_rtx_p): New.
4168 (aarch64_preferred_reload_class):
4169 Remove restrictions of using FP registers for certain SIMD operations.
4170 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
4171 (aarch64_valid_floating_const): Add integer move validation.
4172 (aarch64_simd_imm_scalar_p): Remove.
4173 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
4174 (aarch64_legitimate_constant_p): Expand list of supported cases.
4175 * config/aarch64/aarch64-protos.h
4176 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
4177 (aarch64_reinterpret_float_as_int): New.
4178 (aarch64_simd_imm_scalar_p): Remove.
4179 * config/aarch64/constraints.md (Uvi): New.
4180 (Dd): Split into Ds and new Dd.
4181 * config/aarch64/aarch64.md (*movsi_aarch64):
4183 (*movdi_aarch64): Add SIMD mov case.
4185 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4187 * tree-predcom.c: (struct chain): Handle store-store chain in which
4188 stores for elimination only store loop invariant values.
4189 (execute_pred_commoning_chain): Ditto.
4190 (prepare_initializers_chain_store_elim): Ditto.
4191 (prepare_finalizers): Ditto.
4192 (is_inv_store_elimination_chain): New function.
4193 (initialize_root_vars_store_elim_1): New function.
4195 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4197 * tree-predcom.c: Revise general description of the pass.
4198 (enum chain_type): New enum type for store elimination.
4199 (struct chain): New field supporting store elimination.
4200 (struct component): Ditto.
4201 (dump_chain): Dump store-stores chain.
4202 (release_chain): Release resources.
4203 (split_data_refs_to_components): Compute and create component
4204 contains only stores for elimination.
4205 (get_chain_last_ref_at): New function.
4206 (make_invariant_chain): Initialization.
4207 (make_rooted_chain): Specify chain type in parameter and record it.
4208 (add_looparound_copies): Skip for store-stores chain.
4209 (determine_roots_comp): Compute type of chain and pass it to
4211 (initialize_root_vars_store_elim_2): New function.
4212 (finalize_eliminated_stores): New function.
4213 (remove_stmt): Handle store for elimination.
4214 (execute_pred_commoning_chain): Execute predictive commoning on
4216 (determine_unroll_factor): Skip unroll for store-stores chain.
4217 (prepare_initializers_chain_store_elim): New function.
4218 (prepare_initializers_chain): Hanlde store-store chain.
4219 (prepare_finalizers_chain, prepare_finalizers): New function.
4220 (tree_predictive_commoning_loop): Return integer value indicating
4221 if loop is unrolled or lcssa form is corrupted.
4222 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
4224 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4226 * tree-predcom.c (initialize_root): Delete.
4227 (execute_pred_commoning_chain): Initialize root vars and replace
4228 reference of non-combined chain directly, rather than call above
4231 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4233 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
4234 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
4236 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4238 * tree-predcom.c (struct chain): New field init_seq.
4239 (release_chain): Release init_seq.
4240 (prepare_initializers_chain): Record intialization stmts in above
4242 (insert_init_seqs): New function.
4243 (tree_predictive_commoning_loop): Call insert_init_seqs.
4245 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4247 * tree-predcom.c (determine_roots_comp): Skip trivial components.
4249 2017-07-28 Richard Biener <rguenther@suse.de>
4251 * match.pd: Remove superfluous :c.
4252 * genmatch.c (simplify::id): Add member.
4253 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
4255 (current_id): New global.
4256 (dt_node::parent): Move from ...
4257 (dt_operand::parent): ... here. Add for_id member.
4258 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
4259 (decision_tree::find_node): Relax order requirement when
4260 merging DT_TRUE nodes to ones inbetween the current simplify
4261 and the one we try to merge with. Add diagnostic whenever
4262 we need to enforce pattern order by not merging.
4263 (decision_tree::insert): Set current_id.
4264 (decision_tree::print_node): Dump parent node and for_id.
4265 (parser::last_id): Add member.
4266 (parser::push_simplify): Assign unique id.
4267 (parser::parser): Initialize last_id.
4269 2017-07-28 Martin Liska <mliska@suse.cz>
4272 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
4273 gimple_build_debug_bind.
4275 2017-07-28 Richard Biener <rguenther@suse.de>
4277 PR tree-optimization/81502
4278 * match.pd: Add pattern combining BIT_INSERT_EXPR with
4280 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
4282 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
4283 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
4284 for BIT_FIELD_REF args.
4285 * fold-const.c (make_bit_field_ref): Likewise.
4286 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
4288 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4291 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
4292 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
4293 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
4294 Or it into SANITIZER_UNDEFINED.
4295 * ubsan.c: Include gimple-fold.h and varasm.h.
4296 (ubsan_expand_ptr_ifn): New function.
4297 (instrument_pointer_overflow): New function.
4298 (maybe_instrument_pointer_overflow): New function.
4299 (instrument_object_size): Formatting fix.
4300 (pass_ubsan::execute): Call instrument_pointer_overflow
4301 and maybe_instrument_pointer_overflow.
4302 * internal-fn.c (expand_UBSAN_PTR): New function.
4303 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
4304 * sanitizer.def (__ubsan_handle_pointer_overflow,
4305 __ubsan_handle_pointer_overflow_abort): New builtins.
4306 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
4307 * internal-fn.def (UBSAN_PTR): New internal function.
4308 * opts.c (sanitizer_opts): Add pointer-overflow.
4309 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
4310 * fold-const.c (build_range_check): Compute pointer range check in
4311 integral type if pointer arithmetics would be needed. Formatting
4314 2017-07-28 Martin Liska <mliska@suse.cz>
4317 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
4318 parameters that are of a variable-length.
4320 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
4322 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
4324 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
4325 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
4326 (CRT_CALL_STATIC_FUNCTION): Likewise.
4327 (ASM_DEFAULT_SPEC): New define.
4328 (ASM_SPEC32): Likewise.
4329 (ASM_SPEC64): Likewise.
4330 (ASM_SPEC_COMMON): Likewise.
4331 (ASM_SPEC): Likewise.
4332 (INVALID_64BIT): Likewise.
4333 (LINK_OS_DEFAULT_SPEC): Likewise.
4334 (LINK_OS_SPEC32): Likewise.
4335 (LINK_OS_SPEC64): Likewise.
4336 (POWERPC_LINUX): Likewise.
4337 (PTRDIFF_TYPE): Likewise.
4338 (RESTORE_FP_PREFIX): Likewise.
4339 (RESTORE_FP_SUFFIX): Likewise.
4340 (SAVE_FP_PREFIX): Likewise.
4341 (SAVE_FP_SUFFIX): Likewise.
4342 (SIZE_TYPE): Likewise.
4343 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
4344 (TARGET_64BIT): Likewise.
4345 (TARGET_64BIT): Likewise.
4346 (TARGET_AIX): Likewise.
4347 (WCHAR_TYPE_SIZE): Likewise.
4348 (WCHAR_TYPE): Undefine.
4349 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
4350 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
4351 (CPP_OS_RTEMS_SPEC): Delete.
4352 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
4353 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
4355 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
4357 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4359 PR tree-optimization/81578
4360 * tree-parloops.c (build_new_reduction): Bail out if
4361 reduction_code isn't one of the standard OpenMP reductions.
4362 Move the details printing after that decision.
4364 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4366 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
4367 related to reload_in_progress.
4368 (splat_input_operand): Likewise.
4369 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
4371 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
4373 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
4374 (TARGET_INSTANTIATE_DECLS): Likewise.
4375 (legitimate_indexed_address_p): Delete reload_in_progress code.
4376 (rs6000_debug_legitimate_address_p): Likewise.
4377 (rs6000_eliminate_indexed_memrefs): Likewise.
4378 (rs6000_emit_le_vsx_store): Likewise.
4379 (rs6000_emit_move_si_sf_subreg): Likewise.
4380 (rs6000_emit_move): Likewise.
4381 (register_to_reg_type): Likewise.
4382 (rs6000_pre_atomic_barrier): Likewise.
4383 (rs6000_machopic_legitimize_pic_address): Likewise.
4384 (rs6000_allocate_stack_temp): Likewise.
4385 (rs6000_address_for_fpconvert): Likewise.
4386 (rs6000_address_for_altivec): Likewise.
4387 (rs6000_secondary_memory_needed_rtx): Delete function.
4388 (rs6000_check_sdmode): Likewise.
4389 (rs6000_alloc_sdmode_stack_slot): Likewise.
4390 (rs6000_instantiate_decls): Likewise.
4391 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
4392 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
4393 Delete reload_in_progress.
4394 (*vec_reload_and_plus_<mptrsize>): Likewise.
4395 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
4396 (vsx_div_v2di): Likewise.
4397 (vsx_udiv_v2di): Likewise.
4399 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4401 * config/rs6000/rs6000.opt (mlra): Replace with stub.
4402 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
4403 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
4404 (rs6000_debug_reg_global): Delete print of LRA status.
4405 (rs6000_option_override_internal): Delete dead LRA related code.
4406 (rs6000_lra_p): Delete function.
4407 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
4409 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
4411 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
4412 * config/riscv/rtems.h: New file.
4414 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4415 Sudakshina Das <sudi.das@arm.com>
4417 * config/aarch64/aarch64.md
4418 (define_split for and<mode>3nr_compare): Move
4419 non aarch64_logical_operand to a register.
4420 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
4421 register immediate operand to a register.
4422 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
4424 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4427 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
4429 2017-07-27 Richard Biener <rguenther@suse.de>
4431 PR tree-optimization/81573
4432 PR tree-optimization/81494
4433 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
4434 multi defuse cycle case.
4436 2017-07-27 Richard Biener <rguenther@suse.de>
4438 PR tree-optimization/81571
4439 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
4442 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
4444 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
4445 earlier and only if MASK_FPU is set. Adjust formatting.
4447 2017-07-27 Martin Liska <mliska@suse.cz>
4449 * opt-functions.awk: Add validation of value of Init.
4450 * optc-gen.awk: Pass new argument.
4452 2017-07-27 Martin Liska <mliska@suse.cz>
4454 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
4455 Fix wrong condition.
4457 2017-07-27 Martin Liska <mliska@suse.cz>
4459 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
4460 BBs and edges seen by autoFDO.
4462 2017-07-27 Richard Biener <rguenther@suse.de>
4464 PR tree-optimization/81502
4465 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
4466 with incompatible but same sized type.
4467 (execute_update_addresses_taken): Likewise.
4469 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
4471 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
4472 flag_tree_loop_vectorize rather than flag_tree_vectorize.
4474 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4477 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
4478 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
4479 Change s_operand to memory_operand.
4481 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
4483 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
4484 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
4485 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
4486 Emit instructions rather than returning an expression. Handle TFmode
4487 and KFmode by casting to TImode.
4488 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
4489 (rs6000_emit_le_vsx_store): Likewise.
4490 * config/rs6000/vsx.md (VSX_TI): New iterator.
4491 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
4492 (*vsx_le_undo_permute_<mode>): Likewise.
4493 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
4494 emit the split sequence.
4495 (*vsx_le_perm_store_<mode>): Likewise.
4497 2017-07-27 Jakub Jelinek <jakub@redhat.com>
4499 PR tree-optimization/81555
4500 PR tree-optimization/81556
4501 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
4502 if true, force CHANGED for the recursive invocation.
4503 (reassociate_bb): Remember original length of ops array, pass
4504 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
4506 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
4507 attributes for noipa attribute. For naked attribute use
4508 lookup_attribute first before lookup_attribute_spec.
4509 * final.c (rest_of_handle_final): Disable IPA RA for functions with
4511 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
4512 for functions with noipa attribute.
4513 (cgraph_externally_visible_p): Return true for functions with noipa
4515 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
4516 for functions with noipa attribute.
4517 * doc/extend.texi: Document noipa function attribute.
4518 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
4519 also for functions with noipa attribute.
4520 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
4522 2017-07-26 Andrew Pinski <apinski@cavium.com>
4524 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
4525 vec_unalign_load_cost and vec_unalign_store_cost.
4527 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4529 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
4530 -mvsx-small-integer option.
4531 (ISA_3_0_MASKS_IEEE): Likewise.
4532 (OTHER_VSX_VECTOR_MASKS): Likewise.
4533 (POWERPC_MASKS): Likewise.
4534 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
4535 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
4536 code, only testing for DImode being allowed in non-VSX floating
4538 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
4539 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
4541 (rs6000_option_override_internal): Delete -mvsx-small-integer.
4542 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
4543 TARGET_P8_VECTOR test.
4544 (rs6000_secondary_reload_simple_move): Likewise.
4545 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
4546 since TARGET_P9_VECTOR was already tested.
4547 (rs6000_opt_masks): Remove -mvsx-small-integer.
4548 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
4549 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4551 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
4552 test for TARGET_VEXTRACTUB was used, and that uses
4554 (p9 extract splitter): Likewise.
4555 (vsx_extract_<mode>_di_p9): Likewise.
4556 (vsx_extract_<mode>_store_p9): Likewise.
4557 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
4558 for TARGET_P9_VECTOR was used. Delete code that is now dead with
4559 the elimination of TARGET_VSX_SMALL_INTEGER.
4560 (vsx_extract_<mode>_p8): Likewise.
4561 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
4562 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
4563 (vsx_set_<mode>_p9): Likewise.
4564 (vsx_set_v4sf_p9): Likewise.
4565 (vsx_set_v4sf_p9_zero): Likewise.
4566 (vsx_insert_extract_v4sf_p9): Likewise.
4567 (vsx_insert_extract_v4sf_p9_2): Likewise.
4568 * config/rs6000/rs6000.md (sign extend splitter): Change
4569 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
4570 (floatsi<mode>2_lfiwax_mem): Likewise.
4571 (floatunssi<mode>2_lfiwzx_mem): Likewise.
4572 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
4573 since a test for TARGET_P9_VECTOR was used.
4574 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4575 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
4576 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4577 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
4578 TARGET_P8_VECTOR test.
4579 (fix_trunc<mode>si2_stfiwx): Likewise.
4580 (fix_trunc<mode>si2_internal): Likewise.
4581 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
4582 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4584 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
4585 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
4586 TARGET_P8_VECTOR test.
4587 (fixuns_trunc<mode>si2_stfiwx): Likewise.
4588 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
4589 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4591 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
4592 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
4593 since a test for TARGET_P9_VECTOR was used.
4594 (splitter for loading small constants): Likewise.
4596 2017-07-26 Andrew Pinski <apinski@cavium.com>
4598 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
4601 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
4604 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
4605 (fp_valid_at): Likewise.
4607 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
4609 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
4610 (qdf24xx_addrcost_table): Likewise.
4611 (cortexa57_tunings): Update to use generic_branch_cost.
4612 (cortexa72_tunings): Likewise.
4613 (cortexa73_tunings): Likewise.
4614 (qdf24xx_tunings): Likewise.
4616 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
4618 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
4619 (thunderx2t99_branch_cost): Likewise.
4620 (cortexa35_tunings): Update to use generic_branch_cost.
4621 (cortexa53_tunings): Likewise.
4622 (cortexa57_tunings): Likewise.
4623 (cortexa72_tunings): Likewise.
4624 (cortexa73_tunings): Likewise.
4625 (thunderx2t99_tunings): Likewise.
4627 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
4629 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
4630 (sparc_option_override): Honour MASK_FSMULD.
4631 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
4632 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
4633 * config/sparc/sparc.opt (mfsmuld): New option.
4634 * doc/invoke.texi (mfsmuld): Document option.
4636 2017-07-26 Marek Polacek <polacek@redhat.com>
4639 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
4641 2017-07-26 Richard Biener <rguenther@suse.de>
4643 * gimple-match-head.c (do_valueize): Return OP if valueize
4645 (get_def): New helper to get at the def stmt of a SSA name
4647 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
4648 do_valueize to get at the def stmt.
4649 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
4651 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
4654 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
4656 2017-07-26 Martin Liska <mliska@suse.cz>
4659 * function.c (expand_function_start): Make expansion of
4660 nonlocal_goto_save_area after parm_birth_insn.
4662 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
4664 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
4665 from all CPU target flags enable members.
4667 2017-07-26 Richard Biener <rguenther@suse.de>
4669 * genmatch.c (dt_simplify::gen): Make iterator vars const.
4670 (decision_tree::gen): Make 'type' const.
4671 (write_predicate): Likewise.
4673 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
4675 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
4676 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
4677 (rs6000_option_override_internal): Likewise.
4678 (rs6000_expand_vector_set): Likewise.
4679 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
4680 (TARGET_UPPER_REGS_SF): Likewise.
4681 (TARGET_UPPER_REGS_DI): Likewise.
4682 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
4683 (TARGET_DIRECT_MOVE_64BIT): Likewise.
4684 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
4685 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
4686 (Splitters for DI constants in Altivec registers): Likewise.
4687 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
4688 (vsx_set_v4sf_p9): Likewise.
4689 (vsx_set_v4sf_p9_zero): Likewise.
4690 (vsx_insert_extract_v4sf_p9): Likewise.
4691 (vsx_insert_extract_v4sf_p9_2): Likewise.
4693 2017-07-25 Carl Love <cel@us.ibm.com>
4695 * doc/extend.texi: Update the built-in documentation file for the
4696 existing built-in functions
4697 vector signed char vec_cnttz (vector signed char);
4698 vector unsigned char vec_cnttz (vector unsigned char);
4699 vector signed short vec_cnttz (vector signed short);
4700 vector unsigned short vec_cnttz (vector unsigned short);
4701 vector signed int vec_cnttz (vector signed int);
4702 vector unsigned int vec_cnttz (vector unsigned int);
4703 vector signed long long vec_cnttz (vector signed long long);
4704 vector unsigned long long vec_cnttz (vector unsigned long long);
4706 2017-07-25 Andrew Pinski <apinski@cavium.com>
4708 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
4709 accesses where the use is for the first operand of a BIT_INSERT.
4711 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
4714 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
4715 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
4717 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
4719 * config/i386/gstabs.h: Delete.
4720 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
4722 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
4724 * config/i386/i386.c (ix86_decompose_address): Do not check for
4725 register RTX when looking at index_reg or base_reg.
4726 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
4728 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
4730 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
4731 to update EH info here.
4733 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4735 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
4737 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4739 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
4741 2017-07-25 Torsten Duwe <duwe@suse.de>
4743 * common.opt: Introduce -fpatchable-function-entry
4744 command line option, and its variables function_entry_patch_area_size
4745 and function_entry_patch_area_start.
4746 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
4747 including a two-value parser.
4748 * target.def (print_patchable_function_entry): New target hook.
4749 * targhooks.h (default_print_patchable_function_entry): New function.
4750 * targhooks.c (default_print_patchable_function_entry): Likewise.
4751 * toplev.c (process_options): Switch off IPA-RA if
4752 patchable function entries are being generated.
4753 * varasm.c (assemble_start_function): Look at the
4754 patchable-function-entry command line switch and current
4755 function attributes and maybe generate NOP instructions by
4756 calling the print_patchable_function_entry hook.
4757 * doc/extend.texi: Document patchable_function_entry attribute.
4758 * doc/invoke.texi: Document -fpatchable_function_entry
4759 command line option.
4760 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
4762 * doc/tm.texi: Re-generate.
4764 2017-07-25 Jakub Jelinek <jakub@redhat.com>
4767 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
4768 TARGET_AVX512DQ rather than TARGET_AVX512BW.
4770 2017-07-25 Tamar Christina <tamar.christina@arm.com>
4772 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
4774 2017-07-25 Richard Biener <rguenther@suse.de>
4776 PR tree-optimization/81455
4777 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
4778 not walk in cycles when looking for guards.
4780 2017-07-25 Richard Biener <rguenther@suse.de>
4782 PR tree-optimization/81529
4783 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
4784 when optimizing backedge uses.
4786 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
4788 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
4790 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
4791 to dl_section_ref. On AIX, append an expression to subtract
4792 the size of the section length to dl_section_ref.
4794 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
4796 * configure.ac: If any of the config.* scripts fail, exit 1.
4797 * configure: Regenerate.
4799 2017-07-25 Richard Biener <rguenther@suse.de>
4802 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
4803 of immediate uses, be more verbose on errors.
4805 2017-07-25 Richard Biener <rguenther@suse.de>
4807 PR tree-optimization/81510
4808 * tree-vect-loop.c (vect_is_simple_reduction): When the
4809 reduction stmt is not inside the loop bail out.
4811 2017-07-25 Richard Biener <rguenther@suse.de>
4813 PR tree-optimization/81303
4814 * tree-vect-loop-manip.c (vect_loop_versioning): Build
4815 profitability check against LOOP_VINFO_NITERSM1.
4817 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
4819 * domwalk.c (cmp_bb_postorder): Simplify.
4820 (sort_bbs_postorder): New function. Use it...
4821 (dom_walker::walk): ...here to optimize common cases.
4823 2017-07-25 Martin Liska <mliska@suse.cz>
4826 * ipa-visibility.c (function_and_variable_visibility): Make the
4827 redirection just on target that supports aliasing.
4828 Fix GNU coding style.
4830 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
4833 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
4835 * config/arm/rtems.h: Likewise.
4836 * config/bfin/rtems.h: Likewise.
4837 * config/i386/rtemself.h: Likewise.
4838 * config/lm32/rtems.h: Likewise.
4839 * config/m32c/rtems.h: Likewise.
4840 * config/m68k/rtemself.h: Likewise.
4841 * config/microblaze/rtems.h: Likewise.
4842 * config/mips/rtems.h: Likewise.
4843 * config/moxie/rtems.h: Likewise.
4844 * config/nios2/rtems.h: Likewise.
4845 * config/powerpcspe/rtems.h: Likewise.
4846 * config/rs6000/rtems.h: Likewise.
4847 * config/rtems.h: Likewise.
4848 * config/sh/rtems.h: Likewise.
4849 * config/sh/rtemself.h: Likewise.
4850 * config/sparc/rtemself.h: Likewise.
4852 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
4855 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
4856 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
4857 * tree-ssa-structalias.c (alias_get_name): Same.
4859 2017-07-25 Bin Cheng <bin.cheng@arm.com>
4862 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
4863 instructions if no du chain is found.
4865 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
4867 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
4869 2017-07-25 Richard Biener <rguenther@suse.de>
4872 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
4875 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
4877 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
4879 (ISA_2_7_MASKS_SERVER): Likewise.
4880 (ISA_3_0_MASKS_IEEE): Likewise.
4881 (OTHER_P8_VECTOR_MASKS): Likewise.
4882 (OTHER_VSX_VECTOR_MASKS): Likewise.
4883 (POWERPC_MASKS): Likewise.
4884 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
4885 duplicate list of options.
4886 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
4887 explicit -mupper-regs options.
4888 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
4889 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
4890 alias for -mupper-regs-df.
4891 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
4892 (rs6000_init_hard_regno_mode_ok): Likewise.
4893 (rs6000_option_override_internal): Likewise.
4894 (rs6000_opt_masks): Likewise.
4895 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
4896 options in terms of whether -mvsx or -mpower8-vector was used.
4897 (TARGET_UPPER_REGS_DI): Likewise.
4898 (TARGET_UPPER_REGS_SF): Likewise.
4899 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
4900 -mupper-regs-* options.
4902 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
4904 * passes.c (emergency_dump_function): Print some empty lines and a
4905 header before the RTL dump.
4907 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
4909 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
4911 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
4914 * config/aarch64/aarch64.c (aarch64_classify_symbol):
4915 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
4917 2017-07-24 Carl Love <cel@us.ibm.com>
4919 * config/rs6000/rs6000-c.c: Add support for built-in functions
4920 vector float vec_extract_fp32_from_shorth (vector unsigned short);
4921 vector float vec_extract_fp32_from_shortl (vector unsigned short);
4922 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
4923 vec_extract_fp_from_shortl): Add defines for the two builtins.
4924 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
4925 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
4927 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
4928 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
4929 * doc/extend.texi: Update the built-in documentation file for the
4930 new built-in function.
4932 2017-07-24 Jakub Jelinek <jakub@redhat.com>
4935 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
4937 * doc/generic.texi: Likewise.
4938 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
4939 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
4941 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
4943 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
4944 (aarch64_mls_elt_merge<mode>): Likewise.
4946 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
4948 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
4949 having __cxa_atexit.
4951 2017-07-23 Michael Collison <michael.collison@arm.com>
4953 * config/arm/arm.c (arm_option_override): Deprecate
4954 use of -mstructure-size-boundary.
4955 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
4956 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
4958 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4961 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
4962 Reduce cost estimate for direct moves.
4964 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
4967 * config/i386/i386.c (ix86_option_override_internal): Disable
4968 BMI, BMI2 and TBM instructions for -m16.
4970 2017-07-21 Carl Love <cel@us.ibm.com>
4972 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
4973 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
4974 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
4975 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
4976 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
4977 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
4978 VMULOSW): New enum "unspec" values.
4979 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
4980 altivec_vmulosw): New patterns.
4981 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
4982 VMULOSW): Add definitions.
4984 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
4986 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
4987 (qdf24xx): Likewise.
4988 * config/aarch64/aarch64-options-extensions.def (rdma); New.
4989 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
4990 (AARCH64_FL_V8_1): Renumber.
4991 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
4992 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
4993 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
4994 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
4995 rdma to feature modifiers list.
4997 2017-07-21 Yury Gribov <tetra2005@gmail.com>
5000 * ipa-visibility (function_and_variable_visibility): Convert
5001 recursive PLT call to direct call if appropriate.
5003 2017-07-21 Andrew Pinski <apinski@cavium.com>
5005 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
5006 operand 1 to see if the types precision matches.
5007 * fold-const.c (operand_equal_p): Likewise.
5009 2017-07-21 Richard Biener <rguenther@suse.de>
5011 PR tree-optimization/81303
5012 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
5013 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
5014 (vect_peeling_hash_get_lowest_cost): Adjust.
5015 (vect_enhance_data_refs_alignment): Likewise. Use
5016 vect_get_peeling_costs_all_drs to compute the penalty for no
5017 peeling to match up costs.
5019 2017-07-21 Richard Biener <rguenther@suse.de>
5021 PR tree-optimization/81500
5022 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
5023 we didn't identify a reduction path.
5025 2017-07-21 Tom de Vries <tom@codesourcery.com>
5026 Cesar Philippidis <cesar@codesourcery.com>
5028 PR gcov-profile/81442
5029 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
5032 2017-07-21 Tom de Vries <tom@codesourcery.com>
5035 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
5037 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
5038 nvptx_override_options_after_change.
5040 2017-07-21 Ulrich Drepper <drepper@redhat.com>
5042 * dwarf2out.c (output_file_names): Avoid double testing for
5045 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
5047 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
5049 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5051 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
5053 (try_crossjump_to_edge): Do not punt on partitioned functions.
5055 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5057 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
5058 Put all BBs reachable only via paths crossing cold region to cold
5060 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
5062 2016-07-21 Richard Biener <rguenther@suse.de>
5064 PR tree-optimization/81303
5065 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
5066 into account prologue and epilogue iterations when raising
5067 min_profitable_iters to sth at least covering one vector iteration.
5069 2017-07-21 Tamar Christina <tamar.christina@arm.com>
5071 * config/arm/arm.c (arm_test_cpu_arch_dat):
5074 2017-07-20 Nathan Sidwell <nathan@acm.org>
5076 Remove TYPE_METHODS.
5077 * tree.h (TYPE_METHODS): Delete.
5078 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
5079 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
5080 (dbxout_type_methods): Scan TYPE_FIELDS.
5081 (dbxout_type): Don't check TYPE_METHODS here.
5082 * function.c (use_register_for_decl): Always ignore register for
5083 class types when not optimizing.
5084 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
5085 * tree.c (free_lang_data_in_type): Stitch out member functions and
5086 templates from TYPE_FIELDS.
5087 (build_distinct_type_copy, verify_type_variant,
5088 verify_type): Member fns are on TYPE_FIELDS.
5089 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
5090 * tree-pretty-print.c (dump_generic_node): Likewise.
5092 2017-07-20 Jakub Jelinek <jakub@redhat.com>
5095 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
5096 V2TImode and V4TImode.
5097 (ix86_expand_vector_extract): Likewise.
5098 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
5099 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
5100 (ssescalarmode): Handle V4TImode and V2TImode.
5101 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
5102 (*vec_extractv2ti, *vec_extractv4ti): New insns.
5103 (VEXTRACTI128_MODE): New mode iterator.
5104 (splitter for *vec_extractv?ti first element): New.
5105 (VEC_INIT_MODE): New mode iterator.
5106 (vec_init<mode>): Consolidate 3 expanders into one using
5107 VEC_INIT_MODE mode iterator.
5109 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
5111 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
5112 non_spilled_static_chain_regno_p.
5114 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
5116 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
5118 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
5120 * bb-reorder.c (connect_traces): Allow copying of blocks within
5123 2017-07-20 Richard Biener <rguenther@suse.de>
5125 * gimple.h (gimple_phi_result): Add gphi * overload.
5126 (gimple_phi_result_ptr): Likewise.
5127 (gimple_phi_arg): Likewise. Adjust index assert to only
5128 allow actual argument accesses rather than all slots available
5130 (gimple_phi_arg_def): Add gphi * overload.
5131 * tree-phinodes.c (make_phi_node): Initialize only actual
5133 (resize_phi_node): Clear memory not covered by old node,
5134 do not initialize excess argument slots.
5135 (reserve_phi_args_for_new_edge): Initialize new argument slot
5138 2017-07-20 Bin Cheng <bin.cheng@arm.com>
5140 PR tree-optimization/81388
5142 2016-07-21 Bin Cheng <bin.cheng@arm.com>
5144 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
5145 by removing computation of may_be_zero.
5147 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5148 Tom de Vries <tom@codesourcery.com>
5151 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
5152 when gimple level profile disagrees with what RTL expander did.
5154 2017-07-20 Richard Biener <rguenther@suse.de>
5156 PR tree-optimization/61171
5157 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
5158 (vect_analyze_stmt): Add slp instance parameter.
5159 (vectorizable_reduction): Likewise.
5160 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
5161 (vect_is_simple_reduction): Deal with chains not detected
5162 as SLP reduction chain, specifically not properly associated
5163 chains containing a mix of plus/minus.
5164 (get_reduction_op): Remove.
5165 (get_initial_defs_for_reduction): Simplify, pass in whether
5166 this is a reduction chain, pass in the SLP node for the PHIs.
5167 (vect_create_epilog_for_reduction): Get the SLP instance as
5169 (vectorizable_reduction): Get the SLP instance as arg.
5170 During analysis remember the SLP node with the PHIs in the
5171 instance. Simplify getting at the vectorized reduction PHIs.
5172 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
5173 through SLP instance.
5174 (vect_slp_analyze_operations): Likewise.
5175 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
5176 (vect_transform_stmt): Likewise.
5178 2017-07-20 Tom de Vries <tom@codesourcery.com>
5180 PR tree-optimization/81489
5181 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
5182 read of phi arg location to before loop that modifies phi.
5184 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5186 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
5189 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5192 * except.c (execute): Fix ordering issue.
5194 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
5196 PR rtl-optimization/81423
5197 * combine.c (make_compound_operation_int): Don't try to optimize
5198 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
5200 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
5202 PR rtl-optimization/81423
5203 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
5204 with a constant that is -1 in the truncated to mode.
5206 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5208 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
5209 (determine_unlikely_bbs): ... here.
5210 * predict.h (propagate_unlikely_bbs_forward): Declare.
5211 * cfgexpand.c (pass_expand::execute): Use it.
5212 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
5214 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
5215 propagate_unlikely_bbs_forward.
5217 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5220 * except.c (maybe_add_nop_after_section_switch): New function.
5223 2017-07-19 Tom de Vries <tom@codesourcery.com>
5225 * gimple.h (gimple_phi_set_arg): Make assert more strict.
5227 2017-07-19 Tom de Vries <tom@codesourcery.com>
5229 * gimple.h (gimple_phi_arg): Make assert more strict.
5231 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
5233 * config.gcc (powerpc*-*-*): Add mmintrin.h.
5234 * config/rs6000/mmintrin.h: New file.
5235 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
5237 2017-07-19 Jakub Jelinek <jakub@redhat.com>
5239 PR tree-optimization/81346
5240 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
5242 2017-07-19 Tom de Vries <tom@codesourcery.com>
5244 * config/nvptx/nvptx.md (VECIM): Add V2DI.
5246 2017-07-19 Tom de Vries <tom@codesourcery.com>
5248 * config/nvptx/nvptx-modes.def: Add V2DImode.
5249 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
5250 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
5251 (nvptx_output_mov_insn): Handle lack of mov.b128.
5252 (nvptx_print_operand): Handle 'H' and 'L' codes.
5253 (nvptx_vector_mode_supported): Allow V2DImode.
5254 (nvptx_preferred_simd_mode): New function.
5255 (nvptx_data_alignment): New function.
5256 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
5257 nvptx_preferred_simd_mode.
5258 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
5260 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
5262 2017-07-19 Tom de Vries <tom@codesourcery.com>
5264 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
5265 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
5266 (nvptx_vector_mode_supported): New function. Allow V2SImode.
5267 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
5268 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
5269 (mov<VECIM>_insn): New define_insn.
5270 (define_expand "mov<VECIM>): New define_expand.
5272 2017-07-19 Tom de Vries <tom@codesourcery.com>
5274 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
5276 2017-07-19 Jakub Jelinek <jakub@redhat.com>
5278 PR tree-optimization/81346
5279 * fold-const.h (fold_div_compare, range_check_type): Declare.
5280 * fold-const.c (range_check_type): New function.
5281 (build_range_check): Use range_check_type.
5282 (fold_div_compare): No longer static, rewritten into
5283 a match.pd helper function.
5284 (fold_comparison): Don't call fold_div_compare here.
5285 * match.pd (X / C1 op C2): New optimization using fold_div_compare
5288 2017-07-19 Nathan Sidwell <nathan@acm.org>
5290 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
5291 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
5292 * tree.c (find_decls_types_r, verify_type): Use
5293 TYPE_{MIN,MAX}_VALUE_RAW.
5294 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
5295 (hash_tree): Likewise.
5296 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5298 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5301 2017-07-18 Tom de Vries <tom@codesourcery.com>
5304 * omp-expand.c (expand_omp_for_static_chunk): Handle
5305 equal-argument loop exit phi.
5307 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
5310 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
5311 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
5312 operand 2 predicate.
5313 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
5314 operand 2 predicate.
5315 (ror,rol -> rorx splitters): Use const_int_operand as
5316 operand 2 predicate.
5318 2017-06-18 Richard Biener <rguenther@suse.de>
5320 PR tree-optimization/81410
5321 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
5322 the gap in the ! slp_perm SLP case after each group.
5324 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5327 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
5330 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5333 * predict.c (set_even_probabilities): Cleanup; do not affect
5334 probabilities that are already known.
5335 (combine_predictions_for_bb): Call even when count is set.
5337 2017-07-18 Nathan Sidwell <nathan@acm.org>
5339 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
5342 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5345 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
5346 optimization for loop niter analysis.
5348 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
5351 * config/avr/avr.c (avr_optimize_casesi): Don't use
5352 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
5354 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5356 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
5357 body_cost_vec from _vect_peel_extended_info.
5358 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
5359 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
5362 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5364 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
5366 2017-07-18 Richard Biener <rguenther@suse.de>
5368 PR tree-optimization/80620
5369 PR tree-optimization/81403
5370 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
5371 info when re-using a VN table entry.
5373 2017-07-18 Richard Biener <rguenther@suse.de>
5375 PR tree-optimization/81418
5376 * tree-vect-loop.c (vectorizable_reduction): Properly compute
5377 vectype_in. Verify that with lane-reducing reduction operations
5378 we have a single def-use cycle.
5380 2017-07-17 Carl Love <cel@us.ibm.com>
5382 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
5384 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5385 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5386 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5387 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5388 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5389 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5390 VMULOSW): New enum "unspec" values.
5391 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
5392 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
5393 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5394 altivec_vmulosw): New patterns.
5395 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5396 VMULOSW): Add definitions.
5397 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
5399 * config/alpha/alpha.c: Include predict.h.
5401 2017-07-17 Yury Gribov <tetra2005@gmail.com>
5403 * tree-vrp.c (compare_assert_loc): Fix comparison function
5404 to return predictable results.
5406 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5408 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
5411 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
5412 * doc/invoke.texi (mexpand-adddi): Update text.
5414 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5416 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
5417 that also clobbers the CC register. The old expand code is moved
5419 (*arc_clzsi2): ... here.
5420 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
5421 the CC register. The old expand code is moved to ...
5422 (arc_ctzsi2): ... here.
5424 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5426 * config/arc/arc.opt (mindexed-loads): Use initial value
5427 TARGET_INDEXED_LOADS_DEFAULT.
5428 (mauto-modify-reg): Use initial value
5429 TARGET_AUTO_MODIFY_REG_DEFAULT.
5430 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5431 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5432 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5433 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5435 2017-07-17 Martin Liska <mliska@suse.cz>
5438 * opts.c (finish_options): Do not allow -fgnu-tm
5439 w/ -fsanitize={kernel-,}address. Say sorry.
5441 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5444 * tree-loop-distribution.c (classify_partition): Only assert on
5445 numer of iterations.
5446 (merge_dep_scc_partitions): Delete prameter. Update function call.
5447 (distribute_loop): Remove code handling loop with unknown niters.
5448 (pass_loop_distribution::execute): Skip loop with unknown niters.
5450 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5453 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
5454 function sort_partitions_by_post_order.
5456 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5458 PR tree-optimization/81374
5459 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
5460 the max index of basic blocks, rather than number of basic blocks.
5462 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5464 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
5466 (arc_legitimate_pic_operand_p): Likewise.
5467 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
5469 (arc_needs_pcl_p): Likewise.
5470 (arc_legitimate_pc_offset_p): Likewise.
5471 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
5472 function is also used in constrains.md.
5473 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
5474 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
5475 PLUS. Only return true/false in known cases, otherwise assert.
5476 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
5477 is already called in arc_legitimate_constant_p.
5478 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
5480 (LEGITIMATE_PIC_OPERAND_P): Use
5481 arc_raw_symbolic_reference_mentioned_p function.
5482 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
5487 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5488 Andrew Burgess <andrew.burgess@embecosm.com>
5490 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
5491 (arc_return_address_register): New function.
5492 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
5493 (arc_handle_fndecl_attribute): Add naked attribute.
5494 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5495 (TARGET_WARN_FUNC_RETURN): Likewise.
5496 (arc_allocate_stack_slots_for_args): New function.
5497 (arc_warn_func_return): Likewise.
5498 (machine_function): Change type fn_type.
5499 (arc_compute_function_type): Consider new naked function type,
5500 change function return type.
5501 (arc_must_save_register): Adapt to handle new
5502 arc_compute_function_type's return type.
5503 (arc_expand_prologue): Likewise.
5504 (arc_expand_epilogue): Likewise.
5505 (arc_return_address_regs): Delete.
5506 (arc_return_address_register): New function.
5507 (arc_epilogue_uses): Use above function.
5508 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
5509 (arc_function_type): Change encoding, add naked type.
5510 (ARC_INTERRUPT_P): Change to handle the new encoding.
5511 (ARC_FAST_INTERRUPT_P): Likewise.
5512 (ARC_NORMAL_P): Define.
5513 (ARC_NAKED_P): Likewise.
5514 (arc_compute_function_type): Delete prototype.
5515 * config/arc/arc.md (in_ret_delay_slot): Use
5516 arc_return_address_register function.
5517 (simple_return): Likewise.
5518 (p_return_i): Likewise.
5520 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5522 PR tree-optimization/81428
5523 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
5524 can't be built for those types.
5526 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
5528 Remove stuff dead since r239246.
5530 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
5531 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
5532 (avr_inform_devices): Remove dead stuff.
5534 2017-07-17 Tamar Christina <tamar.christina@arm.com>
5536 * config/arm/arm_neon.h: Fix softp typo.
5538 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5540 PR tree-optimization/81365
5541 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
5542 aggregate moves onto bb predecessor edges, make sure there are no
5543 loads that could alias the lhs in between the start of bb and the
5546 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
5549 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
5550 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
5551 [LSHIFTRT, outer_code = TRUNCATE]: Same.
5553 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5555 PR tree-optimization/81396
5556 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
5557 (init_symbolic_number): Initialize it to 1.
5558 (perform_symbolic_merge): Add n_ops from both operands into the new
5560 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
5561 without base_addr as useless if they need more than one operation.
5562 (bswap_replace): Handle !bswap case for NULL base_addr.
5564 2017-07-17 Tom de Vries <tom@codesourcery.com>
5567 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
5570 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
5572 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
5573 conditional builtin define __FIX_LEON3FT_B2BST.
5575 2017-07-17 Daniel Cederman <cederman@gaisler.com>
5577 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
5578 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
5581 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
5583 PR rtl-optimization/81424
5584 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
5585 to remove potential trapping from operands if -fnon-call-exceptions.
5587 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5589 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
5590 profile_proability for scalling.
5591 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
5593 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5595 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
5597 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5599 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
5600 fixpoint arithmetics.
5602 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5604 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
5605 fixpoint arithmetics.
5607 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5609 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
5610 fixpoint arithmetics.
5612 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
5614 * profile-count.h (profile_probability::from_reg_br_prob_note,
5615 profile_probability::to_reg_br_prob_note): New functions.
5616 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
5617 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
5618 * predict.c (probability_reliable_p): Update.
5619 (edge_probability_reliable_p): Update.
5620 (br_prob_note_reliable_p): Update.
5621 (invert_br_probabilities): Update.
5622 (add_reg_br_prob_note): New function.
5623 (combine_predictions_for_insn): Update.
5624 * asan.c (asan_clear_shadow): Update.
5625 * cfgbuild.c (compute_outgoing_frequencies): Update.
5626 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
5627 (update_br_prob_note): Update.
5628 (rtl_verify_edges): Update.
5629 (purge_dead_edges): Update.
5630 (fixup_reorder_chain): Update.
5631 * emit-rtl.c (try_split): Update.
5632 * ifcvt.c (cond_exec_process_insns): Update.
5633 (cond_exec_process_if_block): Update.
5634 (dead_or_predicable): Update.
5635 * internal-fn.c (expand_addsub_overflow): Update.
5636 (expand_neg_overflow): Update.
5637 (expand_mul_overflow): Update.
5638 * loop-doloop.c (doloop_modify): Update.
5639 * loop-unroll.c (compare_and_jump_seq): Update.
5640 * optabs.c (emit_cmp_and_jump_insn_1): Update.
5641 * predict.h: Update.
5642 * reorg.c (mostly_true_jump): Update.
5644 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
5645 * config/alpha/alpha.c (emit_unlikely_jump): Update.
5646 * config/arc/arc.c: (emit_unlikely_jump): Update.
5647 * config/arm/arm.c: (emit_unlikely_jump): Update.
5648 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
5649 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
5650 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
5651 (ix86_print_operand): Update.
5652 (ix86_split_fp_branch): Update.
5653 (predict_jump): Update.
5654 * config/ia64/ia64.c (ia64_print_operand): Update.
5655 * config/mmix/mmix.c (mmix_print_operand): Update.
5656 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
5657 (rs6000_expand_split_stack_prologue): Update.
5658 * config/rs6000/rs6000.c: Update.
5659 * config/s390/s390.c (s390_expand_vec_strlen): Update.
5660 (s390_expand_vec_movstr): Update.
5661 (s390_expand_cs_tdsi): Update.
5662 (s390_expand_split_stack_prologue): Update.
5663 * config/sh/sh.c (sh_print_operand): Update.
5664 (expand_cbranchsi4): Update.
5665 (expand_cbranchdi4): Update.
5666 * config/sparc/sparc.c (output_v9branch): Update.
5667 * config/spu/spu.c (get_branch_target): Update.
5668 (ea_load_store_inline): Update.
5669 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
5670 * config/tilepro/tilepro.c: Update.
5672 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
5674 * gimplify.c (mostly_copy_tree_r): Revert latest change.
5675 (gimplify_save_expr): Likewise.
5677 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5679 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
5681 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5683 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
5685 * timevar.def (TV_IPA_FNSUMMARY): Define.
5687 2017-07-16 Daniel Cederman <cederman@gaisler.com>
5689 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
5690 to back store errata sensitive sequence from being generated.
5691 (sqrtdf2_fix): Likewise.
5693 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
5695 * tree-ssa-threadupdate.c (compute_path_counts,
5696 update_joiner_offpath_counts): Use profile_probability.
5698 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
5701 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5703 * config/arm/arm-c.c (arm_cpu_builtins): Define
5704 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
5706 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
5708 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5709 array entries to represent __ieee128 versions of the
5710 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
5711 scalar_extract_sig, and scalar_insert_exp built-in functions.
5712 (altivec_resolve_overloaded_builtin): Add special case handling
5713 for the __builtin_scalar_insert_exp function, as represented by
5714 the P9V_BUILTIN_VEC_VSIEDP constant.
5715 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
5716 exponent support for __ieee128 argument.
5717 (VSESQP): Add scalar extract signature support for __ieee128
5719 (VSTDCNQP): Add scalar test negative support for __ieee128
5721 (VSIEQP): Add scalar insert exponent support for __int128 argument
5722 with __ieee128 result.
5723 (VSIEQPF): Add scalar insert exponent support for __ieee128
5724 argument with __ieee128 result.
5725 (VSTDCQP): Add scalar test data class support for __ieee128
5727 (VSTDCNQP): Add overload support for scalar test negative with
5729 (VSTDCQP): Add overload support for scalar test data class
5731 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
5733 (UNSPEC_VSX_SIEXPQP): New constant.
5734 (xsxexpqp): New insn for VSX scalar extract exponent quad
5736 (xsxsigqp): New insn for VSX scalar extract significand quad
5738 (xsiexpqpf): New insn for VSX scalar insert exponent quad
5739 precision with floating point argument.
5740 (xststdcqp): New expand for VSX scalar test data class quad
5742 (xststdcnegqp): New expand for VSX scalar test negative quad
5744 (xststdcqp): New insn to match expansions for VSX scalar test data
5745 class quad precision and VSX scalar test negative quad precision.
5746 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
5747 special case operand checking to enforce that second operand of
5748 VSX scalar test data class with quad precision argument is a 7-bit
5750 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
5751 prototypes and descriptions of __ieee128 versions of
5752 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
5753 scalar_test_data_class, and scalar_test_neg built-in functions.
5755 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5757 PR tree-optimization/81162
5758 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
5759 replace a negate with an add.
5761 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
5763 * doc/invoke.texi (arm/-mcpu): Document +crypto.
5765 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5767 * config/arm/arm-c.c (arm_cpu_builtins): Define
5768 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
5770 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5772 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
5773 (armv8-r): Set ARM Cortex-R52 as default CPU.
5774 * config/arm/arm-tables.opt: Regenerate.
5775 * config/arm/arm-tune.md: Regenerate.
5776 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
5778 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
5779 extension for -mcpu=cortex-r52.
5781 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
5783 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
5784 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
5785 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
5786 (fp-armv8): Define it as FP_ARMv8 only.
5787 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
5788 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
5790 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
5791 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
5792 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
5793 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
5794 than TARGET_FPU_ARMV8.
5795 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
5796 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
5797 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
5799 * config/arm/neon.md (neon_vrint): Likewise.
5800 (neon_vcvt): Likewise.
5801 (neon_<fmaxmin_op><mode>): Likewise.
5802 (<fmaxmin><mode>3): Likewise.
5803 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
5804 * config/arm/predicates.md (arm_cond_move_operator): Check against
5805 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
5807 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
5809 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
5812 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5814 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
5815 loop in comment with memset.
5817 2017-07-14 Martin Liska <mliska@suse.cz>
5819 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
5820 * dwarf2out.c (is_java): Remove the function.
5821 (output_pubname): Remove usage of the function.
5822 (lower_bound_default): Remove usage of DW_LANG_Java.
5823 (gen_compile_unit_die): Likewise.
5824 * gcc.c: Remove compiler defaults for .java and .zip files.
5825 * gimple-expr.c (remove_suffix): Change as there's no longer
5826 extension than 4-letter one.
5827 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
5828 (gimplify_save_expr): Likewise.
5829 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
5830 as it's possible even for other languages than Java.
5831 * langhooks.h (struct lang_hooks): Remove Java from a comment.
5832 * lto-opts.c (lto_write_options): Remove reference to Java.
5833 * opts.c (strip_off_ending): Update file extension handling.
5834 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
5835 * tree-eh.c (lower_resx): Likewise.
5836 * tree.c (free_lang_data_in_type): Remove dead code.
5837 (find_decls_types_r): Likewise.
5838 (build_common_builtin_nodes): Remove Java from a comment.
5839 (verify_type): Remove dead code.
5840 * varasm.c (assemble_external): Remove Java from a comment.
5842 2017-07-14 Martin Liska <mliska@suse.cz>
5844 * opts.c (finish_options): Add quotes.
5845 (common_handle_option): Likewise.
5847 2017-07-14 Martin Liska <mliska@suse.cz>
5849 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
5850 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
5852 * dwarf2out.c (lower_bound_default): Do not handle
5854 (gen_compile_unit_die): Likewise.
5855 * gcc.c: Remove default extension binding for GNU Pascal.
5856 * stmt.c: Remove Pascal language from a comment.
5857 * xcoffout.c: Likewise.
5859 2017-07-13 David Malcolm <dmalcolm@redhat.com>
5862 * diagnostic-show-locus.c (fixit_cmp): New function.
5863 (layout::layout): Sort m_fixit_hints.
5864 (column_range::column_range): Assert that the values are valid.
5865 (struct char_span): New struct.
5866 (correction::overwrite): New method.
5867 (struct source_line): New struct.
5868 (line_corrections::add_hint): Add assertions. Reimplement memcpy
5869 calls in terms of classes source_line and char_span, and
5870 correction::overwrite.
5871 (selftest::test_overlapped_fixit_printing_2): New function.
5872 (selftest::diagnostic_show_locus_c_tests): Call it.
5874 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
5876 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
5877 early if there is no lhs.
5879 2017-07-13 Martin Liska <mliska@suse.cz>
5881 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
5882 (gen_reference_type_die): Likewise.
5883 * stor-layout.c: Remove Pascal-related comment.
5885 2017-07-13 Martin Liska <mliska@suse.cz>
5887 * opts.c (finish_options): Add quotes to error messages.
5888 (parse_sanitizer_options): Likewise.
5890 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
5892 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
5894 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
5896 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
5898 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
5900 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
5902 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
5904 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5907 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
5908 provides the hardware capability bits, define the macro
5909 __BUILTIN_CPU_SUPPORTS__.
5910 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
5911 if GLIBC does not provide the hardware capability bits. Add a
5912 gcc_unreachable call if the built-in cpu function is neither
5913 __builtin_cpu_is nor __builtin_cpu_supports.
5914 (rs6000_get_function_versions_dispatcher): Change the warning
5915 that an old GLIBC is used which does not export the capability
5916 bits to be an error.
5917 * doc/extend.texi (target_clones attribute): Document the
5918 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
5919 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
5920 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
5921 the macros defined by GCC if the newer GLIBC is available.
5923 2017-07-12 Jeff Law <law@redhat.com>
5925 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
5926 remaining includes slightly.
5927 * config/riscv/riscv-builtins.c: Include profile-count.h.
5929 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
5932 * config/avr/avr.c (avr_set_current_function): In diagnostic
5933 messages: Quote keywords and (parts of) identifiers.
5934 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
5937 2017-07-12 Carl Love <cel@us.ibm.com>
5939 * config/rs6000/rs6000-c.c: Add support for built-in functions
5940 vector bool char vec_revb (vector bool char);
5941 vector bool short vec_revb (vector short char);
5942 vector bool int vec_revb (vector bool int);
5943 vector bool long long vec_revb (vector bool long long);
5944 * doc/extend.texi: Update the built-in documentation file for the
5945 new built-in functions.
5947 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5949 * config/s390/s390.md: Remove movcc splitter.
5951 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5953 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
5954 load/store on condition.
5956 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
5959 * config/avr/avr.c (avr_encode_section_info)
5960 [progmem && !TREE_READONLY]: Error if progmem object needs
5963 2017-07-11 Michael Collison <michael.collison@arm.com>
5965 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
5968 2017-07-11 Carl Love <cel@us.ibm.com>
5970 * config/rs6000/rs6000-c.c: Add support for builtins
5971 vector unsigned int vec_parity_lsbb (vector signed int);
5972 vector unsigned int vec_parity_lsbb (vector unsigned int);
5973 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
5974 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
5975 vector unsigned long long vec_parity_lsbb (vector signed long long);
5976 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
5977 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
5978 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
5979 * doc/extend.texi: Update the built-in documentation file for the
5980 new built-in functions.
5982 2017-07-11 David Malcolm <dmalcolm@redhat.com>
5984 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
5985 (layout::m_primary_loc): New field.
5986 (layout::layout): Initialize new field. Move location filtering
5987 logic from here to...
5988 (layout::maybe_add_location_range): ...this new method. Add
5989 support for filtering to just the lines already specified by other
5991 (layout::will_show_line_p): New method.
5992 (gcc_rich_location::add_location_if_nearby): New method.
5993 (selftest::test_add_location_if_nearby): New test function.
5994 (selftest::diagnostic_show_locus_c_tests): Call it.
5995 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
5998 2017-07-11 Tom de Vries <tom@codesourcery.com>
6000 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
6001 (bb_first_real_insn): New function.
6002 (nvptx_single): Add extra initialization of broadcasted condition
6005 2017-07-11 Nathan Sidwell <nathan@acm.org>
6007 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
6009 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
6011 * doc/extend.texi (AVR Function Attributes): Remove weblink to
6012 Binutils doc as TEXI will mess them up.
6013 * doc/invoke.texi (AVR Options): Same here.
6015 2017-07-11 Daniel Cederman <cederman@gaisler.com>
6017 * config/sparc/sparc.opt (mfix-ut700): New option.
6018 (mfix-gr712rc): Likewise.
6019 (sparc_fix_b2bst): New variable.
6020 * doc/invoke.texi (SPARC options): Document them.
6021 (ARM options): Fix warnings.
6022 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
6023 instructions to prevent sequences that can trigger the store-store
6024 errata for certain LEON3FT processors.
6025 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
6026 (sparc_option_override): Set sparc_fix_b2bst appropriately.
6027 * config/sparc/sparc.md (fix_b2bst): New attribute.
6028 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
6030 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
6033 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
6035 (*rsqrtsf2_sse): Ditto.
6037 (div<mode>3): Macroize insn from divdf3 and divsf3
6038 using MODEF mode iterator.
6040 2017-07-10 Martin Sebor <msebor@redhat.com>
6042 PR tree-optimization/80397
6043 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
6044 instead of testing for equality to INTEGER_TYPE.
6046 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
6048 * config.gcc: Remove uclibc from arc target spec.
6050 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
6052 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
6054 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6057 * lto-wrapper.c (remove_option): New function.
6058 (merge_and_complain): Merge PIC/PIE options more realistically.
6060 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
6062 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
6066 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
6067 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
6068 * config.in: Regenerate.
6069 * configure: Regenerate.
6070 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
6071 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
6072 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
6073 (TARGET_GASISR_PROLOGUES): ...target mask.
6074 * common/config/avr/avr-common.c
6075 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
6076 Set -mgas-isr-prologues.
6077 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
6078 INSERT_PASS_BEFORE for it.
6079 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
6080 * config/avr/avr.c (avr_option_override)
6081 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
6082 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
6083 (avr_attribute_table) <no_gccisr>: Add new function attribute.
6084 (avr_set_current_function) <is_no_gccisr>: Init machine field.
6085 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
6087 (make_avr_pass_pre_proep): New function.
6088 (emit_push_sfr) <treg>: Add argument to function and use it
6090 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
6091 and set machine->gasisr.yes.
6092 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
6093 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
6094 __gcc_isr.n_pushed to .L__stack_usage.
6095 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
6096 (avr_asm_final_postscan_insn): ...this new static function.
6097 * config/avr/avr.h (machine_function)
6098 <is_no_gccisr, use_L__stack_usage>: New fields.
6099 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
6100 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
6101 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
6102 (gasisr, *gasisr): New expander and insn.
6103 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
6104 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
6105 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
6107 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
6109 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
6112 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
6114 Move jump-tables out of .text again.
6117 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
6118 (ASM_OUTPUT_ADDR_VEC): New function.
6119 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
6120 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
6121 INSN_ADDRESSes as asm comment.
6122 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
6123 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
6124 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
6125 * config/avr/avr.md (*tablejump): Adjust comment.
6126 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
6127 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
6129 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
6130 (avr_output_addr_vec): New proto.
6131 (avr_log_t) <insn_addresses>: New field.
6133 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
6136 * config/i386/i386.c (ix86_function_arg_advance): Set
6137 outgoing_args_on_stack to true if there are outgoing arguments
6139 (ix86_function_arg): Likewise.
6140 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
6141 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
6142 * config/i386/i386.h (machine_function): Add
6143 outgoing_args_on_stack.
6145 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
6147 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
6149 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
6151 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
6153 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
6154 (REAL_H): Remove $(MACHMODE_H).
6155 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
6157 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
6158 $(MACHMODE_H) and double-int.h.
6159 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
6161 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
6164 2017-07-07 Andrew Pinski <apinski@cavium.com>
6166 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
6167 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
6169 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6171 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
6172 Add warning if GCC was not configured to link against a GLIBC that
6173 exports the hardware capability bits.
6174 (make_resolver_func): Make resolver function private and not a
6175 COMDAT function. Create the name with clone_function_name instead
6176 of make_unique_name.
6179 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
6180 correct operand in doing the split.
6182 2017-07-07 Carl Love <cel@us.ibm.com>
6184 * config/rs6000/rs6000-c: Add support for built-in function
6185 vector unsigned short vec_pack_to_short_fp32 (vector float,
6187 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
6188 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
6189 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
6190 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
6191 (convert_4f32_8i16): Add define_expand.
6192 * doc/extend.texi: Update the built-in documentation file for the
6193 new built-in function.
6195 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6197 * config/sparc/m8.md: New file.
6198 * config/sparc/sparc.md: Include m8.md.
6200 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6202 * config/sparc/sparc.opt: New option -mvis4b.
6203 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
6204 (sparc_option_override): Handle VIS4B.
6205 (enum sparc_builtins): Define
6206 SPARC_BUILTIN_DICTUNPACK{8,16,32},
6207 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
6208 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
6209 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
6210 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
6211 (check_constant_argument): New function.
6212 (sparc_vis_init_builtins): Define builtins
6213 __builtin_vis_dictunpack{8,16,32},
6214 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
6215 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
6216 __builtin_vis_fpcmpde{8,16,32}shl and
6217 __builtin_vis_fpcmpur{8,16,32}shl.
6218 (sparc_expand_builtin): Check that the constant operands to
6219 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
6220 constant and in range.
6221 * config/sparc/sparc-c.c (sparc_target_macros): Handle
6223 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
6224 (SPARC_IMM5_P): Likewise.
6225 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
6226 (enabled): Handle vis4b.
6227 (UNSPEC_DICTUNPACK): New unspec.
6228 (UNSPEC_FPCMPSHL): Likewise.
6229 (UNSPEC_FPUCMPSHL): Likewise.
6230 (UNSPEC_FPCMPDESHL): Likewise.
6231 (UNSPEC_FPCMPURSHL): Likewise.
6232 (cpu_feature): New CPU feature `vis4b'.
6233 (dictunpack{8,16,32}): New insns.
6234 (FPCSMODE): New mode iterator.
6235 (fpcscond): New code iterator.
6236 (fpcsucond): Likewise.
6237 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
6238 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
6239 (fpcmpde{8,16,32}{si,di}shl): Likewise.
6240 (fpcmpur{8,16,32}{si,di}shl): Likewise.
6241 * config/sparc/constraints.md: Define constraints `q' for unsigned
6242 2-bit integer constants and `t' for unsigned 5-bit integer
6244 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
6246 (imm5_operand_dictunpack16): Likewise.
6247 (imm5_operand_dictunpack32): Likewise.
6248 (imm2_operand): Likewise.
6249 * doc/invoke.texi (SPARC Options): Document -mvis4b.
6250 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
6251 ditunpack* and fpcmp*shl builtins.
6253 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6255 * config.gcc: Handle m8 in --with-{cpu,tune} options.
6256 * config.in: Add HAVE_AS_SPARC6 define.
6257 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
6259 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
6261 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
6262 (CPP_CPU_SPEC): Handle m8.
6263 (ASM_CPU_SPEC): Likewise.
6264 * config/sparc/sparc-opts.h (enum processor_type): Add
6266 * config/sparc/sparc.c (m8_costs): New struct.
6267 (sparc_option_override): Handle TARGET_CPU_m8.
6268 (sparc32_initialize_trampoline): Likewise.
6269 (sparc64_initialize_trampoline): Likewise.
6270 (sparc_issue_rate): Likewise.
6271 (sparc_register_move_cost): Likewise.
6272 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
6273 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
6274 (ASM_CPU64_DEFAULT_SPEC): Likewise.
6275 (CPP_CPU_SPEC): Handle M8.
6276 (ASM_CPU_SPEC): Likewise.
6277 (AS_M8_FLAG): Define.
6278 * config/sparc/sparc.md: Add m8 to the cpu attribute.
6279 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
6280 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
6282 * configure: Regenerate.
6283 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
6286 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6288 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
6290 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
6291 ("*movdi_insn_sp32"): Do not set v3pipe.
6292 ("*movsi_insn"): Likewise.
6293 ("*movdi_insn_sp64"): Likewise.
6294 ("*movsf_insn"): Likewise.
6295 ("*movdf_insn_sp32"): Likewise.
6296 ("*movdf_insn_sp64"): Likewise.
6297 ("*zero_extendsidi2_insn_sp64"): Likewise.
6298 ("*sign_extendsidi2_insn"): Likewise.
6299 ("*mov<VM32:mode>_insn"): Likewise.
6300 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6301 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6302 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6303 ("<vlop:code><VL:mode>3"): Likewise.
6304 ("*not_<vlop:code><VL:mode>3"): Likewise.
6305 ("*nand<VL:mode>_vis"): Likewise.
6306 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
6307 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
6308 ("one_cmpl<VL:mode>2"): Likewise.
6309 ("faligndata<VM64:mode>_vis"): Likewise.
6310 ("alignaddrsi_vis"): Likewise.
6311 ("alignaddrdi_vis"): Likweise.
6312 ("alignaddrlsi_vis"): Likewise.
6313 ("alignaddrldi_vis"): Likewise.
6314 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6315 ("bmaskdi_vis"): Likewise.
6316 ("bmasksi_vis"): Likewise.
6317 ("bshuffle<VM64:mode>_vis"): Likewise.
6318 ("cmask8<P:mode>_vis"): Likewise.
6319 ("cmask16<P:mode>_vis"): Likewise.
6320 ("cmask32<P:mode>_vis"): Likewise.
6321 ("pdistn<P:mode>_vis"): Likewise.
6322 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6324 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6326 * config/sparc/sparc.md ("subtype"): New insn attribute.
6327 ("*wrgsr_sp64"): Set insn subtype.
6328 ("*rdgsr_sp64"): Likewise.
6329 ("alignaddrsi_vis"): Likewise.
6330 ("alignaddrdi_vis"): Likewise.
6331 ("alignaddrlsi_vis"): Likewise.
6332 ("alignaddrldi_vis"): Likewise.
6333 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6334 ("fexpand_vis"): Likewise.
6335 ("fpmerge_vis"): Likewise.
6336 ("faligndata<VM64:mode>_vis"): Likewise.
6337 ("bshuffle<VM64:mode>_vis"): Likewise.
6338 ("cmask8<P:mode>_vis"): Likewise.
6339 ("cmask16<P:mode>_vis"): Likewise.
6340 ("cmask32<P:mode>_vis"): Likewise.
6341 ("fchksm16_vis"): Likewise.
6342 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
6343 ("fmean16_vis"): Likewise.
6344 ("fp<plusminus_insn>64_vis"): Likewise.
6345 ("<plusminus_insn>v8qi3"): Likewise.
6346 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6347 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
6348 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
6349 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
6350 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
6351 ("*movqi_insn"): Likewise.
6352 ("*movhi_insn"): Likewise.
6353 ("*movsi_insn"): Likewise.
6354 ("movsi_pic_gotdata_op"): Likewise.
6355 ("*movdi_insn_sp32"): Likewise.
6356 ("*movdi_insn_sp64"): Likewise.
6357 ("movdi_pic_gotdata_op"): Likewise.
6358 ("*movsf_insn"): Likewise.
6359 ("*movdf_insn_sp32"): Likewise.
6360 ("*movdf_insn_sp64"): Likewise.
6361 ("*zero_extendhisi2_insn"): Likewise.
6362 ("*zero_extendqihi2_insn"): Likewise.
6363 ("*zero_extendqisi2_insn"): Likewise.
6364 ("*zero_extendqidi2_insn"): Likewise.
6365 ("*zero_extendhidi2_insn"): Likewise.
6366 ("*zero_extendsidi2_insn_sp64"): Likewise.
6367 ("ldfsr"): Likewise.
6368 ("prefetch_64"): Likewise.
6369 ("prefetch_32"): Likewise.
6370 ("tie_ld32"): Likewise.
6371 ("tie_ld64"): Likewise.
6372 ("*tldo_ldub_sp32"): Likewise.
6373 ("*tldo_ldub1_sp32"): Likewise.
6374 ("*tldo_ldub2_sp32"): Likewise.
6375 ("*tldo_ldub_sp64"): Likewise.
6376 ("*tldo_ldub1_sp64"): Likewise.
6377 ("*tldo_ldub2_sp64"): Likewise.
6378 ("*tldo_ldub3_sp64"): Likewise.
6379 ("*tldo_lduh_sp32"): Likewise.
6380 ("*tldo_lduh1_sp32"): Likewise.
6381 ("*tldo_lduh_sp64"): Likewise.
6382 ("*tldo_lduh1_sp64"): Likewise.
6383 ("*tldo_lduh2_sp64"): Likewise.
6384 ("*tldo_lduw_sp32"): Likewise.
6385 ("*tldo_lduw_sp64"): Likewise.
6386 ("*tldo_lduw1_sp64"): Likewise.
6387 ("*tldo_ldx_sp64"): Likewise.
6388 ("*mov<VM32:mode>_insn"): Likewise.
6389 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6390 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6392 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6394 * config/sparc/sparc.md ("type"): New insn type viscmp.
6395 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
6397 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
6398 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
6399 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6400 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
6402 ("n7_vis_logical_11cycle"): Likewise.
6403 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
6404 * config/sparc/niagara2.md ("niag3_vis": Likewise.
6405 * config/sparc/niagara.md ("niag_vis"): Likewise.
6406 * config/sparc/ultra3.md ("us3_fga"): Likewise.
6407 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
6409 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6411 * config/sparc/sparc.md: New instruction type `bmask'.
6412 (bmaskdi_vis): Use the `bmask' type.
6413 (bmasksi_vis): Likewise.
6414 * config/sparc/ultra3.md (us3_array): Likewise.
6415 * config/sparc/niagara7.md (n7_array): Likewise.
6416 * config/sparc/niagara4.md (n4_array): Likewise.
6417 * config/sparc/niagara2.md (niag2_vis): Likewise.
6418 (niag3_vis): Likewise.
6419 * config/sparc/niagara.md (niag_vis): Likewise.
6421 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6423 * ipa-comdats.c: Remove optimize check from gate.
6424 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
6425 for functions not optimized.
6426 (ipa_fn_summary_read): Skip optimize check.
6427 (ipa_fn_summary_write): Likewise.
6428 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
6430 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
6432 (can_inline_edge_p): Check flag_pcc_struct_return for match.
6433 (check_callers): Give up on caller which is not optimized.
6434 (inline_small_functions): Likewise.
6435 (ipa_inline): Do not give up when not optimizing.
6436 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
6437 away unoptimizes cdtors.
6438 (whole_program_function_and_variable_visibility): Do
6439 ipa_discover_readonly_nonaddressable_vars in LTO mode.
6440 * ipa.c (process_references): Do not check optimize.
6441 (symbol_table::remove_unreachable_nodes): Update optimize check.
6442 (set_writeonly_bit): Update optimize check.
6443 (pass_ipa_cdtor_merge::gate): Do not check optimize.
6444 (pass_ipa_single_use::gate): Remove.
6446 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
6448 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
6449 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
6450 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
6451 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
6452 permute_load, permute_store, adjust_extract, adjust_splat,
6453 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
6454 replace_swap_with_copy, dump_swap_insn_table,
6455 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
6456 recombine_lvx_pattern, recombine_stvx_pattern,
6457 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
6458 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
6459 to file rs6000-p8swap.c.
6460 * config/rs6000/rs6000-p8swap.c: New file.
6461 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
6462 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
6463 and rs6000*-*-* targets.
6465 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6467 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
6469 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6471 * lto-wrapper.c (merge_and_complain): Do not merge
6472 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
6473 fsigned_zeros, ftrapping_math, fwrapv.
6474 (append_compiler_options): Do not track these options.
6475 (append_linker_options): Likewie
6477 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6479 * cgraphunit.c (cgraph_node::finalize_function): When
6480 !flag_toplevel_reorde set no_reorder flag.
6481 (varpool_node::finalize_decl): Likewise.
6482 (symbol_table::compile): Drop no toplevel reorder path.
6484 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6486 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
6487 edges; zero probability is not better than uninitialized.
6489 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6491 * asan.h (asan_sanitize_allocas_p): Declare.
6492 * asan.c (asan_sanitize_allocas_p): New function.
6493 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
6494 (handle_builtin_alloca): Likewise.
6495 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
6496 if !asan_sanitize_allocas_p.
6497 * params.def (asan-instrument-allocas): Add new option.
6498 * params.h (ASAN_PROTECT_ALLOCAS): Define.
6499 * opts.c (common_handle_option): Disable allocas sanitization for
6502 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6504 * asan.c: Include gimple-fold.h.
6505 (get_last_alloca_addr): New function.
6506 (handle_builtin_stackrestore): Likewise.
6507 (handle_builtin_alloca): Likewise.
6508 (asan_emit_allocas_unpoison): Likewise.
6509 (get_mem_refs_of_builtin_call): Add new parameter, remove const
6510 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
6511 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
6512 (instrument_builtin_call): Pass gimple iterator to
6513 get_mem_refs_of_builtin_call.
6514 (last_alloca_addr): New global.
6515 * asan.h (asan_emit_allocas_unpoison): Declare.
6516 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
6517 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
6518 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
6519 if function calls alloca.
6520 * gimple-fold.c (replace_call_with_value): Remove static keyword.
6521 * gimple-fold.h (replace_call_with_value): Declare.
6522 * internal-fn.c: Include asan.h.
6523 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
6524 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
6526 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6528 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
6529 (C_SELFTEST_FLAGS): New.
6530 (CPP_SELFTEST_FLAGS): New.
6531 (SELFTEST_DEPS): New, from deps of s-selftest.
6532 (C_SELFTEST_DEPS): New, from deps of s-selftest.
6533 (CPP_SELFTEST_DEPS): New.
6534 (selftest): Add dependency on s-selftest-c++.
6535 (s-selftest): Rename to...
6536 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
6537 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
6538 than SELFTEST_FLAGS.
6539 (selftest-gdb): Rename to...
6540 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
6542 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
6543 (selftest-valgrind): Rename to...
6544 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
6546 (selftest-valgrind): Reintroduce as an alias for
6547 selftest-c-valgrind.
6548 (s-selftest-c++): New.
6549 (selftest-c++-gdb): New.
6550 (selftest-c++-valgrind): New.
6552 2017-07-06 Olivier Hainque <hainque@adacore.com>
6554 * gcc.c (process_command): When deciding if undefined variables
6555 should be ignored when processing specs, accept "gcc -v" as well.
6557 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6559 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
6560 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
6562 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6564 * config/arm/arm-cpus.in (armv8-r): Add new entry.
6565 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
6566 * config/arm/arm-tables.opt: Regenerate.
6567 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
6569 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
6571 2017-07-06 Carl Love <cel@us.ibm.com>
6573 * ChangeLog: Clean up from mid air collision
6575 2017-07-06 Carl Love <cel@us.ibm.com>
6577 * config/rs6000/rs6000-c.c: Add support for built-in functions
6578 vector signed int vec_subc (vector signed int, vector signed int);
6579 vector signed __int128 vec_subc (vector signed __int128,
6580 vector signed __int128);
6581 vector unsigned __int128 vec_subc (vector unsigned __int128,
6582 vector unsigned __int128);
6583 vector signed int vec_sube (vector signed int, vector signed int,
6585 vector unsigned int vec_sube (vector unsigned int,
6586 vector unsigned int,
6587 vector unsigned int);
6588 vector signed __int128 vec_sube (vector signed __int128,
6589 vector signed __int128,
6590 vector signed__int128);
6591 vector unsigned __int128 vec_sube (vector unsigned __int128,
6592 vector unsigned __int128,
6593 vector unsigned __int128);
6594 vector signed int vec_subec (vector signed int, vector signed int,
6596 vector unsigned int vec_subec (vector unsigned int,
6597 vector unsigned int,
6598 vector unsigned int);
6599 vector signed __int128 vec_subec (vector signed __int128,
6600 vector signed __int128,
6601 vector signed__int128);
6602 vector unsigned __int128 vec_subec (vector unsigned __int128,
6603 vector unsigned __int128,
6604 vector unsigned __int128);
6605 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
6606 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
6607 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
6608 BU_ALTIVEC_OVERLOAD_X definitions.
6609 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
6610 * doc/extend.texi: Update the built-in documentation file for the new
6613 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6616 * diagnostic-show-locus.c (layout::layout): Use start and finish
6617 spelling location for the start and finish of each range.
6618 * genmatch.c (linemap_client_expand_location_to_spelling_point):
6619 Add unused aspect param.
6620 * input.c (expand_location_1): Add "aspect" param, and use it
6621 to access the correct part of the location.
6622 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
6624 (expand_location_to_spelling_point): Likewise.
6625 (linemap_client_expand_location_to_spelling_point): Add "aspect"
6626 param, and pass it to expand_location_1.
6628 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
6630 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
6631 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
6632 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
6633 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
6634 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
6635 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
6636 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
6637 _mm_maskz_getmant_ss): New intrinsics.
6638 (__builtin_ia32_getexpss128_mask): Changed to ...
6639 __builtin_ia32_getexpss128_round ... this.
6640 (__builtin_ia32_getexpsd128_mask): Changed to ...
6641 __builtin_ia32_getexpsd128_round ... this.
6642 * config/i386/i386-builtin-types.def
6643 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
6644 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
6645 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
6646 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
6647 __builtin_ia32_getmantss_mask_round): New builtins.
6648 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
6649 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
6650 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
6651 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
6652 * config/i386/sse.md
6653 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
6654 avx512f_sgetexp<mode><mask_scalar_name>
6655 <round_saeonly_scalar_name> ... this.
6656 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
6657 %0, %1, %2<round_saeonly_op3>}): Changed to ...
6658 vgetexp<ssescalarmodesuffix>
6659 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6660 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
6661 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
6662 avx512f_vgetmant<mode><mask_scalar_name>
6663 <round_saeonly_scalar_name> ... this.
6664 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
6665 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
6666 vgetmant<ssescalarmodesuffix>
6667 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
6668 %0<mask_scalar_operand4>, %1, %2
6669 <round_saeonly_scalar_mask_op4>, %3} ... this.
6670 * config/i386/subst.md (mask_scalar_operand4,
6671 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
6672 round_saeonly_scalar_nimm_predicate): New subst attributes.
6674 2017-07-06 Julia Koval <julia.koval@intel.com>
6676 * config/i386/i386.c (ix86_erase_embedded_rounding):
6677 Remove code for old rounding pattern.
6679 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
6681 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
6683 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
6685 * doc/sourcebuild.texi (Test Directives, Variants of
6686 dg-require-support): Add documentation for dg-require-stack-check.
6688 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
6690 * config/i386/subst.md (mask_scalar, round_scalar,
6691 round_saeonly_scalar): New meta-templates.
6692 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
6693 round_scalar_mask_operand3, round_scalar_mask_op3,
6694 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
6695 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
6696 round_saeonly_scalar_constraint,
6697 round_saeonly_scalar_prefix): New subst attribute.
6698 * config/i386/sse.md
6699 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
6700 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
6701 <round_scalar_name> ... this.
6702 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
6703 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
6704 <round_scalar_name> ... this.
6705 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
6706 <sse>_vm<code><mode>3<mask_scalar_name>
6707 <round_saeonly_scalar_name> ... this.
6708 (v<plusminus_mnemonic><ssescalarmodesuffix>
6709 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
6710 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
6711 v<plusminus_mnemonic><ssescalarmodesuffix>
6712 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6713 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
6714 (v<multdiv_mnemonic><ssescalarmodesuffix>
6715 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
6716 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
6717 v<multdiv_mnemonic><ssescalarmodesuffix>
6718 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6719 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
6720 (v<maxmin_float><ssescalarmodesuffix>
6721 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
6722 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
6723 v<maxmin_float><ssescalarmodesuffix>
6724 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
6725 %0<mask_scalar_operand3>, %1, %<iptr>2
6726 <round_saeonly_scalar_mask_op3>} ... this.
6728 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
6730 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
6731 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
6733 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
6734 Alan Hayward <alan.hayward@arm.com>
6735 David Sherwood <david.sherwood@arm.com>
6737 * combine.c (simplify_if_then_else): Remove "enum" before
6739 * compare-elim.c (can_eliminate_compare): Likewise.
6740 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
6742 (aarch64_lookup_simd_builtin_type): Likewise.
6743 (aarch64_simd_builtin_type): Likewise.
6744 (aarch64_init_simd_builtin_types): Likewise.
6745 (aarch64_simd_expand_args): Likewise.
6746 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
6748 (aarch64_reverse_mask): Likewise.
6749 (aarch64_simd_emit_reg_reg_move): Likewise.
6750 (aarch64_gen_adjusted_ldpstp): Likewise.
6751 (aarch64_ccmp_mode_to_code): Likewise.
6752 (aarch64_operands_ok_for_ldpstp): Likewise.
6753 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6754 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
6756 (aarch64_min_divisions_for_recip_mul): Likewise.
6757 (aarch64_reassociation_width): Likewise.
6758 (aarch64_get_condition_code_1): Likewise.
6759 (aarch64_simd_emit_reg_reg_move): Likewise.
6760 (aarch64_simd_attr_length_rglist): Likewise.
6761 (aarch64_reverse_mask): Likewise.
6762 (aarch64_operands_ok_for_ldpstp): Likewise.
6763 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
6764 (aarch64_gen_adjusted_ldpstp): Likewise.
6765 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
6767 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
6768 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
6769 (arm_lookup_simd_builtin_type): Likewise.
6770 (arm_simd_builtin_type): Likewise.
6771 (arm_init_simd_builtin_types): Likewise.
6772 (arm_expand_builtin_args): Likewise.
6773 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
6774 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
6775 (ft32_setup_incoming_varargs): Likewise.
6776 (ft32_function_arg): Likewise.
6777 (ft32_function_arg_advance): Likewise.
6778 (ft32_pass_by_reference): Likewise.
6779 (ft32_arg_partial_bytes): Likewise.
6780 (ft32_valid_pointer_mode): Likewise.
6781 (ft32_addr_space_pointer_mode): Likewise.
6782 (ft32_addr_space_legitimate_address_p): Likewise.
6783 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
6785 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
6786 (ix86_emit_outlined_ms2sysv_restore): Likewise.
6787 (iamcu_alignment): Likewise.
6788 (canonicalize_vector_int_perm): Likewise.
6789 (ix86_noce_conversion_profitable_p): Likewise.
6790 (ix86_mpx_bound_mode): Likewise.
6791 (ix86_operands_ok_for_move_multiple): Likewise.
6792 * config/microblaze/microblaze-protos.h
6793 (microblaze_expand_conditional_branch_reg): Likewise.
6794 * config/microblaze/microblaze.c
6795 (microblaze_expand_conditional_branch_reg): Likewise.
6796 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
6798 (rs6000_reassociation_width): Likewise.
6799 (rs6000_invalid_binary_op): Likewise.
6800 (fusion_p9_p): Likewise.
6801 (emit_fusion_p9_load): Likewise.
6802 (emit_fusion_p9_store): Likewise.
6803 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
6805 (riscv_hard_regno_mode_ok_p): Likewise.
6806 (riscv_address_insns): Likewise.
6807 (riscv_split_symbol): Likewise.
6808 (riscv_legitimize_move): Likewise.
6809 (riscv_function_value): Likewise.
6810 (riscv_hard_regno_nregs): Likewise.
6811 (riscv_expand_builtin): Likewise.
6812 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
6813 (riscv_build_integer): Likewise.
6814 (riscv_split_integer): Likewise.
6815 (riscv_legitimate_constant_p): Likewise.
6816 (riscv_cannot_force_const_mem): Likewise.
6817 (riscv_regno_mode_ok_for_base_p): Likewise.
6818 (riscv_valid_base_register_p): Likewise.
6819 (riscv_valid_offset_p): Likewise.
6820 (riscv_valid_lo_sum_p): Likewise.
6821 (riscv_classify_address): Likewise.
6822 (riscv_legitimate_address_p): Likewise.
6823 (riscv_address_insns): Likewise.
6824 (riscv_load_store_insns): Likewise.
6825 (riscv_force_binary): Likewise.
6826 (riscv_split_symbol): Likewise.
6827 (riscv_force_address): Likewise.
6828 (riscv_legitimize_address): Likewise.
6829 (riscv_move_integer): Likewise.
6830 (riscv_legitimize_const_move): Likewise.
6831 (riscv_legitimize_move): Likewise.
6832 (riscv_address_cost): Likewise.
6833 (riscv_subword): Likewise.
6834 (riscv_output_move): Likewise.
6835 (riscv_canonicalize_int_order_test): Likewise.
6836 (riscv_emit_int_order_test): Likewise.
6837 (riscv_function_arg_boundary): Likewise.
6838 (riscv_pass_mode_in_fpr_p): Likewise.
6839 (riscv_pass_fpr_single): Likewise.
6840 (riscv_pass_fpr_pair): Likewise.
6841 (riscv_get_arg_info): Likewise.
6842 (riscv_function_arg): Likewise.
6843 (riscv_function_arg_advance): Likewise.
6844 (riscv_arg_partial_bytes): Likewise.
6845 (riscv_function_value): Likewise.
6846 (riscv_pass_by_reference): Likewise.
6847 (riscv_setup_incoming_varargs): Likewise.
6848 (riscv_print_operand): Likewise.
6849 (riscv_elf_select_rtx_section): Likewise.
6850 (riscv_save_restore_reg): Likewise.
6851 (riscv_for_each_saved_reg): Likewise.
6852 (riscv_register_move_cost): Likewise.
6853 (riscv_hard_regno_mode_ok_p): Likewise.
6854 (riscv_hard_regno_nregs): Likewise.
6855 (riscv_class_max_nregs): Likewise.
6856 (riscv_memory_move_cost): Likewise.
6857 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
6858 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
6859 (rl78_addr_space_address_mode): Likewise.
6860 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6862 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
6863 (rs6000_reassociation_width): Likewise.
6864 (rs6000_invalid_binary_op): Likewise.
6865 (fusion_p9_p): Likewise.
6866 (emit_fusion_p9_load): Likewise.
6867 (emit_fusion_p9_store): Likewise.
6868 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
6869 (ok_for_simple_move_operands): Likewise.
6870 (ok_for_simple_move_strict_operands): Likewise.
6871 (ok_for_simple_arith_logic_operands): Likewise.
6872 (visium_legitimize_reload_address): Likewise.
6873 (visium_select_cc_mode): Likewise.
6874 (output_cbranch): Likewise.
6875 (visium_split_double_move): Likewise.
6876 (visium_expand_copysign): Likewise.
6877 (visium_expand_int_cstore): Likewise.
6878 (visium_expand_fp_cstore): Likewise.
6879 * config/visium/visium.c (visium_pass_by_reference): Likewise.
6880 (visium_function_arg): Likewise.
6881 (visium_function_arg_advance): Likewise.
6882 (visium_libcall_value): Likewise.
6883 (visium_setup_incoming_varargs): Likewise.
6884 (visium_legitimate_constant_p): Likewise.
6885 (visium_legitimate_address_p): Likewise.
6886 (visium_legitimize_address): Likewise.
6887 (visium_secondary_reload): Likewise.
6888 (visium_register_move_cost): Likewise.
6889 (visium_memory_move_cost): Likewise.
6890 (prepare_move_operands): Likewise.
6891 (ok_for_simple_move_operands): Likewise.
6892 (ok_for_simple_move_strict_operands): Likewise.
6893 (ok_for_simple_arith_logic_operands): Likewise.
6894 (visium_function_value_1): Likewise.
6895 (rtx_ok_for_offset_p): Likewise.
6896 (visium_legitimize_reload_address): Likewise.
6897 (visium_split_double_move): Likewise.
6898 (visium_expand_copysign): Likewise.
6899 (visium_expand_int_cstore): Likewise.
6900 (visium_expand_fp_cstore): Likewise.
6901 (visium_split_cstore): Likewise.
6902 (visium_select_cc_mode): Likewise.
6903 (visium_split_cbranch): Likewise.
6904 (output_cbranch): Likewise.
6905 (visium_print_operand_address): Likewise.
6906 * expmed.c (flip_storage_order): Likewise.
6907 * expmed.h (emit_cstore): Likewise.
6908 (flip_storage_order): Likewise.
6909 * genrecog.c (validate_pattern): Likewise.
6910 * hsa-gen.c (gen_hsa_addr): Likewise.
6911 * internal-fn.c (expand_arith_overflow): Likewise.
6912 * ira-color.c (allocno_copy_cost_saving): Likewise.
6913 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6914 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
6915 (process_invariant_for_inheritance): Likewise.
6916 * lra-eliminations.c (move_plus_up): Likewise.
6917 * omp-low.c (lower_oacc_reductions): Likewise.
6918 * simplify-rtx.c (simplify_subreg): Likewise.
6919 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
6920 (TARGET_CHKP_BOUND_MODE): Likewise..
6921 * targhooks.c (default_chkp_bound_mode): Likewise.
6922 (default_setup_incoming_vararg_bounds): Likewise.
6923 * targhooks.h (default_chkp_bound_mode): Likewise.
6924 (default_setup_incoming_vararg_bounds): Likewise.
6925 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
6926 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
6927 (have_whole_vector_shift): Likewise.
6928 * tree-vect-stmts.c (vectorizable_load): Likewise.
6929 * doc/tm.texi: Regenerate.
6931 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
6933 Graceful degrade if Binutils PR21472 is not available.
6936 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
6937 .rodata in flash test fails.
6938 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
6939 * confgure: Regenerate.
6940 * config.in: Regenerate.
6941 * config/avr/avr.c (avr_asm_named_section)
6942 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
6943 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
6944 (avr_asm_init_sections): Same.
6946 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
6948 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
6949 (fma<VH:mode>4_intrinsic): Likewise.
6950 (*fmsub<VCVTF:mode>4): Likewise.
6951 (*fmsub<VH:mode>4_intrinsic): Likewise.
6953 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
6956 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
6957 Don't depend on "optimize > 0".
6958 (out_movhi_r_mr, out_movqi_mr_r): Same.
6959 (out_movhi_mr_r, out_movqi_r_mr): Same.
6960 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
6961 io_address_operand on "optimize > 0".
6963 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6965 * tree-loop-distribution.c: Add general explanantion on the pass.
6966 (generate_loops_for_partition): Mark distributed loop.
6967 (pg_add_dependence_edges): New parameter. Handle alias data
6968 dependence specially and record it in the parameter if asked.
6969 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
6970 (init_partition_graph_vertices, add_partition_graph_edge): New.
6971 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
6972 (free_partition_graph_vdata, build_partition_graph): New.
6973 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
6974 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
6975 (data_ref_segment_size, latch_dominated_by_data_ref): New.
6976 (compute_alias_check_pairs, version_loop_by_alias_check): New.
6977 (version_for_distribution_p, finalize_partitions): New.
6978 (distribute_loop): Handle alias data dependence specially. Factor
6979 out loop fusion code as functions and call these functions.
6981 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6983 * tree-loop-distribution.c (classify_partition): New parameter and
6984 better handle reduction statement.
6985 (rdg_build_partitions): Revise comment.
6986 (distribute_loop): Compute statements in all partitions and pass it
6987 to classify_partition.
6989 2017-07-05 Bin Cheng <bin.cheng@arm.com>
6991 * tree-loop-distribution.c (enum partition_type): New.
6992 (struct partition): New field type.
6993 (partition_merge_into): Add parameter. Update partition type.
6994 (data_dep_in_cycle_p, update_type_for_merge): New functions.
6995 (build_rdg_partition_for_vertex): Compute partition type.
6996 (rdg_build_partitions): Dump partition type.
6997 (distribute_loop): Update calls to partition_merge_into.
6999 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7001 * tree-loop-distribution.c (struct ddr_hasher): New.
7002 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
7004 (classify_partition): Call get_data_dependence.
7005 (pg_add_dependence_edges): Ditto.
7006 (distribute_loop): Release data dependence hash table.
7008 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7010 * tree-loop-distribution.c (ref_base_address): Delete.
7011 (similar_memory_accesses): Rename ...
7012 (share_memory_accesses): ... to this. Check if partitions access
7013 the same memory reference.
7014 (distribute_loop): Call share_memory_accesses.
7016 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7018 * tree-loop-distribution.c (struct partition): New field recording
7020 (partition_alloc, partition_free): Init and release data refs.
7021 (partition_merge_into): Merge data refs.
7022 (build_rdg_partition_for_vertex): Collect data refs for partition.
7023 (pg_add_dependence_edges): Change parameters from vector to bitmap.
7025 (distribute_loop): Remve data refs from vertice data of partition
7028 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7030 * tree-loop-distribution.c (params.h): Include header file.
7031 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
7032 (datarefs_vec): New global var.
7033 (create_rdg_vertices): Use datarefs_vec directly.
7034 (free_rdg): Don't free data references.
7035 (build_rdg): Update use. Don't free data references.
7036 (distribute_loop): Compute global variable for data references.
7037 Bail out if there are too many data references.
7039 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7041 * tree-loop-distribution.c (loop_nest): New global var.
7042 (build_rdg): Use loop directly, rather than loop nest.
7043 (pg_add_dependence_edges): Remove loop nest parameter. Use global
7045 (distribute_loop): Compute global variable loop nest. Update use.
7047 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7049 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
7050 (partition_merge_into): New parameter. Dump reason for fusion.
7051 (distribute_loop): Update use of partition_merge_into.
7053 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7055 * tree-loop-distribution.c (bb_top_order_index): New.
7056 (bb_top_order_index_size, bb_top_order_cmp): New.
7057 (stmts_from_loop): Use topological order.
7058 (pass_loop_distribution::execute): Compute and release topological
7059 order for basic blocks.
7061 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7063 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
7066 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7068 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
7069 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
7070 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
7071 * internal-fn.def (LOOP_DIST_ALIAS): New.
7072 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
7073 (fold_loop_internal_call): ... this.
7074 (vect_loop_dist_alias_call): New function.
7075 (set_uid_loop_bbs): Call fold_loop_internal_call.
7076 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
7079 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7082 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
7083 Require dead FLAGS_REG at the beginning of a peephole.
7085 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7088 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
7089 arguments in the call to __builtin_ia32_sbb_u32.
7090 (_subborrow_u64): Swap _X and _Y arguments in the call to
7091 __builtin_ia32_sbb_u64.
7093 2017-07-04 Jakub Jelinek <jakub@redhat.com>
7096 * tree-vrp.c (compare_assert_loc): Turn into a function template
7097 with stable template parameter. Only test if a->e is NULL,
7098 !a->e == !b->e has been verified already. Use e == NULL or
7099 e != NULL instead of e or ! e tests. If stable is true, don't use
7100 iterative_hash_expr, on the other side allow a or b or both NULL
7101 and sort the NULLs last.
7102 (process_assert_insertions): Sort using compare_assert_loc<false>
7103 instead of compare_assert_loc, later sort using
7104 compare_assert_loc<true> before calling process_assert_insertions_for
7105 in a loop. Use break instead of continue once seen NULL pointer.
7107 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
7109 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7110 Cortex-R7 and Cortex-R8 processors.
7112 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7114 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
7115 uninitialized while src is not.
7117 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
7119 * common/config/arm/arm-common.c: Adjust include path for
7121 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
7122 (arm-cpu.h): Create in build directory. Adjust dependency rules.
7123 (arm-cpu-data.h): Likewise.
7124 (arm-cpu-cdata.h): Likewise.
7125 * config/arm/arm-cpu.h: Delete.
7126 * config/arm/arm-cpu-cdata.h: Delete.
7127 * config/arm/arm-cpu-data.h: Delete.
7129 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
7131 * config/arm/arm-cpus.in (cortex-a55): New.
7132 (cortex-a75): Likewise.
7133 (cortex-a75.cortex-a55): Likewise.
7134 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
7136 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
7137 * config/arm/arm-cpu-cdata.h: Regenerate.
7138 * config/arm/arm-cpu-data.h: Regenerate.
7139 * config/arm/arm-cpu.h: Regenerate.
7140 * config/arm/arm-tables.opt: Regenerate.
7141 * config/arm/arm-tune.md: Regenerate.
7143 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7145 * haifa-sched.c (sched_create_recovery_edges): Update profile.
7147 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7149 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
7152 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
7154 PR tree-optimization/81292
7155 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
7156 full_string_p, also call adjust_related_strinfos if the adjustment
7157 is simple, otherwise invalidate related strinfos.
7159 2017-07-04 Martin Liska <mliska@suse.cz>
7162 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
7163 newly created variable as DECL_IGNORED_P.
7165 2017-07-04 Martin Liska <mliska@suse.cz>
7168 * ipa-inline.c (inline_small_functions):
7169 Use xstrdup_for_dump.
7171 2017-07-04 Tom de Vries <tom@codesourcery.com>
7173 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
7175 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
7178 * config/darwin.c (darwin_function_switched_text_sections):
7181 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
7183 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
7185 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
7187 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
7189 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7191 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
7192 min_profitable_iters, and th as inclusive lower bounds.
7193 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
7194 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
7195 for min_profitable_iters and min_profitable_estimate.
7196 (vect_transform_loop): Treat th as an inclusive lower bound.
7197 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
7199 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
7202 * config/darwin.c (darwin_function_switched_text_sections):
7203 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
7204 in two pieces, and suppress the use of buf.
7206 2017-07-03 Nathan Sidwell <nathan@acm.org>
7208 * hash-table.h (hash_table_mod1): Fix indentation.
7210 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7213 * predict.c (force_edge_cold): Be more careful about propagation
7215 * profile-count.h (profile_probability::guessed,
7216 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
7218 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
7220 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
7222 * doc/invoke.texi (rcpc architecture extension): Document it.
7224 2017-07-03 Richard Biener <rguenther@suse.de>
7226 PR tree-optimization/60510
7227 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
7228 the scalar reduction PHI and use it.
7229 (vectorizable_reduction): Properly guard the single_defuse_cycle
7230 path for non-SLP reduction chains where we cannot use it.
7231 Rework reduc_def/index and vector type deduction. Rework
7232 vector operand gathering during reduction op code-gen.
7233 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
7234 chains dissolve the chain and leave it to non-SLP reduction
7237 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7239 * tree-data-ref.h (dr_alignment): Declare.
7240 * tree-data-ref.c (dr_alignment): New function.
7241 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
7242 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
7244 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
7246 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7248 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
7249 and base_misalignment fields.
7250 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
7251 * tree-data-ref.c: Include builtins.h.
7252 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
7253 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
7254 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
7255 * tree-vect-data-refs.c: Include tree-cfg.h.
7256 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
7257 fields instead of calculating an alignment here.
7258 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
7259 innermost_loop_behavior fields.
7261 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7263 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
7265 (DR_STEP_ALIGNMENT): New macro.
7266 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
7267 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
7268 (create_data_ref): Print it.
7269 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
7270 to tell whether the step preserves vector (mis)alignment.
7271 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7272 Move the check for an integer step and generalise to all INTEGER_CST.
7273 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
7274 Print the outer step alignment.
7276 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7278 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
7279 with offset_alignment.
7280 (DR_ALIGNED_TO): Delete.
7281 (DR_OFFSET_ALIGNMENT): New macro.
7282 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
7283 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
7284 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
7285 (create_data_ref): Likewise.
7286 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7287 (vect_analyze_data_refs): Likewise.
7288 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
7289 creating dummy innermost behavior.
7291 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7293 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
7294 with a "innermost_loop_behavior *" and refeence tree.
7295 * tree-data-ref.c (dr_analyze_innermost): Likewise.
7296 (create_data_ref): Update call accordingly.
7297 * tree-predcom.c (find_looparound_phi): Likewise.
7299 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7301 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
7302 fields with dr_wrt_vec_loop.
7303 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
7304 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
7305 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
7306 (vect_dr_behavior): New function.
7307 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7308 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
7309 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
7310 track whether the step preserves the misalignment.
7311 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7312 Use vect_dr_behavior.
7313 (vect_setup_realignment): Update call accordingly.
7314 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
7315 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
7316 call to vect_create_addr_base_for_vector_ref.
7317 (vect_create_cond_for_align_checks): Likewise.
7318 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
7319 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
7320 (vect_recog_mask_conversion_pattern): Likewise.
7321 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
7322 (new_stmt_vec_info): Remove redundant zeroing.
7324 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
7326 * common/config/arm/arm-common.c (arm_be8_option): New function.
7327 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
7328 (ISA_ARMv6): Add isa_bit_be8.
7329 * config/arm/arm.h (arm_be8_option): Add prototype.
7330 (BE8_SPEC_FUNCTION): New define.
7331 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
7332 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
7333 (mlittle-endian): Similarly.
7334 (mbe8, mbe32): New options.
7335 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
7336 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
7338 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7340 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
7342 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7344 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
7345 (cleanup_tree_cfg_bb): Use it.
7346 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
7348 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
7350 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7353 * loop-doloop.c (add_test): Update profile.
7355 2017-07-03 Martin Liska <mliska@suse.cz>
7358 * sanopt.c (rewrite_usage_of_param): New function.
7359 (sanitize_rewrite_addressable_params): Likewise.
7360 (pass_sanopt::execute): Call rewrite_usage_of_param.
7362 2017-07-03 Richard Biener <rguenther@suse.de>
7364 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
7365 back to using VIEW_CONVERT_EXPR.
7367 2017-07-03 Martin Liska <mliska@suse.cz>
7370 * doc/extend.texi: Document when a resolver function is
7371 generated for target_clones.
7373 2017-07-03 Martin Liska <mliska@suse.cz>
7375 * asan.c (asan_emit_stack_protection): Unpoison just red zones
7376 and shadow memory of auto variables which are subject of
7377 use-after-scope sanitization.
7378 (asan_expand_mark_ifn): Add do set only when is_poison.
7380 2016-07-03 Richard Biener <rguenther@suse.de>
7382 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
7384 (vect_force_simple_reduction): Record reduction def -> phi mapping.
7385 (vectorizable_reduction): Perform reduction PHI creation when
7386 visiting a reduction PHI and adjust and simplify code generation
7387 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
7388 (vect_transform_loop): Visit reduction PHIs.
7389 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
7390 defs into the SLP tree.
7391 (vect_build_slp_tree): Reduction defs terminate the recursion.
7392 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
7394 (vect_get_vec_defs_for_stmt_copy): Export.
7395 (vect_get_vec_defs): Likewise.
7396 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
7398 (vect_get_vec_defs_for_stmt_copy): Declare.
7399 (vect_get_vec_defs): Likewise.
7401 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7403 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
7404 parameter with a "loop" parameter and use it instead of the
7405 loop containing DR_STMT. Don't check simple_iv when doing
7406 BB analysis. Describe the two analysis modes in the comment.
7408 2017-07-03 Tom de Vries <tom@codesourcery.com>
7410 PR tree-optimization/69468
7411 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
7412 (find_same_succ_bb): Handle ignore_edge_flags.
7414 2017-07-03 Tom de Vries <tom@codesourcery.com>
7416 PR tree-optimization/81192
7417 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
7419 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
7421 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
7423 2017-07-03 Tom de Vries <tom@codesourcery.com>
7425 PR tree-optimization/81192
7426 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
7427 BB_SAME_SUCC (bb) == NULL.
7429 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7431 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
7434 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7436 * dumpfile.c: Include profile-count.h
7437 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
7439 (insert_cond_bb): Update profile.
7440 * tree-cfg.h (insert_cond_bb): Update prototype.
7441 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
7442 * tree-dump.c: Do not include tree-cfg.
7444 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7446 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
7448 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7450 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
7453 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7455 * tree-complex.c (expand_complex_div_wide): update profile.
7457 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7458 Alan Hayward <alan.hayward@arm.com>
7459 David Sherwood <david.sherwood@arm.com>
7461 * Makefile.in (MACHMODE_H): Remove insn-modes.h
7462 (CORETYPES_H): New define.
7463 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
7464 (insn-modes-inline.h, s-modes-inline-h): New rules.
7465 (generated_files): Add insn-modes-inline.h.
7466 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
7467 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
7468 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
7469 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
7470 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
7471 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
7472 (build/gencodes.o, build/genconditions.o): Likewise.
7473 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
7474 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
7475 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
7476 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
7477 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
7478 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
7479 * coretypes.h: Include everything up to real.h for generators.
7480 Include insn-modes.h first. Include wide-int-print.h after
7481 wide-int.h. Include insn-modes-inline.h and then machmode.h.
7482 * machmode.h: Don't include insn-modes.h here.
7483 * function-tests.c: Remove includes of signop.h, machmode.h,
7484 double-int.h and wide-int.h.
7486 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
7488 * optc-save-gen.awk: Likewise.
7489 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
7490 * godump.c: Remove include of wide-int-print.h.
7491 * pretty-print.h: Likewise.
7492 * wide-int-print.cc: Likewise.
7493 * wide-int.cc: Likewise.
7494 * hash-map-tests.c: Remove include of signop.h.
7495 * hash-set-tests.c: Likewise.
7496 * rtl-tests.c: Likewise.
7497 * mkconfig.sh: Remove include of machmode.h.
7498 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
7500 (emit_insn_modes_inline_h): ...this new function. Emit the code
7501 into an insn-modes-inline.h header file, adding appropriate
7502 include guards and end comments.
7503 (emit_insn_modes_c_header): Remove include of machmode.h.
7504 (emit_min_insn_modes_c_header): Include coretypes.h rather than
7506 (main): Handle -i flag and call emit_insn_modes_inline_h when
7509 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7511 * tree-ssa-strlen.c (strinfo): Rename the length field to
7512 nonzero_chars. Add a full_string_p field.
7513 (compare_nonzero_chars, zero_length_string_p): New functions.
7514 (get_addr_stridx): Add an offset_out parameter.
7515 Use compare_nonzero_chars.
7516 (get_stridx): Update accordingly. Use compare_nonzero_chars.
7517 (new_strinfo): Update after above changes to strinfo.
7518 (set_endptr_and_length): Set full_string_p.
7519 (get_string_length): Update after above changes to strinfo.
7520 (unshare_strinfo): Update call to new_strinfo.
7521 (maybe_invalidate): Likewise.
7522 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
7523 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
7524 as a uhwi instead of an shwi. Update after above changes to
7525 strinfo and new_strinfo.
7526 (zero_length_string): Assert that chainsi contains full strings.
7527 Use zero_length_string_p. Update call to new_strinfo.
7528 (adjust_related_strinfos): Update after above changes to strinfo.
7529 Copy full_string_p from origsi.
7530 (adjust_last_stmt): Use zero_length_string_p.
7531 (handle_builtin_strlen): Update after above changes to strinfo and
7532 new_strinfo. Install the lhs as the string length if the previous
7533 entry didn't describe a full string.
7534 (handle_builtin_strchr): Update after above changes to strinfo
7536 (handle_builtin_strcpy): Likewise.
7537 (handle_builtin_strcat): Likewise.
7538 (handle_builtin_malloc): Likewise.
7539 (handle_pointer_plus): Likewise.
7540 (handle_builtin_memcpy): Likewise. Track nonzero characters
7541 that aren't necessarily followed by a nul terminator.
7542 (handle_char_store): Likewise.
7544 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7546 PR tree-optimization/80769
7547 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
7548 for malloc and calloc. Document the new invariant that all related
7549 strinfos have delayed lengths or none do.
7550 (verify_related_strinfos): Move earlier in file.
7551 (set_endptr_and_length): New function, split out from...
7552 (get_string_length): ...here. Also set the lengths of related
7554 (zero_length_string): Assert that chainsi has known (rather than
7556 (adjust_related_strinfos): Likewise.
7558 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7560 PR tree-optimization/81136
7561 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
7562 assert that two references with the same misalignment have the same
7563 compile-time misalignment if those compile-time misalignments
7566 2017-07-01 Andi Kleen <ak@linux.intel.com>
7568 * print-tree.c (print_node): Print all attributes.
7570 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7572 * cfg.c (scale_bbs_frequencies): New function.
7573 * cfg.h (scale_bbs_frequencies): Declare it.
7574 * cfgloopanal.c (single_likely_exit): Cleanup.
7575 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
7577 (scale_loop_profile): Likewise.
7578 (loop_version): Likewise.
7579 (create_empty_loop_on_edge): Update.
7580 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
7581 scale_loop_frequencies, scale_loop_profile, loopify,
7582 loop_version): Update prototypes.
7583 * modulo-sched.c (sms_schedule): Update.
7584 * predict.c (unlikely_executed_edge_p): Also check probability.
7585 (probably_never_executed_edge_p): Fix typo.
7586 * tree-if-conv.c (version_loop_for_if_conversion): Update.
7587 * tree-parloops.c (gen_parallel_loop): Update.
7588 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
7589 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7590 * tree-ssa-loop-split.c (split_loop): Update.
7591 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
7592 * tree-vect-loop-manip.c (vect_do_peeling): Update.
7593 (vect_loop_versioning): Update.
7594 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
7596 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7598 * trans-mem.c (split_bb_make_tm_edge): Update profile.
7600 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7602 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
7603 to keep profile consistent.
7605 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7607 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
7608 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
7609 * profile-count.h (max_safe_multiplier): Make unsigned.
7610 (profile_count::guessed_zero): New.
7612 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7614 * bb-reorder.c (fix_up_crossing_landing_pad,
7615 fix_crossing_conditional_branches): Use make_single_succ_edge
7616 to keep profile consistent.
7618 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
7620 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
7623 2017-07-01 Jakub Jelinek <jakub@redhat.com>
7626 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
7627 the right scopes, make sure cond_jump isn't preserved between multiple
7628 iterations. Search for fallthru edge whenever there are 3+ edges and
7629 use find_fallthru_edge for it.
7631 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7633 Patch by Alexander Monakov <amonakov@ispras.ru>
7634 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
7635 probabilities consistently.
7637 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7639 * pa.c (pa_expand_compare_and_swap_loop): Update call of
7640 emit_cmp_and_jump_insns.
7642 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7645 * tree-inline.c (expand_call_inline): Combine profile statuses.
7647 2017-06-30 Andrew Pinski <apinski@cavium.com>
7649 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
7650 fold_stmt returned true.
7652 2017-06-30 Nathan Sidwell <nathan@acm.org>
7654 * ggc.h (empty_string): Delete.
7655 * cfgexpand.c (expand_asm_stmt): Use plain "".
7656 * optabs.c (expand_asm_memory_barrier): Likewise.
7657 * stringpool.c (empty_string): Delete.
7658 (digit_vector, digit_string): Delete.
7659 (ggc_alloc_string): Use plain "", don't optimize single digit
7660 strings. Use ggc_alloc_atomic.
7662 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
7664 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
7665 comparison set and one other set, use the cost of the non-comparison
7668 2017-06-30 Nathan Sidwell <nathan@acm.org>
7670 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
7673 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
7675 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
7676 loops. Remove now unneeded calls to gimple_switch_set_label() that
7677 just set removed labels to NULL_TREE.
7679 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
7681 * tree-ssanames.c (set_range_info_raw): Abstract from ...
7682 (set_range_info): ...here. Only call set_range_info_raw if domain
7684 (set_nonzero_bits): Call set_range_info_raw.
7685 * tree-ssanames.h (set_range_info_raw): New.
7687 2017-06-30 Jakub Jelinek <jakub@redhat.com>
7690 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
7691 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
7692 of nonimmediate_operand and <store_mask_constraint> instead of m
7693 for the input operand. For V8FI iterator, always split if input
7694 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
7695 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
7696 <store_mask_predicate> instead of register_operand and
7697 <store_mask_constraint> instead of v for the input operand. Make
7698 sure both operands aren't MEMs for if not <mask_applied>.
7700 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
7702 * lto-wrapper.c (copy_file) Close both file descriptors before
7705 2017-06-30 Martin Liska <mliska@suse.cz>
7708 * multiple_target.c (create_dispatcher_calls): Make ifunc
7709 also for function that don't have calls or are not referenced.
7711 2017-06-30 Richard Biener <rguenther@suse.de>
7713 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
7714 analyze the first scalar stmt. Move vector type computation
7715 for the BB case here from ...
7716 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
7717 live operation processing in the SLP case properly.
7719 2017-06-30 Richard Biener <rguenther@suse.de>
7721 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
7723 2017-06-30 Martin Liska <mliska@suse.cz>
7726 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
7727 before BUILT_IN_UNWIND_RESUME when ASAN is used.
7729 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
7731 * doc/invoke.texi (AArch64): Add missing options and remove redundant
7734 2017-06-30 Richard Biener <rguenther@suse.de>
7736 PR tree-optimization/81249
7737 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
7738 condition reduction result to original scalar type.
7740 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7742 * profile-count.h (enum profile_quality): Fix typos and whitespace
7745 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7747 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
7748 type for branch probabilities.
7750 2017-06-29 Julian Brown <julian@codesourcery.com>
7751 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7753 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
7754 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
7755 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
7756 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
7758 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
7760 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
7761 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
7762 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
7763 CC usage from generic code to here.
7764 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
7765 CC usage into the target macros.
7767 2017-06-29 Maya Rashish <coypu@sdf.org>
7769 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
7772 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7774 * arm/arm-builtins.c: Include profile-count.h
7775 * except.c (sjlj_emit_function_enter): Use
7776 profile_probability::unlikely.
7778 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7780 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
7781 and tocrel_offset be pointer args rather than implicitly using
7783 (legitimate_constant_pool_address_p, rs6000_emit_move,
7784 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
7785 tocrel_offset and use in toc_relative_expr_p call.
7786 (print_operand, print_operand_address): Use static tocrel_base_oac
7787 and tocrel_offset_oac.
7788 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
7791 2017-06-29 Maya Rashish <coypu@sdf.org>
7793 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
7795 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
7797 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
7798 objects, take into account only the alignment of 'op0' and 'mode1' if
7801 2017-06-29 Steve Ellcey <sellcey@cavium.com>
7803 * ccmp.c (ccmp_tree_comparison_p): New function.
7804 (ccmp_candidate_p): Update to use above function.
7805 (get_compare_parts): New function.
7806 (expand_ccmp_next): Update to use new functions.
7807 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
7809 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
7810 take mode as argument.
7811 * ccmp.h (expand_ccmp_expr): Add mode as argument.
7812 * expr.c (expand_expr_real_1): Pass mode as argument.
7814 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
7816 * combine.c (combine_instructions): Print insns to dump_file, together
7819 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
7821 * asan.c (asan_emit_stack_protection): Update.
7822 (create_cond_insert_point): Update.
7823 * auto-profile.c (afdo_propagate_circuit): Update.
7824 * basic-block.h (struct edge_def): Turn probability to
7825 profile_probability.
7826 (EDGE_FREQUENCY): Update.
7827 * bb-reorder.c (find_traces_1_round): Update.
7828 (better_edge_p): Update.
7829 (sanitize_hot_paths): Update.
7830 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
7831 (make_single_succ_edge): Update.
7832 (check_bb_profile): Update.
7833 (dump_edge_info): Update.
7834 (update_bb_profile_for_threading): Update.
7835 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
7836 probabilitycount to 0.
7837 * cfgbuild.c (compute_outgoing_frequencies): Update.
7838 * cfgcleanup.c (try_forward_edges): Update.
7839 (outgoing_edges_match): Update.
7840 (try_crossjump_to_edge): Update.
7841 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
7842 (expand_gimple_tailcall): Update.
7843 (construct_init_block): Use make_single_succ_edge.
7844 (construct_exit_block): Use make_single_succ_edge.
7845 * cfghooks.c (verify_flow_info): Update.
7846 (redirect_edge_succ_nodup): Update.
7847 (split_edge): Update.
7848 (account_profile_record): Update.
7849 * cfgloopanal.c (single_likely_exit): Update.
7850 * cfgloopmanip.c (scale_loop_profile): Update.
7851 (set_zero_probability): Remove.
7852 (duplicate_loop_to_header_edge): Update.
7853 * cfgloopmanip.h (loop_version): Update prototype.
7854 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
7855 (force_nonfallthru_and_redirect): Update.
7856 (update_br_prob_note): Update.
7857 (rtl_verify_edges): Update.
7858 (purge_dead_edges): Update.
7859 (rtl_lv_add_condition_to_bb): Update.
7860 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
7861 * cgraphunit.c (init_lowered_empty_function): Update.
7862 (cgraph_node::expand_thunk): Update.
7863 * cilk-common.c: Include profile-count.h
7864 * dojump.c (inv): Remove.
7865 (jumpifnot): Update.
7866 (jumpifnot_1): Update.
7867 (do_jump_1): Update.
7869 (do_jump_by_parts_greater_rtx): Update.
7870 (do_compare_rtx_and_jump): Update.
7871 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
7872 do_jump_1. do_compare_rtx_and_jump): Update prototype.
7873 * dwarf2cfi.c: Include profile-count.h
7874 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
7875 (sjlj_emit_dispatch_table): Likewise.
7876 * explow.c: Include profile-count.h
7877 * expmed.c (emit_store_flag_force): Update.
7878 (do_cmp_and_jump): Update.
7879 * expr.c (compare_by_pieces_d::generate): Update.
7880 (compare_by_pieces_d::finish_mode): Update.
7881 (emit_block_move_via_loop): Update.
7882 (store_expr_with_bounds): Update.
7883 (store_constructor): Update.
7884 (expand_expr_real_2): Update.
7885 (expand_expr_real_1): Update.
7886 * expr.h (try_casesi, try_tablejump): Update prototypes.
7887 * gimple-pretty-print.c (dump_probability): Update.
7888 (dump_profile): New.
7889 (dump_gimple_label): Update.
7890 (dump_gimple_bb_header): Update.
7891 * graph.c (draw_cfg_node_succ_edges): Update.
7892 * hsa-gen.c (convert_switch_statements): Update.
7893 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
7894 (find_if_case_1): Update.
7895 (find_if_case_2): Update.
7896 * internal-fn.c (expand_arith_overflow_result_store): Update.
7897 (expand_addsub_overflow): Update.
7898 (expand_neg_overflow): Update.
7899 (expand_mul_overflow): Update.
7900 (expand_vector_ubsan_overflow): Update.
7901 * ipa-cp.c (good_cloning_opportunity_p): Update.
7902 * ipa-split.c (split_function): Use make_single_succ_edge.
7903 * ipa-utils.c (ipa_merge_profiles): Update.
7904 * loop-doloop.c (add_test): Update.
7905 (doloop_modify): Update.
7906 * loop-unroll.c (compare_and_jump_seq): Update.
7907 (unroll_loop_runtime_iterations): Update.
7908 * lra-constraints.c (lra_inheritance): Update.
7909 * lto-streamer-in.c (input_cfg): Update.
7910 * lto-streamer-out.c (output_cfg): Update.
7911 * mcf.c (adjust_cfg_counts): Update.
7912 * modulo-sched.c (sms_schedule): Update.
7913 * omp-expand.c (expand_omp_for_init_counts): Update.
7914 (extract_omp_for_update_vars): Update.
7915 (expand_omp_ordered_sink): Update.
7916 (expand_omp_for_ordered_loops): Update.
7917 (expand_omp_for_generic): Update.
7918 (expand_omp_for_static_nochunk): Update.
7919 (expand_omp_for_static_chunk): Update.
7920 (expand_cilk_for): Update.
7921 (expand_omp_simd): Update.
7922 (expand_omp_taskloop_for_outer): Update.
7923 (expand_omp_taskloop_for_inner): Update.
7924 * omp-simd-clone.c (simd_clone_adjust): Update.
7925 * optabs.c (expand_doubleword_shift): Update.
7926 (expand_abs): Update.
7927 (emit_cmp_and_jump_insn_1): Update.
7928 (expand_compare_and_swap_loop): Update.
7929 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
7930 * predict.c (predictable_edge_p): Update.
7931 (edge_probability_reliable_p): Update.
7932 (set_even_probabilities): Update.
7933 (combine_predictions_for_insn): Update.
7934 (combine_predictions_for_bb): Update.
7935 (propagate_freq): Update.
7936 (estimate_bb_frequencies): Update.
7937 (force_edge_cold): Update.
7938 * profile-count.c (profile_count::dump): Add missing space into dump.
7939 (profile_count::debug): Add newline.
7940 (profile_count::differs_from_p): Explicitly convert to unsigned.
7941 (profile_count::stream_in): Update.
7942 (profile_probability::dump): New member function.
7943 (profile_probability::debug): New member function.
7944 (profile_probability::differs_from_p): New member function.
7945 (profile_probability::differs_lot_from_p): New member function.
7946 (profile_probability::stream_in): New member function.
7947 (profile_probability::stream_out): New member function.
7948 * profile-count.h (profile_count_quality): Rename to ...
7949 (profile_quality): ... this one.
7950 (profile_probability): New.
7951 (profile_count): Update.
7952 * profile.c (compute_branch_probabilities): Update.
7953 * recog.c (peep2_attempt): Update.
7954 * sched-ebb.c (schedule_ebbs): Update.
7955 * sched-rgn.c (find_single_block_region): Update.
7956 (compute_dom_prob_ps): Update.
7957 (schedule_region): Update.
7958 * sel-sched-ir.c (compute_succs_info): Update.
7959 * stmt.c (struct case_node): Update.
7960 (do_jump_if_equal): Update.
7961 (get_outgoing_edge_probs): Update.
7962 (conditional_probability): Update.
7963 (emit_case_dispatch_table): Update.
7964 (expand_case): Update.
7965 (expand_sjlj_dispatch_table): Update.
7966 (emit_case_nodes): Update.
7967 * targhooks.c: Update.
7968 * tracer.c (better_p): Update.
7969 (find_best_successor): Update.
7970 * trans-mem.c (expand_transaction): Update.
7971 * tree-call-cdce.c: Update.
7972 * tree-cfg.c (gimple_split_edge): Upate.
7973 (move_sese_region_to_fn): Upate.
7974 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
7975 * tree-eh.c (lower_resx): Upate.
7976 (cleanup_empty_eh_move_lp): Upate.
7977 * tree-if-conv.c (version_loop_for_if_conversion): Update.
7978 * tree-inline.c (copy_edges_for_bb): Update.
7979 (copy_cfg_body): Update.
7980 * tree-parloops.c (gen_parallel_loop): Update.
7981 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
7982 (gimple_gen_time_profiler): Update.
7983 * tree-ssa-dce.c (remove_dead_stmt): Update.
7984 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
7985 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
7986 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
7987 (unloop_loops): Update.
7988 (try_peel_loop): Update.
7989 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
7990 * tree-ssa-loop-split.c (connect_loops): Update.
7991 (split_loop): Update.
7992 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
7993 (hoist_guard): Update.
7994 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
7995 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
7996 (value_replacement): Update.
7997 * tree-ssa-reassoc.c (branch_fixup): Update.
7998 * tree-ssa-tail-merge.c (replace_block_by): Update.
7999 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8000 (create_edge_and_update_destination_phis): Update.
8001 (compute_path_counts): Update.
8002 (recompute_probabilities): Update.
8003 (update_joiner_offpath_counts): Update.
8004 (freqs_to_counts_path): Update.
8005 (duplicate_thread_path): Update.
8006 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8007 (struct switch_conv_info): Update.
8008 (gen_inbound_check): Update.
8009 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8010 (vect_do_peeling): Update.
8011 (vect_loop_versioning): Update.
8012 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8013 (optimize_mask_stores): Update.
8014 * ubsan.c (ubsan_expand_null_ifn): Update.
8015 * value-prof.c (gimple_divmod_fixed_value): Update.
8016 (gimple_divmod_fixed_value_transform): Update.
8017 (gimple_mod_pow2): Update.
8018 (gimple_mod_pow2_value_transform): Update.
8019 (gimple_mod_subtract): Update.
8020 (gimple_mod_subtract_transform): Update.
8021 (gimple_ic): Update.
8022 (gimple_stringop_fixed_value): Update.
8023 (gimple_stringops_transform): Update.
8024 * value-prof.h: Update.
8026 2017-06-29 Carl Love <cel@us.ibm.com>
8028 * config/rs6000/rs6000-c.c: Add support for built-in functions
8029 vector signed int vec_signed (vector float);
8030 vector signed long long vec_signed (vector double);
8031 vector signed int vec_signed2 (vector double, vector double);
8032 vector signed int vec_signede (vector double);
8033 vector signed int vec_signedo (vector double);
8034 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
8035 instruction generator.
8036 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8037 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
8038 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
8040 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
8041 vunsignede_v2df): Add define_expands.
8042 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
8043 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
8044 VEC_UNSIGNEDO): Add definitions.
8045 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8046 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
8047 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
8048 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
8049 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
8050 * config/rs6000/altivec.h (vec_signed, vec_signed2,
8051 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
8052 vec_unsignede, vec_unsignedo): Add builtin defines.
8053 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
8055 * doc/extend.texi: Update the built-in documentation file for the
8056 new built-in functions.
8058 2017-06-29 Richard Biener <rguenther@suse.de>
8060 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
8061 reduction chains to LOOP_VINFO_REDUCTIONS.
8062 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
8063 SLP reductions after processing reduction chains.
8065 2017-06-29 Nathan Sidwell <nathan@acm.org>
8067 * builtins.c (fold_builtin_FUNCTION): Use
8068 lang_hooks.decl_printable_name.
8070 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
8073 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
8074 with only one label.
8075 * stmt.c (expand_case): Assert NCASES is greater than one.
8077 2017-06-29 Richard Biener <rguenther@suse.de>
8079 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
8081 (group_case_labels): Likewise.
8082 (find_taken_edge): Push sanity checking on val to workers...
8083 (find_taken_edge_cond_expr): ... here
8084 (find_taken_edge_switch_expr): ... and here, handle cases
8085 with just a default label.
8086 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
8087 (group_case_labels): Likewise.
8088 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
8089 group_case_labels does anything cleanup the CFG again.
8091 2017-06-29 Bin Cheng <bin.cheng@arm.com>
8093 PR tree-optimization/81196
8094 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
8095 exit condition comparing two IVs.
8097 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
8099 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
8100 profile to the dummy entry at the end of the list of architectures.
8101 * config/arm/arm-cpu-cdata.h: Regenerated.
8103 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8104 Michael Collison <michael.collison@arm.com>
8107 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
8109 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
8110 (*aarch64_reg_<mode>3_minus_mask): New pattern.
8111 (*aarch64_<optab>_reg_di3_mask2): New pattern.
8112 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
8113 of shift when the shift amount is masked with constant equal to
8114 the size of the mode.
8115 * config/aarch64/predicates.md (subreg_lowpart_operator): New
8118 2017-06-29 Martin Liska <mliska@suse.cz>
8120 * config/i386/i386.opt: Change range from [1,5] to [0,5].
8122 2017-06-29 Yury Gribov <tetra2005@gmail.com>
8125 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
8128 (edge_growth_cache_entry::edge_growth_cache_entry): New
8130 (reset_edge_growth_cache): Update to use constructor.
8132 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8134 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
8135 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
8136 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
8138 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
8140 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
8141 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
8143 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
8145 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
8146 (*-linux-uclibc*): Add t-uclibc tmake_file.
8147 * config/t-musl: New.
8148 * config/t-uclibc: New.
8150 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
8152 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
8154 (gen_comm_data): Emit architectural setting of arch_prof.
8155 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
8157 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
8158 (armv8-m.base, armv8-m.main): Likewise.
8159 * arm-protos.h (arm_build_target): Add profile field.
8160 (arch_option): Likewise.
8161 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
8163 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
8164 arm_active_target.profile.
8166 2017-06-28 Richard Biener <rguenther@suse.de>
8169 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
8170 TYPE_OVERFLOW_WRAPS.
8171 * match.pd (negate_expr_p): Likewise.
8172 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
8173 fold_build2, not fold_binary.
8175 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
8177 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
8178 Convert memory address to Pmode.
8179 (aarch64_print_operand): Assert MEM operands are always Pmode.
8181 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
8184 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
8185 Remove redundant if.
8186 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
8187 * config/arm/aarch-common-protos.h
8188 (aarch_forward_to_shift_is_not_shifted_re): Remove.
8189 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
8191 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
8194 * multiple_target.c (create_dispatcher_calls): Set the default
8195 clone to be static, not public.
8197 2017-06-28 Richard Biener <rguenther@suse.de>
8199 * tree-vect-loop.c (vectorizable_reduction): Move special
8200 cond reduction IV var creation ...
8201 (vect_create_epilog_for_reduction): ... here. Remove induction_index
8202 parameter. Use STMT_VINFO_VECTYPE.
8203 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
8206 2017-06-28 Martin Liska <mliska@suse.cz>
8209 * ipa-visibility.c (non_local_p): Handle visibility.
8211 2017-06-28 Martin Liska <mliska@suse.cz>
8214 * common.opt: Add IntegerRange to various options.
8215 * opt-functions.awk (integer_range_info): New function.
8216 * optc-gen.awk: Add integer_range_info to cl_options struct.
8217 * opts-common.c (decode_cmdline_option): Handle
8218 CL_ERR_INT_RANGE_ARG.
8219 (cmdline_handle_error): Likewise.
8220 * opts.c (print_filtered_help): Show valid interval in
8221 when --help is provided.
8222 * opts.h (struct cl_option): Add range_min and range_max fields.
8223 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
8225 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
8227 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
8228 (x * C EQ/NE y * C): New transformation.
8230 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
8232 * genmultilib (combination_space): Accept '+' in option names.
8234 2017-06-28 Martin Liska <mliska@suse.cz>
8237 * asan.c (instrument_derefs): Bail out inner references
8238 that are hard register variables.
8240 2017-06-28 Jakub Jelinek <jakub@redhat.com>
8243 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
8244 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
8246 2017-06-28 Richard Biener <rguenther@suse.de>
8248 * tree-vectorizer.h (vect_get_vec_defs): Remove.
8249 (vect_get_slp_defs): Adjust.
8250 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
8252 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
8254 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
8255 get_initial_defs_for_reduction instead of vect_get_vec_defs.
8256 (vectorizable_reduction): Adjust.
8257 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
8259 (vect_get_slp_defs): Likewise.
8260 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
8261 (vectorizable_bswap): Adjust.
8262 (vectorizable_call): Likewise.
8263 (vectorizable_conversion): Likewise.
8264 (vectorizable_assignment): Likewise.
8265 (vectorizable_shift): Likewise.
8266 (vectorizable_operation): Likewise.
8267 (vectorizable_store): Likewise.
8268 (vectorizable_condition): Likewise.
8269 (vectorizable_comparison): Likewise.
8271 2017-06-28 Michael Collison <michael.collison@arm.com>
8274 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
8276 (arm_gen_movmemqi): Removed unused variable 'i'.
8277 Convert 'for' loop into 'while' loop.
8278 (arm_expand_prologue): Remove last unnecessary set of insn.
8279 (thumb_pop): Remove unused variable 'pushed_words'.
8280 (thumb_exit): Remove last unnecessary set of regs_to_pop.
8282 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8284 * config/s390/predicates.md: Use s390_rel_address_ok_p.
8285 * config/s390/s390-protos.h: Add prototype of
8286 s390_rel_address_ok_p.
8287 * config/s390/s390.c (s390_got_symbol): New function.
8288 (s390_rel_address_ok_p): New function.
8289 (legitimize_pic_address): Use s390_rel_address_ok_p.
8290 (s390_load_got): Use s390_got_symbol.
8291 (s390_option_override): Issue error if
8292 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
8293 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
8295 * config/s390/s390.opt: New option mpic-data-is-text-relative.
8297 2017-06-27 Andrew Pinski <apinski@cavium.com>
8299 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
8300 (X * copysign (1.0, X)): New pattern.
8301 (X * copysign (1.0, -X)): New pattern.
8302 (copysign (-1.0, CST)): New pattern.
8304 2017-06-27 Joseph Myers <joseph@codesourcery.com>
8306 * genmultilib (combination_space): Remove variable.
8307 Validate reuse rules against regular expression for any sequence
8308 of multilib options in any order.
8310 2017-06-27 Michael Collison <michael.collison@arm.com>
8312 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
8313 call aarch64_split_simd_combine.
8314 * (aarch64_combine_internal<mode>): Delete pattern.
8315 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
8316 Allow register and subreg operands.
8318 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8320 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
8321 specific need, just fallback on defaults.
8322 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
8324 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8325 Olivier Hainque <hainque@adacore.com>
8327 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
8329 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
8330 targets. Pick a default if no particular attempt applied.
8331 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
8334 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8336 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
8337 (x86_64-wrs-vxworks7): Likewise.
8339 2017-06-27 Marek Polacek <polacek@redhat.com>
8342 * ubsan.c (instrument_null): Check get_base_address's result for null.
8344 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8346 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
8348 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8350 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
8351 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
8352 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
8353 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
8355 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
8356 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
8357 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
8358 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
8359 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
8360 BUILT_IN_FEUPDATEENV): New builtins.
8361 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
8362 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
8363 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
8364 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
8366 (builtin_structptr_types): Adjust size.
8367 * tree.c (builtin_structptr_types): Add four entries.
8369 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8370 Olivier Hainque <hainque@adacore.com>
8372 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
8373 (TLS_SYM): New local macro, forcing reference to __tls__ on
8374 link command lines for VxWorks 7 RTPs, triggering initialization
8376 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
8377 OS features TLS support, true for RTPs on VxWorks 7.
8378 * config/vxworks.c (vxworks_override_options): Setup emutls
8381 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8383 * predict.c (test_prediction_value_range): Use -1U instead of -1
8384 to avoid narrowing conversion warning.
8385 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
8386 to avoid narrowing conversion warning.
8387 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
8389 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
8391 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8393 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
8394 64bit configurations.
8395 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
8396 (SIZE_TYPE): Likewise.
8397 * config/vxworks.c (vxworks_emutls_var_fields): Use
8398 long_unsigned_type_node instead of unsigned_type_node as the offset
8399 field type, which is "pointer" mode in emutls.c.
8401 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8404 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
8407 * gimple-fold.c (replace_call_with_call_and_fold): Handle
8408 gimple_vuse copying separately from gimple_vdef copying.
8410 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8412 * value-prof.c (free_hist): Remove call to memset and the enclosing if
8415 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8416 Olivier Hainque <hainque@adacore.com>
8418 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
8419 for all vxworks7 targets.
8420 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
8421 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
8422 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
8423 variations for VX6/VX7 and 32/64bits later on in ...
8424 (VXWORKS_LIB_SPEC): Leverage new macros.
8425 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
8426 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
8428 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8430 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
8431 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
8433 2017-06-26 Carl Love <cel@us.ibm.com>
8435 * config/rs6000/rs6000-c.c: Add support for built-in functions
8436 vector bool char vec_reve (vector bool char);
8437 vector signed char vec_reve (vector signed char);
8438 vector unsigned char vec_reve (vector unsigned char);
8439 vector bool int vec_reve (vector bool int);
8440 vector signed int vec_reve (vector signed int);
8441 vector unsigned int vec_reve (vector unsigned int);
8442 vector bool long long vec_reve (vector bool long long);
8443 vector signed long long vec_reve (vector signed long long);
8444 vector unsigned long long vec_reve (vector unsigned long long);
8445 vector bool short vec_reve (vector bool short);
8446 vector signed short vec_reve (vector signed short);
8447 vector double vec_reve (vector double);
8448 vector float vec_reve (vector float);
8449 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
8450 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
8451 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
8452 (altivec_vreve): New pattern.
8453 * config/rs6000/altivec.h (vec_reve): New define.
8454 * doc/extend.texi (vec_rev): Update the built-in documentation file
8455 for the new built-in functions.
8457 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8459 PR tree-optimization/71815
8460 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
8462 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
8464 (slsr_process_phi): Likewise.
8465 (replace_uncond_cands_and_profitable_phis): Don't replace a
8466 multiply candidate with a stride of 1 (copy or cast).
8467 (phi_incr_cost): Call uses_consumed_by_stmt rather than
8469 (lowest_cost_path): Likewise.
8470 (total_savings): Likewise.
8472 2017-06-26 Richard Biener <rguenther@suse.de>
8475 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
8476 Use def_builtin_pure for all gather builtins.
8478 2017-06-26 Richard Biener <rguenther@suse.de>
8480 PR tree-optimization/81203
8481 * tree-tailcall.c (find_tail_calls): Do not move stmts into
8484 2017-06-26 Marek Polacek <polacek@redhat.com>
8487 * doc/invoke.texi: Document -Wmultistatement-macros.
8489 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
8491 * doc/sourcebuild.texi (ARM-specific attributes): Document new
8492 arm_neon_ok_no_float_abi effective target.
8494 2017-06-26 Richard Biener <rguenther@suse.de>
8496 PR tree-optimization/80928
8497 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
8498 (copy_bbs): Set BB_DUPLICATED flag early.
8499 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
8501 (execute_on_shrinking_pred): Likewise.
8502 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
8503 BB_DUPLICATED blocks.
8504 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
8505 iterate over all PHIs considering removal of *gsi.
8507 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
8509 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
8512 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8514 * config/rs6000/rs6000-string.c: (expand_block_clear,
8515 do_load_for_compare, select_block_compare_mode,
8516 compute_current_alignment, expand_block_compare,
8517 expand_strncmp_align_check, expand_strn_compare,
8518 expand_block_move, rs6000_output_load_multiple)
8519 Move functions related to string/block move/compare
8521 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
8522 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
8523 for this function which is now used in two files.
8524 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
8525 * config.gcc: Add rs6000-string.o to extra_objs for
8526 targets powerpc*-*-* and rs6000*-*-*.
8528 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
8531 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
8532 32-bit, since indexed is not valid for DImode.
8533 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
8534 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
8535 (define_peephole2 for Altivec d-form load): Add 32-bit support.
8536 (define_peephole2 for Altivec d-form store): Likewise.
8539 * multiple_target.c (create_dispatcher_calls): Only create the
8540 dispatcher call if the function is the default clone of a
8543 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
8546 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
8547 a call, force the call to not be a tail call.
8549 2017-06-23 Jeff Law <law@redhat.com>
8551 * doc/contrib.texi: Add entry for Steven Pemberton's work on
8554 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
8556 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
8557 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
8558 handling for early expansion of vector shifts (sl,sr,sra,rl).
8559 (builtin_function_type): Add vector shift right instructions
8560 to the unsigned argument list.
8562 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
8564 rtl-optimizatoin/79286
8565 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
8566 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
8567 trap. PIC register plus a const unspec without offset can never trap.
8569 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
8571 * tree.h (builtin_structptr_type): New type.
8572 (builtin_structptr_types): Declare new array.
8573 * tree.c (builtin_structptr_types): New array.
8574 (free_lang_data, build_common_tree_nodes): Use it.
8576 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
8579 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
8582 2017-06-22 Matt Turner <mattst88@gmail.com>
8584 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
8585 Lake models to skylake case. Assume skylake for unknown
8586 models with clflushopt.
8588 2017-06-22 Jeff Law <law@redhat.com>
8590 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
8591 frame sizes that do not satisfy aarch64_uimm12_shift.
8593 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
8595 * profile-count.h (apply_probability,
8596 apply_scale, probability_in): Fix checks for zero.
8598 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8600 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
8601 * doc/cppdiropts.texi (-I @var{dir}): Document it.
8603 2016-06-22 Richard Biener <rguenther@suse.de>
8605 * tree-vect-loop.c (vect_model_reduction_cost): Handle
8606 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
8607 REDUC_MAX_EXPR support.
8608 (vectorizable_reduction): Likewise.
8609 (vect_create_epilog_for_reduction): Likewise.
8611 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
8613 * match.pd (A / (1 << B) -> A >> B): New.
8614 * generic-match-head.c: Include optabs-tree.h.
8615 * gimple-match-head.c: Likewise.
8616 * optabs-tree.h (target_supports_op_p): New.
8617 * optabs-tree.c (target_supports_op_p): New.
8619 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8621 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
8622 $gcc_cv_ld --help output.
8623 (gcc_cv_ld_demangle): Likewise.
8624 (gcc_cv_ld_eh_frame_hdr): Likewise.
8625 (gcc_cv_ld_pie): Likewise.
8626 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
8627 (gcc_cv_ld_buildid): Likewise.
8628 (gcc_cv_ld_sysroot): Likewise.
8629 (ld_bndplt_support): Likewise.
8630 (ld_pushpopstate_support): Likewise.
8631 * configure: Regenerate.
8632 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
8634 2017-06-21 Jakub Jelinek <jakub@redhat.com>
8637 * config/i386/sse.md (round<mode>2): Renumber match_dup and
8638 operands indexes to avoid gap between operands and match_dups.
8640 2017-06-21 Andrew Pinski <apinski@cavium.com>
8642 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
8643 Increment Arith_shift and Arith_shift_reg by 1.
8644 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
8646 * config/aarch64/aarch64.c (thunderx_tunings): Enable
8647 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
8648 (aarch64_strip_extend): Add new argument and test for it.
8649 (aarch64_cheap_mult_shift_p): New function.
8650 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
8651 add a cost if it is true.
8652 Update calls to aarch64_strip_extend.
8653 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
8655 2017-06-21 Andrew Pinski <apinski@cavium.com>
8657 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
8659 (thunderxt88): Likewise.
8660 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
8661 (thunderx_prefetch_tune): New variable.
8662 (thunderx2t99_prefetch_tune): Update for the correct values.
8663 (thunderxt88_tunings): New variable.
8664 (thunderx_tunings): Use thunderx_prefetch_tune instead of
8665 generic_prefetch_tune.
8666 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
8668 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8670 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
8671 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
8672 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
8673 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
8674 (aarch64_atomic_cas<mode>, GPI): Likewise.
8676 2017-06-21 Martin Liska <mliska@suse.cz>
8678 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
8679 statements on cold and hot labels.
8680 * predict.c (tree_estimate_probability_bb): Remove the
8681 prediction from this place.
8683 2017-06-21 Martin Liska <mliska@suse.cz>
8685 PR tree-optimization/79489
8686 * gimplify.c (maybe_add_early_return_predict_stmt): New
8688 (gimplify_return_expr): Call the function.
8689 * predict.c (tree_estimate_probability_bb): Remove handling
8691 * predict.def: Update comment about early return predictor.
8692 * gimple-predict.h (is_gimple_predict): New function.
8693 * predict.def: Change default value of early return to 66.
8694 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
8696 * passes.def: Put pass_strip_predict_hints to the beginning of
8699 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
8701 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
8702 FUNCTION_DECL declarations.
8703 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
8705 (dwaf2out_decl): Likewise.
8706 * godump.c (go_early_global_decl): Skip call to the real debug hook
8707 for FUNCTION_DECL declarations.
8708 * passes.c (rest_of_decl_compilation): Skip call to the
8709 early_global_decl debug hook for FUNCTION_DECL declarations, unless
8710 -fdump-go-spec is passed.
8712 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
8714 * config/i386/i386.c (struct builtin_isa): New field pure_p.
8715 Reorder for compactness.
8716 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
8717 (def_builtin_pure, def_builtin_pure2): New functions.
8718 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
8720 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
8722 * match.pd (nop_convert): New predicate.
8723 ((A +- CST1) +- CST2): Allow some NOP conversions.
8725 2017-06-21 Jakub Jelinek <jakub@redhat.com>
8728 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
8729 with ctors/dtors if GOVD_SHARED is set.
8731 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8733 * config/aarch64/aarch64.md (movti_aarch64):
8734 Emit mov rather than orr.
8735 (movtf_aarch64): Likewise.
8736 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
8737 Emit mov rather than orr.
8739 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8741 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
8742 Swap alternatives, make integer dup more expensive.
8744 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
8746 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
8747 Return true for non-tls symbols.
8749 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
8751 * config/aarch64/aarch64-cores.def (cortex-a55): New.
8752 (cortex-a75): Likewise.
8753 (cortex-a75.cortex-a55): Likewise.
8754 * config/aarch64/aarch64-tune.md: Regenerate.
8755 * doc/invoke.texi (-mtune): Document new values for -mtune.
8757 2017-06-21 Tom de Vries <tom@codesourcery.com>
8759 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
8761 (Effective-Target Keywords, Other attributes): Suggest using
8762 dg-add-options stack_size feature to get stack limit in stack_size
8763 effective target documentation.
8765 2017-06-21 Julian Brown <julian@codesourcery.com>
8766 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8768 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
8769 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
8770 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
8772 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
8773 attribute type list for neon_multiply.
8774 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
8775 attribute type list for neon_multiply.
8776 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
8777 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
8778 attribute type list for neon_multiply.
8779 * config/arm/types.md (crypto_pmull): Add.
8780 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
8781 attribute type list.
8783 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
8785 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
8788 2017-06-20 Jakub Jelinek <jakub@redhat.com>
8790 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
8791 to make sure not to dereference a NULL cost_classes_ptr pointer.
8793 2017-06-20 Carl Love <cel@us.ibm.com>
8795 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8796 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
8797 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
8798 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
8799 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
8800 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
8801 VMULOSW): New enum "unspec" values.
8802 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
8803 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
8804 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
8805 altivec_vmulosw): New patterns.
8806 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
8807 VMULOSW): Add definitions.
8809 2017-06-20 Julia Koval <julia.koval@intel.com>
8811 * config/i386/i386.c: Fix rounding expand for new pattern.
8812 * config/i386/subst.md: Fix pattern (parallel -> unspec).
8814 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8816 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
8817 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
8819 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8821 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
8824 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
8826 * config/aarch64/aarch64-cores.def: Rearrange to sort by
8827 architecture, then by implementer ID.
8828 * config/aarch64/aarch64-tune.md: Regenerate.
8830 2017-06-20 Richard Biener <rguenther@suse.de>
8833 * fold-const.c (split_tree): Fold to type before negating.
8835 2017-06-20 David Malcolm <dmalcolm@redhat.com>
8837 * diagnostic-show-locus.c
8838 (selftest::test_fixit_deletion_affecting_newline): New function.
8839 (selftest::diagnostic_show_locus_c_tests): Call it.
8841 2017-06-20 Andreas Schwab <schwab@suse.de>
8844 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
8847 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8849 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
8850 __ARM_FEATURE_COPROC according to support.
8852 2017-06-20 Jakub Jelinek <jakub@redhat.com>
8854 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
8855 Rewritten to avoid overflow for > 32-bit pointers.
8858 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
8859 by removing enum keyword.
8860 (ubsan_type_descriptor): Likewise. Formatting fix.
8863 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
8864 splitter): Require TARGET_SSE2 in the condition.
8866 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
8869 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
8870 for doing vector set of SFmode on ISA 3.0.
8871 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
8872 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
8874 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
8875 SFmode value into a V4SF variable that was extracted from another
8876 V4SF variable without converting the element to double precision
8877 and back to single precision vector format.
8878 (vsx_insert_extract_v4sf_p9_2): Likewise.
8880 2017-06-19 Jakub Jelinek <jakub@redhat.com>
8882 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
8883 in UWHI to avoid undefined overflow.
8886 * ubsan.h (enum ubsan_encode_value_phase): New.
8887 (ubsan_encode_value): Change second argument to
8888 enum ubsan_encode_value_phase with default value of
8889 UBSAN_ENCODE_VALUE_GENERIC.
8890 * ubsan.c (ubsan_encode_value): Change second argument to
8891 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
8892 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
8893 create_tmp_var_raw instead of create_tmp_var and use a
8895 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
8896 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
8897 ubsan_encode_value callers.
8900 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
8901 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
8902 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
8904 2017-06-19 Richard Biener <rguenther@suse.de>
8907 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
8908 estimates if we changed anything.
8910 2017-06-19 Richard Biener <rguenther@suse.de>
8912 PR tree-optimization/80887
8913 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
8914 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
8915 simplified lookups, then reset mprts_hook.
8916 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
8918 (try_to_simplify): Likewise.
8920 2017-06-19 Martin Liska <mliska@suse.cz>
8923 * gimplify.c (gimplify_switch_expr):
8924 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
8926 2017-06-19 Martin Liska <mliska@suse.cz>
8928 * doc/install.texi: Document that PGO runs in 4 stages.
8930 2017-06-19 Martin Liska <mliska@suse.cz>
8933 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
8934 to dispatcher function name.
8935 * multiple_target.c (replace_function_decl): New function.
8936 (create_dispatcher_calls): Redirect both edges and references.
8938 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
8940 * profile-count.c (profile_count::dump): Dump quality.
8941 (profile_count::differs_from_p): Update for unsigned val.
8942 * profile-count.h (profile_count_quality): New enum.
8943 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
8945 2017-06-19 Richard Biener <rguenther@suse.de>
8947 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
8948 struct function as arg.
8949 (estimate_numbers_of_iterations): Export overload with loop arg.
8950 (free_numbers_of_iterations_estimates_loop): Use an overload of
8951 free_numbers_of_iterations_estimates instead.
8952 * tree-cfg.c (remove_bb): Adjust.
8953 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
8954 * tree-parloops.c (gen_parallel_loop): Likewise.
8955 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
8957 (tree_unroll_loops_completely): Likewise.
8958 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
8959 Use an overload instead and export.
8960 (estimated_loop_iterations): Adjust.
8961 (max_loop_iterations): Likewise.
8962 (likely_max_loop_iterations): Likewise.
8963 (estimate_numbers_of_iterations): Take struct function as arg
8965 (loop_exits_before_overflow): Adjust.
8966 (free_numbers_of_iterations_estimates_loop): Use an overload.
8967 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
8968 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
8970 2017-06-19 Richard Biener <rguenther@suse.de>
8973 * ipa-prop.c (find_constructor_constant_at_offset): Handle
8974 RANGE_EXPR conservatively.
8976 2017-06-16 Carl Love <cel@us.ibm.com>
8978 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8979 definitions for vec_float, vec_float2, vec_floato,
8980 vec_floate built-ins.
8981 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
8982 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
8984 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
8985 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
8986 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
8987 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
8988 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
8989 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
8990 vec_floato): Add builtin defines.
8991 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
8992 Update the built-in documentation file for the new built-in
8995 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
8997 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
8998 (mthumb): Mark as the negative of -marm.
9000 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9002 * doc/invoke.texi (ARM Options, -mcpu): Document supported
9004 (ARM Options, -mtune): Document that this accepts the same
9005 extension options as -mcpu.
9006 (ARM Options, -mfpu): Document addition of -mfpu=auto.
9008 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9010 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
9011 permitted extensions.
9013 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9015 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
9016 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
9017 (armv8-m.main): Add option +nodsp.
9018 * config/arm/arm-cpu-cdata.h: Regenerated.
9020 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9022 * config/arm/t-fuchsia: New file.
9023 * config.gcc (arm*-*-fuchsia*): Use it.
9025 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9027 * config/arm/t-symbian: Rewrite for new option infrastructure.
9029 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9031 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
9032 (MULTILIB_REQUIRED): Likewise.
9034 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9036 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
9037 (MULTILIB_RESUE): Likewise.
9038 (MULTILIB_MATCHES): Likewise.
9039 (MULTLIB_REQUIRED): Likewise.
9041 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9043 * config/arm/t-rtems: Rewrite for new option framework.
9045 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9047 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
9048 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
9049 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
9050 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
9051 * config/arm/t-multilib: ... here.
9052 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
9053 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
9054 armv7-a and armv8*-a when A-profile libraries have not been built.
9055 * config/arm/t-rmprofile: Rewrite.
9057 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9059 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
9060 with a backslash. Remove the backslash after substituting unescaped
9062 * doc/fragments.texi (MULTILIB_REUSE): Document it.
9064 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9066 * config.gcc: (arm*-*-*): When building a-profile libraries, force
9067 the driver to pass through the default setting of -mfloat-abi.
9068 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
9070 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
9071 (all_feat_combs): New rule.
9072 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
9074 * config/arm/t-aprofile: Rewrite.
9076 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9078 * config/arm/arm.h (FPUTYPE_AUTO): Define.
9079 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
9080 fpu is not specified by the user/command-line.
9081 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
9082 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
9083 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
9084 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
9085 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
9086 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
9088 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9090 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
9091 * config/arm/t-arm-elf: Rewritten.
9093 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9095 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
9096 have some floating-point instructions.
9097 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
9098 (TARGET_MAYBE_HARD_FLOAT): New macro.
9099 * config/arm/arm-builtins.c (arm_init_builtins): Use
9100 TARGET_MAYBE_HARD_FLOAT.
9101 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
9103 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9105 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
9106 (configargs.h): Include it.
9107 (arm_print_hint_for_fpu_option): New function.
9108 (arm_parse_fpu_option): New function.
9109 (candidate_extension): New class.
9110 (arm_canon_for_multilib): New function.
9111 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
9112 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
9113 (ARCH_CANONICAL_SPECS): New macro.
9114 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
9116 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9118 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
9119 are set after handling multilib fragments. Set target_cpu_default2
9122 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9124 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
9126 (arm*-*-*): Set target_cpu_default2 to a quoted string.
9127 * config/arm/parsecpu.awk (check_cpu): Validate any extension
9129 (check_arch): Likewise.
9130 * config/arm/arm.c (arm_configure_build_target): Handle
9131 TARGET_CPU_DEFAULT being a string constant. Scan any feature
9132 options in the default.
9134 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9136 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
9137 when an option is an alias of another.
9138 * config/arm/parsecpu.awk (optalias): New parser token.
9139 (gen_comm_data): Mark non-alias options as such. Emit entries
9140 for extension aliases.
9141 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
9142 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
9143 (armv6kz, armv6zk, armv6t2): Likewise.
9144 (armv7): Make vfpv3-d16 an alias.
9145 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
9147 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
9148 Sort in canonical order.
9149 (armv8-a): Sort in canonical order.
9150 (armv8.1-a, armv8.2-a): Likewise.
9151 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
9153 (cortex-a9): Sort in canonical order.
9154 * config/arm/arm.c (selftests.h): Include it.
9155 (arm_test_cpu_arch_data): New function.
9156 (arm_run_self_tests): New function.
9157 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
9158 (targetm): Move declaration to the end of the file.
9159 * arm-cpu-cdata.h: Regenerated.
9161 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9163 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
9164 call to target_mode_check describing the type of option passed.
9165 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
9166 (arm_target_thumb_only): Use arm_parse_arch_option_name or
9167 arm_parse_cpu_option_name to match parameters against list of
9169 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
9170 arm_arch_core_flags data structure.
9171 * config/arm/arm-cpu_cdata.h: Regenerated.
9173 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9175 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
9177 (arm_print_hint_for_cpu_option): Likewise.
9178 (arm_print_hint_for_arch_option): Likewise.
9179 (arm_parse_cpu_option_name): Likewise.
9180 (arm_parse_arch_option_name): Likewise.
9181 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
9182 of entries in the all_fpus list.
9183 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
9184 (arm_parse_cpu_option_name): Declare.
9185 (arm_parse_arch_option_name): Declare.
9186 (arm_parse_option_features): Declare.
9187 (arm_intialize_isa): Declare.
9188 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
9190 (gen_comm_data): ... here. Make definitions non-static.
9191 * config/arm/arm-cpu-data.h: Regenerated.
9192 * config/arm/arm-cpu-cdata.h: Regenerated.
9194 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9196 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
9197 (cpu_arch_extension): New structure.
9198 (cpu_arch_option, arch_option, cpu_option): New structures.
9199 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
9201 (gen_data): Generate new format data tables.
9202 * config/arm/arm.c (cpu_tune): New structure.
9203 (cpu_option, processors): Delete.
9204 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
9205 (arm_print_hint_for_cpu_option): ... this and ...
9206 (arm_print_hint_for_arch_option): ... this.
9207 (arm_parse_arch_cpu_name): Delete. Replace with ...
9208 (arm_parse_cpu_option_name): ... this and ...
9209 (arm_parse_arch_option_name): ... this.
9210 (arm_unrecognized_feature): Change type of target parameter to
9212 (arm_parse_arch_cpu_features): Delete. Replace with ...
9213 (arm_parse_option_features): ... this.
9214 (arm_configure_build_target): Rework to use new configuration data
9216 (arm_print_tune_info): Rework for new configuration data tables.
9217 * config/arm/arm-cpu-data.h: Regenerated.
9218 * config/arm/arm-cpu.h: Regenerated.
9220 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9222 * Makefile.in (OBJS): Move sbitmap.o from here ...
9223 (OBJS-libcommon): ... to here.
9225 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9227 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
9228 (ISA_ALL_CRYPTO): New macro.
9229 (ISA_ALL_SIMD): New macro
9230 (ISA_ALL_FP): New macro.
9231 * config/arm/arm.c (fpu_bitlist): Update initializer.
9232 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
9234 (arm9e): Add fpu. Add option for nofp
9235 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
9236 (arm926ej-s, arm1026ej-s): Likewise.
9237 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
9238 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
9239 neon-fp16, neon-vfpv4, nofp and nosimd.
9240 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
9241 (cortex-a8): Add fpu. Add option for nofp.
9242 (cortex-a9): Add fpu. Add options for nosimd and nofp.
9243 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
9244 (cortex-r4f): Add fpu.
9245 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
9246 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
9248 (cortex-r8): Likewise.
9249 (cortex-m4): Add fpu. Add option for nofp.
9250 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
9251 (cortex-a17.cortex-a7): Likewise.
9252 (cortex-a32): Add fpu. Add options for crypto and nofp.
9253 (cortex-a35, cortex-a53): Likewise.
9254 (cortex-a57): Add fpu. Add option for crypto.
9255 (cortex-a72, cortex-a73): Likewise.
9256 (exynos-m1): Likewise.
9257 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
9258 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
9259 (cortex-m33): Add fpu. Add option for nofp.
9260 * config/arm/arm-cpu-cdata.h: Regenerated
9261 * config/arm/arm-cpu-data.h: Regenerated.
9263 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9265 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
9266 (armv5te, armv5tej): Likewise.
9267 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
9268 (armv7): Add options fp and vfpv3-d16.
9269 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
9270 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
9272 (armv7ve): Likewise.
9273 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
9274 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
9275 (armv8-a): Add nocrypto option.
9276 (armv8.1-a, armv8.2-a): Likewise.
9277 (armv8-m.main): add options fp, fp.dp and nofp.
9279 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9281 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
9283 (armv8-a+crc): Delete.
9284 (armv8.1-a): Add options simd, crypto and nofp.
9285 (armv8.2-a): Add options fp16, simd, crypto and nofp.
9286 (armv8.2-a+fp16): Delete.
9287 (armv8-m.main): Add option dsp.
9288 (armv8-m.main+dsp): Delete.
9289 (cortex-a8): Add fpu. Add nofp option.
9290 (cortex-a9): Add fpu. Add nofp and nosimd options.
9291 * config/arm/parsecpu.awk (gen_data): Generate option tables and
9292 link to main cpu and architecture data structures.
9293 (gen_comm_data): Only put isa attributes from the main architecture
9295 (option): New statement for architecture and CPU entries.
9296 * arm.c (struct cpu_option): New structure.
9297 (struct processors): Add entry for options.
9298 (arm_unrecognized_feature): New function.
9299 (arm_parse_arch_cpu_name): Ignore any characters after the first
9301 (arm_parse_arch_cpu_feature): New function.
9302 (arm_configure_build_target): Separate out any CPU and architecture
9303 features and parse separately. Don't error out if -mfpu=auto is
9304 used with only an architecture string.
9305 (arm_print_asm_arch_directives): New function.
9306 (arm_file_start): Call it.
9307 * config/arm/arm-cpu-cdata.h: Regenerated.
9308 * config/arm/arm-cpu-data.h: Likewise.
9309 * config/arm/arm-tables.opt: Likewise.
9311 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9313 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
9314 assembler when it is not -mfpu=auto.
9316 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9318 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
9319 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
9320 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
9321 (ASM_CPU_SPEC): Rewrite.
9322 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
9323 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
9324 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
9325 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
9326 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
9327 copied string is NUL-terminated. Also strip any characters prefixed
9329 (arm_rewrite_selected_arch): New function.
9330 (arm_rewrite_march): New function.
9332 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9334 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
9335 (x_arm_cpu_string, x_arm_tune_string): Likewise.
9336 (march, mcpu, mtune): Convert to string-based options.
9337 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
9338 (arm_parse_arch_cpu_name): New function.
9339 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
9340 identify selected architecture or CPU.
9341 (arm_option_save): New function.
9342 (TARGET_OPTION_SAVE): Redefine.
9343 (arm_option_restore): Restore string options.
9344 (arm_option_print): Print string options.
9346 2017-06-16 Martin Sebor <msebor@redhat.com>
9348 PR tree-optimization/80933
9349 PR tree-optimization/80934
9350 * builtins.c (fold_builtin_3): Do not handle bcmp here.
9351 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
9352 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
9353 (gimple_fold_builtin): Call them.
9355 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9357 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
9358 as unlikely; update profile.
9360 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9362 * predict.c (force_edge_cold): Handle declaring edges impossible
9365 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9367 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
9369 (try_unroll_loop_completely): Fix reporting.
9371 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9373 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
9375 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
9378 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
9379 if given a non-constant argument for an intrinsic which requires a
9382 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9384 * profile.c (compare_freqs): New function.
9385 (branch_prob): Sort edge list.
9386 (find_spanning_tree): Assume that the list is priority sorted.
9388 2017-06-16 Richard Biener <rguenther@suse.de>
9390 PR tree-optimization/81090
9391 * passes.def (pass_record_bounds): Remove.
9392 * tree-pass.h (make_pass_record_bounds): Likewise.
9393 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
9394 make_pass_record_bounds): Likewise.
9395 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
9396 not free niter estimates at the beginning but at the end.
9397 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
9399 2017-06-16 Richard Biener <rguenther@suse.de>
9401 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
9402 initializer to workaround ICE in host GCC 4.8.
9404 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9406 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
9408 (clone_inlined_nodes): Update.
9410 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9412 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
9413 prefetch settings, and enable prefetching by default at -O3.
9415 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9417 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9418 Set flag_prefetch_loop_arrays according to tuning data.
9420 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9422 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
9424 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
9425 [Unrelated to main purpose of the patch] Place the pointer field last
9426 to enable type checking errors when tune structure are wrongly merged.
9427 * config/aarch64/aarch64.c (generic_prefetch_tune,)
9428 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
9429 (thunderx2t99_prefetch_tune): New tune constants.
9430 (tune_params *_tunings): Update all tunings (no functional change).
9431 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
9432 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
9433 from tunings structures.
9435 2017-06-16 Jakub Jelinek <jakub@redhat.com>
9438 * ubsan.c (instrument_null): Add T argument, use it instead
9439 of computing it based on IS_LHS.
9440 (instrument_object_size): Likewise.
9441 (pass_ubsan::execute): Adjust instrument_null and
9442 instrument_object_size callers to pass gimple_get_lhs or
9443 gimple_assign_rhs1 result to it. Use instrument_null instead of
9444 calling get_base_address and instrument_mem_ref. Handle
9445 aggregate call arguments for object-size sanitization.
9447 2017-06-16 Yury Gribov <tetra2005@gmail.com>
9449 PR tree-optimization/81089
9450 * tree-vrp.c (is_masked_range_test): Validate operands of
9453 2017-06-15 Martin Sebor <msebor@redhat.com>
9456 * dumpfile.c (dump_register): Avoid calling memset to initialize
9457 a class with a default ctor.
9458 * gcc.c (struct compiler): Remove const qualification.
9459 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
9460 * hash-table.h: Ditto.
9461 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
9463 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
9464 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
9466 * params.h (struct param_info): Make struct members non-const.
9467 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
9468 with default initialization.
9469 * vec.h (vec_copy_construct, vec_default_construct): New helper
9471 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
9472 with vec_copy_construct.
9473 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
9474 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
9475 * doc/invoke.texi (-Wclass-memaccess): Document.
9477 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9479 * emit-rtl.h (is_leaf): Update comment about local
9482 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9485 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
9486 for a variable to have a section before checking if the section has a
9488 Set section to.persistent if persistent attribute is set.
9489 Warn if .persistent attribute is used on an automatic variable.
9491 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
9493 PR rtl-optimization/80474
9494 * reorg.c (update_block): Do not ignore instructions in a delay slot.
9496 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
9498 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
9501 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
9503 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
9504 (casesi): Emit bounds checking as RTL.
9505 (casesi_internal_mips16_<mode>): Remove bounds checking.
9507 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
9509 * config/xtensa/xtensa.c (xtensa_option_override): Append
9510 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
9511 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
9512 xtensa_doloop_hooks): Define unconditionally.
9513 (xtensa_reorg_loops): Only call reorg_loops in the presence of
9515 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
9516 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
9517 for it in xtensa_option_override.
9518 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
9519 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
9521 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
9523 * doc/cppopts.texi: Document '-' special value to -MF.
9525 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
9527 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
9528 (cortex_a53_fconst): Likewise.
9529 (cortex_a53_fpmul): Likewise.
9530 (cortex_a53_f_load_64): Likewise.
9531 (cortex_a53_f_load_many): Likewise.
9532 (cortex_a53_advsimd_alu): Likewise.
9533 (cortex_a53_advsimd_alu_q): Likewise.
9534 (cortex_a53_advsimd_mul): Likewise.
9535 (cortex_a53_advsimd_mul_q): Likewise.
9536 (fpmac bypass): Add new bypass for fpmac-fpmac case.
9537 Add missing fmul, r2f_cvt and fconst cases.
9539 2017-06-14 Richard Biener <rguenther@suse.de>
9542 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
9544 (fold_binary_loc): When associating do not treat pre-existing
9545 TREE_OVERFLOW on literal constants as a reason to allow
9546 TREE_OVERFLOW on associated literal constants.
9548 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9550 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
9551 (MASK_FEATURES): New macro.
9552 * config/sparc/sparc.c (sparc_option_override): Remove the special
9553 handling of -mfpu and generalize it to all MASK_FEATURES switches.
9555 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9557 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
9558 a division of 0 if non-call exceptions are enabled.
9560 2017-06-14 Andrew Pinski <apinski@cavium.com>
9561 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9564 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
9565 Improve vector initialization code gen for only variable case.
9567 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9569 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
9571 2017-06-14 Richard Biener <rguenther@suse.de>
9573 PR tree-optimization/81083
9574 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
9577 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9579 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
9580 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
9581 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
9582 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
9583 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
9584 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
9586 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9588 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
9589 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
9591 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9593 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
9595 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9597 * config/rs6000/t-rtems: Don't handle SPE.
9599 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9601 * config/rs6000/t-linux: Don't handle SPE.
9603 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9605 * config/rs6000/eabispe.h: Delete file.
9607 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9609 * config/rs6000/t-spe: Delete file.
9611 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9613 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
9614 (rs6000_legitimate_offset_address_p): Return false for anything in
9615 V2SImode or V2SFmode.
9617 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
9619 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
9620 except V2SF and V2SI. Rearrange the vector modes, and add comments.
9621 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
9623 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
9624 (rs6000_legitimate_offset_address_p): Ditto.
9625 (rs6000_emit_move): Ditto.
9626 (rs6000_init_builtins): Remove V4HI_type_node.
9628 2017-06-13 Martin Liska <mliska@suse.cz>
9631 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
9632 (gate_asan): Likewise.
9633 * asan.h (asan_no_sanitize_address_p): Remove the function.
9634 (sanitize_flags_p): New function.
9635 * builtins.def: Fix coding style.
9636 * common.opt: Use renamed enum value.
9637 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
9638 * doc/extend.texi: Document no_sanitize attribute.
9639 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
9640 to SANITIZE_UNDEFINED_NONDEFAULT.
9641 * gcc.c (sanitize_spec_function): Use the renamed enum value.
9642 * gimple-fold.c (optimize_atomic_compare_exchange_p):
9643 Use sanitize_flags_p.
9644 * gimplify.c (gimplify_function_tree): Likewise.
9645 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
9646 * opts.c (parse_no_sanitize_attribute): New function.
9647 (common_handle_option): Use renamed enum value.
9648 * opts.h (parse_no_sanitize_attribute): Declare.
9649 * tree.c (sanitize_flags_p): New function.
9650 * tree.h: Declared here.
9651 * tsan.c: Use sanitize_flags_p.
9652 * ubsan.c (ubsan_expand_null_ifn): Likewise.
9653 (instrument_mem_ref): Likewise.
9654 (instrument_bool_enum_load): Likewise.
9655 (do_ubsan_in_current_function): Remove the function.
9656 (pass_ubsan::execute): Use sanitize_flags_p.
9657 * ubsan.h: Remove do_ubsan_in_current_function
9658 * tree-cfg.c (print_no_sanitize_attr_value): New function.
9659 (dump_function_to_file): Use it here.
9661 2017-06-13 Martin Jambor <mjambor@suse.cz>
9663 PR tree-optimization/80803
9664 PR tree-optimization/81063
9665 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
9666 (propagate_subaccesses_across_link): Enqueue subtree whenever
9667 necessary instead of relying on the caller.
9669 2017-06-13 Martin Jambor <mjambor@suse.cz>
9671 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
9672 that have a first_link.
9673 (sort_and_splice_var_accesses): Do not check first_link before
9675 (subtree_mark_written_and_enqueue): Likewise.
9676 (propagate_all_subaccesses): Likewise and do not stop at first
9677 parent with a first_link.
9679 2017-06-13 Martin Jambor <mjambor@suse.cz>
9681 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
9684 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9686 * match.pd: New pattern.
9688 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9690 * tree-vrp.c (is_masked_range_test): New function.
9691 (register_edge_assert_for): Determine ranges for
9694 2017-06-13 Yury Gribov <tetra2005@gmail.com>
9696 PR tree-optimization/67328
9697 * fold-const.c (maskable_range_p): New function.
9698 (build_range_check): Generate bittests if possible.
9700 2017-06-13 Martin Liska <mliska@suse.cz>
9702 * gimple-pretty-print.c (dump_probability): Add new argument.
9703 (dump_edge_probability): Dump both probability and count.
9704 (dump_gimple_label): Likewise.
9705 (dump_gimple_bb_header): Likewise.
9707 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
9710 * config/avr/avr-devices.c: Fix indentation.
9711 * config/avr/gen-avr-mmcu-specs.c: Dito.
9713 2017-06-13 Richard Biener <rguenther@suse.de>
9715 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
9716 instead get vector type from stmt_info.
9717 (vectorizable_reduction): Adjust. Remove dead code.
9719 2017-06-13 Richard Biener <rguenther@suse.de>
9722 * fold-const.c (extract_muldiv_1): Remove bogus distribution
9723 case of C * (x * C2 + C3).
9724 (fold_addr_of_array_ref_difference): Properly fold index difference.
9726 2017-06-12 David S. Miller <davem@davemloft.net>
9729 * config/sparc/sparc.md (return expander): Emit frame blockage if
9730 function uses alloca.
9732 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9734 * combine.c (make_field_assignment): Check len rather than the mode
9735 precision when calling force_to_mode.
9737 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
9739 Support multilibs and devices that see flash in RAM address range.
9742 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
9743 (avr_mcu_t) <flash_pm_offset>: New field.
9744 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
9745 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
9746 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
9747 (AVR_TINY_PM_OFFSET): Remove macro.
9748 * config/avr/avr.opt (-mshort-calls): New option.
9749 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
9750 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
9751 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
9752 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
9753 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
9754 instead of avr_arch->have_jmp_call.
9755 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
9756 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
9757 avr_arch->flash_pm_offset to define.
9758 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
9759 new field flash_pm_offset. Add entry for avrxmega3.
9760 (avr_texinfo): Add entry for avrxmega3.
9761 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
9762 attiny212, attiny214,
9763 attiny412, attiny414, attiny416, attiny417,
9764 attiny814, attiny816, attiny817,
9765 attiny1614, attiny1616, attiny1617,
9766 attiny3214, attiny3216, attiny3217.
9767 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
9768 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
9769 (avr_print_operand_address) [AVR_TINY]: Same.
9770 (avr_asm_init_sections) <readonly_data_section>: Only patch
9771 callback if avr_arch->flash_pm_offset = 0.
9772 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
9773 for rodata if avr_arch->flash_pm_offset != 0.
9774 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
9775 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
9776 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
9777 (m_options): Append opt_rcall.
9778 (m_dirnames): Append dir_rcall.
9779 * config/avr/t-multilib: Regenerate.
9781 * configure.ac [target=avr]: Check whether avrxmega3 default
9782 linker description file works as needed.
9783 * configure: Regenerate.
9784 * doc/avr-mmcu.texi: Regenerate.
9785 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
9786 <__AVR_ARCH__>: Document avrxmega3 and 103.
9787 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
9788 <__AVR_SHORT_CALLS__>: Document it.
9789 <__AVR_PM_BASE_ADDRESS__>: Document it.
9790 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
9791 (AVR Variable Attributes) <progmem>: Document this is
9792 not needed for avrxmega3.
9793 (AVR Named Address Spaces) <__flash>: Dito.
9795 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
9797 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
9799 2017-06-12 Doug Rupp <rupp@adacore.com>
9801 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
9802 Append vxworks-stdint.h to the tm_file list.
9803 * config/vxworks-stdint.h: New file.
9805 2017-06-12 Martin Liska <mliska@suse.cz>
9807 PR tree-optimization/81041
9808 * tree-profile.c (gimple_gen_ic_func_profiler):
9809 Create an extra BB in profile-generate
9810 (gimple_gen_time_profiler): Likewise.
9812 2017-06-12 Jakub Jelinek <jakub@redhat.com>
9814 PR tree-optimization/81003
9815 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
9816 (update_range_test): Use it instead of force_gimple_operand_gsi.
9818 2017-06-12 Richard Biener <rguenther@suse.de>
9820 PR tree-optimization/81053
9821 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
9822 with backedge value not defined in loop. Simplify def stmt
9825 2017-06-11 Tom de Vries <tom@codesourcery.com>
9828 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
9830 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
9831 nvptx_cannot_force_const_mem.
9833 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9835 * opts.c (finish_options): Move test for flag_split_stack after
9836 it has been initialized.
9838 2017-06-11 Jason Merrill <jason@redhat.com>
9840 * tree.h (id_equal): New.
9841 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
9842 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
9843 instead of strcmp of IDENTIFIER_POINTER.
9845 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9847 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
9848 (mark_all_inlined_calls_cdtor): Fix formating.
9849 (inline_transform): Rescale profile before inlining.
9851 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9853 * cgraph.h (cgraph_edge::clone): Update prototype.
9854 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
9855 (cgraph_node::create_clone): Update.
9856 (cgraph_node::create_version_clone): Update.
9857 * tree-inline.c (copy_bb): Update.
9858 (expand_call_inline): Update.
9860 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
9862 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
9863 factored out from ...
9864 (rs6000_emit_prologue): ... here.
9866 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
9868 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
9869 factored out from ...
9870 (rs6000_emit_prologue): ... here.
9872 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9874 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
9876 (handle_missing_profiles): Fix computation of tp_first_run.
9877 (counts_to_freqs): Do not touch freqs when count is 0.
9879 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
9881 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
9884 2017-06-10 Tom de Vries <tom@codesourcery.com>
9886 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
9887 attributes): Document signal effective target.
9889 2017-06-10 Tom de Vries <tom@codesourcery.com>
9891 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
9892 Document effective target stack_size.
9894 2017-06-09 David Malcolm <dmalcolm@redhat.com>
9896 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
9897 to the edit_context if they can be auto-applied.
9899 2017-06-9 Ian Lance Taylor <iant@golang.org>
9901 * opts.c (finish_options): If -fsplit-stack, disable implicit
9902 -forder-blocks-and-partition.
9903 * doc/invoke.texi (Optimize Options): Document that when using
9904 -fsplit-stack -forder-blocks-and-partition is not implicitly
9907 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9909 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
9910 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
9911 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
9912 * builtins.def (abort, trap, unreachable): Declare cold.
9913 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
9914 * tree-core.h (ECF_COLD): New.
9915 * tree.c (set_call_expr_flags): Handle ECF_COLD.
9916 (build_common_builtin_nodes): Mark unreachable and abort as cold.
9918 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9920 * predict.c (unlikely_executed_stmt_p): Cleanup.
9922 2017-06-09 Richard Biener <rguenther@suse.de>
9924 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
9925 model if the ref is always written to.
9927 2017-06-09 Tamar Christina <tamar.christina@arm.com>
9929 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
9931 2017-06-09 Tamar Christina <tamar.christina@arm.com>
9933 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
9936 2017-06-09 Tom de Vries <tom@codesourcery.com>
9939 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
9940 "target cannot support label values" when encountering LABEL_REF.
9942 2017-06-09 Martin Liska <mliska@suse.cz>
9944 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
9945 (gimple_gen_ic_func_profiler): Emit direct comparison
9946 of __gcov_indirect_call_callee with NULL.
9947 (gimple_gen_time_profiler): Change probability from
9948 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
9950 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9952 * profile.c (edge_gcov_counts): Turn to pointer.
9953 (compute_branch_probabilities, compute_branch_probabilities): Update.
9954 (branch_prob): Do not clear edge_gcov_count.
9955 * profile.h (edge_gcov_counts): Turn to pointer.
9956 (edge_gcov_count): Update.
9958 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
9960 * gimple.h (gimple_check_failed): Mark cold.
9962 2017-06-09 Richard Biener <rguenther@suse.de>
9964 PR tree-optimization/66623
9965 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
9966 refactor check_reduction into two parts, properly computing
9967 whether we have to check reduction validity for outer loop
9970 2017-06-09 Richard Biener <rguenther@suse.de>
9972 PR tree-optimization/79483
9973 * graphite-scop-detection.c (order): New global.
9974 (get_order): Compute bb to order mapping that satisfies code
9975 generation constraints.
9976 (cmp_pbbs): New helper.
9977 (build_scops): Start domwalk at entry block, sort generated
9980 2017-06-09 Richard Biener <rguenther@suse.de>
9983 * ipa-polymorphic-call.c
9984 (ipa_polymorphic_call_context::restrict_to_inner_class):
9985 Skip FIELD_DECLs with error_mark_node type.
9986 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
9989 2017-06-09 Martin Liska <mliska@suse.cz>
9991 * predict.c (struct branch_predictor): New struct.
9992 (test_prediction_value_range): New test.
9993 (predict_c_tests): New function.
9994 * selftest-run-tests.c (selftest::run_tests): Run the function.
9995 * selftest.h: Declare new tests.
9997 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
10000 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
10001 gen_add3_insn did not fail.
10002 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
10003 r0, construct that number in a temporary reg and add that reg to r0.
10004 If asked to put the result in r0 as well, fail.
10006 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
10008 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10009 for early expansion of vec_eqv.
10011 2017-06-08 Jakub Jelinek <jakub@redhat.com>
10013 PR middle-end/81005
10014 * ubsan.c (instrument_null): Avoid pointless code temporary.
10015 (pass_ubsan::execute): Instrument aggregate arguments of calls.
10017 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
10021 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
10024 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
10027 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10029 * predict.c (unlikely_executed_edge_p): Move ahead.
10030 (probably_never_executed_edge_p): Use it.
10032 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10034 PR middle-end/79988
10035 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
10036 gimple_call_builtin_p call.
10038 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10040 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
10041 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
10042 rtl_check_failed_type2, rtl_check_failed_code1,
10043 rtl_check_failed_code2, rtl_check_failed_code_mode,
10044 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
10045 rtvec_check_failed_bounds, rtl_check_failed_flag,
10046 _fatal_insn_not_found, _fatal_insn): Likewise.
10047 * tree.h (tree_contains_struct_check_failed,
10048 tree_check_failed, tree_not_check_failed,
10049 tree_class_check_failed, tree_range_check_failed,
10050 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
10051 tree_vec_elt_check_failed, phi_node_elt_check_failed,
10052 tree_operand_check_failed, omp_clause_check_failed,
10053 omp_clause_operand_check_failed, omp_clause_range_check_failed):
10056 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10058 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
10059 flag_branch_probabilities.
10060 * ipa-inline.c (edge_badness): Likewise.
10061 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
10062 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
10063 * predict.c (maybe_hot_frequency_p): Likewise.
10064 (probably_never_executed): Likewise.
10065 * sched-ebb.c (schedule_ebbs): Likewise.
10066 * sched-rgn.c (find_single_block_region): Likewise.
10067 * tracer.c (tail_duplicate): Likewise.
10069 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10071 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
10072 longer requires x_flag_profile_use.
10074 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10076 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
10077 instead of flag_reorder_blocks_and_partition.
10078 * dbxout.c (dbxout_function_end): Likewise.
10079 * dwarf2out.c (gen_subprogram_die): Likewise.
10080 * haifa-sched.c (sched_create_recovery_edges): Likewise.
10081 * hw-doloop.c (reorg_loops): Likewise.
10082 * varasm.c (assemble_start_function,
10083 assemble_end_function): Likewise.
10084 (decide_function_section): Do not check for
10085 flag_reorder_blocks_and_partition.
10087 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10089 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10091 (chkp_get_hard_register_fake_addr_expr): Ditto.
10092 (chkp_build_addr_expr): Add check for hard reg case.
10093 (chkp_parse_array_and_component_ref): Ditto.
10094 (chkp_find_bounds_1): Ditto.
10095 (chkp_process_stmt): Don't generate bounds store for
10098 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10100 * predict.c (maybe_hot_bb_p): Do not check profile status.
10101 (maybe_hot_edge_p): Likewise.
10102 (probably_never_executed): Check for zero counts even if profile
10104 (unlikely_executed_edge_p): New function.
10105 (unlikely_executed_stmt_p): New function.
10106 (unlikely_executed_bb_p): New function.
10107 (set_even_probabilities): Use unlikely predicates.
10108 (combine_predictions_for_bb): Likewise.
10109 (predict_paths_for_bb): Likewise.
10110 (predict_paths_leading_to_edge): Likewise.
10111 (determine_unlikely_bbs): New function.
10112 (estimate_bb_frequencies): Use it.
10113 (compute_function_frequency): Use zero counts even if profile is
10115 * profile-count.h: Fix typo.
10117 2017-08-08 Julia Koval <julia.koval@intel.com>
10119 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
10120 _mm512_mask_cvtsepi16_storeu_epi8,
10121 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
10122 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
10123 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
10124 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
10125 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
10126 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
10127 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
10128 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
10129 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
10130 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
10131 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
10132 __builtin_ia32_pmovuswb256mem_mask,
10133 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
10134 __builtin_ia32_pmovwb512mem_mask): New builtins.
10136 2017-08-08 Julia Koval <julia.koval@intel.com>
10138 PR target/73350,80862
10139 * config/i386/subst.md (round): Fix round pattern.
10140 * config/i386/i386.c (ix86_erase_embedded_rounding):
10141 Fix erasing rounding for the fixed pattern.
10143 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10145 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
10147 2017-06-08 Martin Liska <mliska@suse.cz>
10149 PR gcov-profile/80911
10150 * gcov.c (block_info::block_info): New constructor.
10152 2017-06-07 Carl Love <cel@us.ibm.com>
10154 * config/rs6000/rs6000-c: The return type of the following
10155 built-in functions was implemented as int not long long. Fix sign
10156 of return value for the unsigned version of vec_mulo and vec_mule.
10157 vector unsigned long long vec_bperm (vector unsigned long long,
10158 vector unsigned char)
10159 vector signed long long vec_mule (vector signed int,
10161 vector unsigned long long vec_mule (vector unsigned int,
10162 vector unsigned int)
10163 vector signed long long vec_mulo (vector signed int,
10165 vector unsigned long long vec_mulo (vector unsigned int,
10166 vector unsigned int)
10167 * doc/extend.texi: Fix the documentation for the built-in
10170 2017-06-07 Carl Love <cel@us.ibm.com>
10173 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
10176 2017-06-07 Carl Love <cel@us.ibm.com>
10178 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
10179 support, Generate doublehv for signed int/float for BE case only.
10181 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
10183 * doc/invoke.texi (mcx16): Rewrite.
10185 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10187 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
10188 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
10189 *mov<mode>_softfloat, and an anonymous splitter): Use
10190 nonimmediate_operand instead of rs6000_nonimmediate_operand.
10192 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10194 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
10196 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
10197 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
10198 (rs6000_debug_reg_global): Adjust.
10199 (rs6000_init_hard_regno_mode_ok): Adjust.
10200 (rs6000_dbx_register_number): Adjust.
10201 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
10202 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
10203 Remove SPE_ACC and SPEFSCR.
10204 (REG_ALLOC_ORDER): Ditto.
10205 (FRAME_POINTER_REGNUM): Change to 111.
10206 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
10207 (REG_CLASS_NAMES): Ditto.
10208 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
10209 (REGISTER_NAMES): Ditto.
10210 (ADDITIONAL_REG_NAMES): Ditto.
10211 (rs6000_reg_names): Ditto.
10212 * config/rs6000/rs6000.md: Renumber some register number
10215 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10217 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
10219 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
10220 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
10222 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
10223 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
10224 Delete the SPE high registers.
10225 (REG_ALLOC_ORDER): Ditto.
10226 (enum reg_class): Remove SPE_HIGH_REGS.
10227 (REG_CLASS_NAMES): Ditto.
10228 (REG_CLASS_CONTENTS): Delete the SPE high registers.
10229 (REGISTER_NAMES): Ditto.
10230 (rs6000_reg_names): Ditto.
10231 * doc/tm.texi.in: Remove SPE as example.
10232 * doc/tm.texi: Regenerate.
10234 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10236 * config/rs6000/8540.md (ppc8540_brinc): Delete.
10237 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
10238 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
10239 * config/rs6000/rs6000.md (type): Remove "brinc".
10241 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10243 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
10244 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
10245 * config/rs6000/linuxspe.h: Delete file.
10246 * config/rs6000/rs6000.md: Don't include spe.md.
10247 * config/rs6000/spe.h: Delete file.
10248 * config/rs6000/spe.md: Delete file.
10249 * config/rs6000/t-rs6000: Remove spe.md.
10251 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10253 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
10254 (reg_or_none500mem_operand): Delete.
10255 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
10256 instead of reg_or_none500mem_operand.
10258 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10260 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
10261 handling of SPE flags.
10262 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
10264 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10266 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
10268 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
10269 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
10270 paired_divv2sf3): Similar.
10271 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
10272 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
10273 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
10275 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
10276 Rename the paired_* instruction patterns.
10277 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
10279 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
10280 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
10281 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
10282 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
10283 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
10284 PAIRED_VECTOR_MODE.
10285 (struct machine_function): Delete field spe_insn_chain_scanned_p.
10286 (spe_func_has_64bit_regs_p): Delete.
10287 (spe_expand_predicate_builtin): Delete.
10288 (spe_expand_evsel_builtin): Delete.
10289 (TARGET_DWARF_REGISTER_SPAN): Do not define.
10290 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
10291 (invalid_e500_subreg): Delete.
10292 (rs6000_legitimize_address): Always force_reg op2 as well, for
10293 paired single memory accesses.
10294 (rs6000_member_type_forces_blk): Delete.
10295 (rs6000_spe_function_arg): Delete.
10296 (rs6000_expand_unop_builtin): Delete SPE handling.
10297 (rs6000_expand_binop_builtin): Ditto.
10298 (spe_expand_stv_builtin): Delete.
10299 (bdesc_2arg_spe): Delete.
10300 (spe_expand_builtin): Delete.
10301 (spe_expand_predicate_builtin): Delete.
10302 (spe_expand_evsel_builtin): Delete.
10303 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
10304 (spe_init_builtins): Delete.
10305 (spe_func_has_64bit_regs_p): Delete.
10306 (savres_routine_name): Delete "info" parameter. Adjust callers.
10307 (rs6000_emit_stack_reset): Ditto.
10308 (rs6000_dwarf_register_span): Delete.
10309 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
10310 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
10311 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
10313 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
10315 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
10316 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
10317 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
10318 mulv2sf3, divv2sf3): Delete expanders.
10320 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10322 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
10324 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10326 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
10327 * config/rs6000/rs6000.c: Ditto.
10329 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10331 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
10332 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
10333 comparison_operator.
10335 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10337 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
10338 * config/rs6000/rs6000.opt: Ditto.
10339 * config/rs6000/t-rtems: Ditto.
10341 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10343 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
10344 TARGET_E500_SINGLE by 0, simplify.
10345 * config/rs6000/rs6000.c: Ditto.
10346 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
10347 (spe_build_register_parallel): Delete.
10348 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
10349 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
10350 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
10351 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
10352 (E500_CONVERT): Delete.
10353 * config/rs6000/spe.md: Remove many patterns and all define_constants.
10355 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10357 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
10358 * config/rs6000/dfp.md: Ditto.
10359 (negdd2, *negdd2_fpr): Merge.
10360 (absdd2, *absdd2_fpr): Merge.
10361 (negtd2, *negtd2_fpr): Merge.
10362 (abstd2, *abstd2_fpr): Merge.
10363 * config/rs6000/e500.h: Delete file.
10364 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
10365 TARGET_FPRS by 1 and simplify.
10366 * config/rs6000/rs6000-c.c: Ditto.
10367 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
10368 TARGET_DF_SPE by 0.
10369 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
10371 * config/rs6000/rs6000.md: Ditto.
10372 (floatdidf2, *floatdidf2_fpr): Merge.
10373 (move_from_CR_gt_bit): Delete.
10374 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
10375 (E500_CR_IOR_COMPARE): Delete.
10376 (All patterns that require !TARGET_FPRS): Delete.
10377 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
10379 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10381 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
10383 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10385 * graphds.c (add_edge): Intitialize edge's attached data.
10386 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
10387 pointer parameter. Call pointed function on each edge during
10388 graph traversing. Skip traversing the edge when the function
10390 (graphds_dfs, graphds_scc): Ditto.
10391 (for_each_edge): New parameter. Pass the new parameter to callback
10393 * graphds.h (skip_edge_callback): New function pointer type.
10394 (graphds_dfs, graphds_scc): New function pointer parameter.
10395 (graphds_edge_callback, for_each_edge): New parameter.
10397 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10399 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
10400 out code checking if runtime alias check is possible to below ...
10401 Call the new function.
10402 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
10403 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
10405 2017-06-07 Marek Polacek <polacek@redhat.com>
10408 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
10409 TYPE_OVERFLOW_WRAPS check.
10411 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10413 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
10414 if versioning is required.
10415 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
10416 peeling with the check for versioning.
10418 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10420 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
10421 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
10422 Set true to new parameter if new ssa variable is defined.
10423 (vect_gen_vector_loop_niters): Refactor. Set range information
10424 for the new vector loop bound variable.
10425 (vect_do_peeling): Ditto.
10427 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10429 * tree-affine.c (ssa.h): Include header file.
10430 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
10431 has wrapping overflow behavior.
10433 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10435 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
10437 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10439 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
10440 (tree_to_aff_combination): ... here.
10442 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10444 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
10445 reg_pressure model function.
10446 (ivopts_global_cost_for_size): Delete.
10447 (determine_set_costs, iv_ca_recount_cost): Call new model function
10448 ivopts_estimate_reg_pressure.
10450 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10452 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
10453 expensive than udiv. Remove floating point cases from mod.
10455 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10457 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
10458 Increase idiv cost.
10460 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10462 * config/aarch64/aarch64.md
10463 (copysignsf3): Fix mask generation.
10465 2017-06-07 Jakub Jelinek <jakub@redhat.com>
10467 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
10469 (class dump_manager): Add register_dumps method.
10470 * dumpfile.c: Include langhooks.h.
10471 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
10472 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
10473 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
10474 (dump_manager::dump_register): Start with 512 entries instead of 32.
10475 (dump_manager::register_dumps): New method.
10476 * toplev.c (general_init): Instead of invoking register_dumps
10477 langhook, invoke register_dumps method on the dump manager.
10478 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
10481 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
10483 * doc/md.texi: Clarify the restrictions on a define_insn condition.
10484 Say that # requires an associated define_split to exist, and that
10485 the define_split must be suitable for use after register allocation.
10487 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10489 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
10490 (compute_outgoing_frequencies): Also initialize zero counts.
10491 (find_many_sub_basic_blocks): Do not produce uninitialized profile
10492 around loops; preserve more of profile when nothing changes.
10494 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
10496 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
10498 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
10499 * config/arm/arm-cpu-cdata.h: Regenerate.
10500 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
10501 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
10502 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
10503 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
10504 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
10506 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
10508 * config/arm/t-rmprofile: Likewise.
10509 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
10511 2017-06-06 David S. Miller <davem@davemloft.net>
10514 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
10515 blockage if function uses alloca.
10517 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10519 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
10520 New "uid" fields to hold pretty-print IDs of group and ref.
10521 Memory references are now identified as <group_id>:<ref_id>
10522 instead of using [random] addresses.
10523 (dump_mem_details): Simplify, no functional change.
10524 (dump_mem_ref): Simplify and make output more concise.
10525 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
10526 (find_or_create_group): Initialize group uid.
10527 (record_ref): Initialize ref uid. Improve debug output.
10528 (prune_group_by_reuse, should_issue_prefetch_p,)
10529 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
10530 (mark_nontemporal_store, determine_loop_nest_reuse):
10531 Improve debug output.
10533 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10535 * dbgcnt.def (prefetch): New debug counter.
10536 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
10537 (schedule_prefetches): Stop issueing prefetches if debug counter
10540 2017-06-06 Tom de Vries <tom@codesourcery.com>
10542 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
10543 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
10545 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10547 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
10548 Use aarch64_reg_or_zero predicate for operand 4.
10549 (aarch64_compare_and_swap<mode> define_insn_and_split):
10550 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
10551 (aarch64_store_exclusive<mode>): Likewise for operand 2.
10553 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10555 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
10556 (arm_compute_save_core_reg_mask): This.
10557 (thumb1_compute_save_reg_mask): Rename into ...
10558 (thumb1_compute_save_core_reg_mask): This.
10559 (arm_compute_save_reg0_reg12_mask): Adapt comment.
10560 (arm_compute_frame_layout): Likewise.
10562 2017-06-06 Richard Biener <rguenther@suse.de>
10564 PR tree-optimization/80974
10565 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
10566 keep or clear leaders SSA info.
10568 2017-06-06 Tom de Vries <tom@codesourcery.com>
10570 * config/nvptx/nvptx.c (split_mode_p): New function.
10571 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
10573 2017-06-06 Tom de Vries <tom@codesourcery.com>
10575 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
10577 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10580 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
10583 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10585 * shrink-wrap.c (handle_simple_exit): Update profile.
10586 (try_shrink_wrapping): Upate profile.
10588 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10590 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
10591 (tree_guess_outgoing_edge_probabilities): New.
10592 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
10593 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
10595 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10597 * ipa-split.c (split_function): Initialize return bb profile.
10599 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10601 * profile.c (compute_branch_probabilities): Also initialize
10602 EXIT_BLOCK profile.
10604 2017-06-06 Richard Biener <rguenther@suse.de>
10606 PR tree-optimization/80928
10607 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
10608 (vect_analyze_loop_operations): Properly guard analysis for
10610 (vect_transform_loop): Likewise.
10611 (vect_analyze_loop_2): Also reset SLP type on PHIs.
10612 (vect_model_induction_cost): Do not cost for pure SLP.
10613 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
10614 of induction in inner loop vectorization.
10615 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
10616 (vect_get_and_check_slp_defs): Handle vect_induction_def.
10617 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
10619 (vect_analyze_slp_cost_1): Cost induction.
10620 (vect_detect_hybrid_slp_stmts): Handle PHIs.
10621 (vect_get_slp_vect_defs): Likewise.
10622 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
10623 (vect_transform_stmt): Handle SLP reductions.
10624 * tree-vectorizer.h (vectorizable_induction): Adjust.
10626 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
10628 * config/rs6000/rs6000.c (make_resolver_func): Update
10629 init_lowered_empty_function call.
10631 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
10633 * doc/invoke.texi: Document the -fprofile-abs-path option.
10634 * common.opt (fprofile-abs-path): New option.
10635 * gcov-io.h (gcov_write_filename): Declare.
10636 * gcov-io.c (gcov_write_filename): New function.
10637 * coverage.c (coverage_begin_function): Use gcov_write_filename.
10638 * profile.c (output_location): Likewise.
10640 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10642 * shring-wrap.c: Revert accidental commit.
10644 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
10646 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
10648 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10650 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
10652 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
10653 profile in callgraph edge.
10654 * profile-count.h (apply_probability): If THIS is 0, then result is 0
10655 (apply_scale): Likewise.
10656 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
10657 Also scale profile when inlining function with zero profile.
10658 (initialize_cfun): Update exit block profile even when it is zero.
10659 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
10660 when profile is read.
10662 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
10664 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
10665 (CLONE_*): New constants to define the processors we can generate
10666 code for with the target_clone attribute.
10667 (rs6000_clone_map): New array to identify which clone processors
10668 the current program is running on.
10669 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
10670 target_clone attribute.
10671 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
10672 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
10673 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
10674 (cpu_expand_builtin): Add support for target_clone attribute.
10675 (rs6000_valid_attribute_p): Allow "default" attribute.
10676 (get_decl_name): New debug function to simplify printing the
10677 current function name in debugging statements.
10678 (rs6000_clone_priority): New functions to support the target_clone
10679 attribute, and be able to generate code to switch between ISA 2.05
10680 through ISA 3.0 (power6 through power9).
10681 (rs6000_compare_version_priority): Likewise.
10682 (rs6000_get_function_versions_dispatcher): Likewise.
10683 (make_resolver_func): Likewise.
10684 (add_condition_to_bb): Likewise.
10685 (dispatch_function_versions): Likewise.
10686 (rs6000_generate_version_dispatcher_body): Likewise.
10687 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
10688 (fusion_gpr_load_p): Fix a spacing issue.
10689 * doc/extend.texi (Common Function Attributes): Document that the
10690 PowerPC supports the target_clone attribute.
10692 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
10694 * config/arm/arm.h: explain F symbol found in description of ARM
10695 register allocation in its legend.
10697 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10699 * config/mips/frame-header-opt.c: Include profile-count.h.
10700 * config/riscv/riscv.c: Include profile-count.h
10702 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
10704 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
10706 (sm_set_flag_if_changed): Add bbs field.
10707 (execute_sm_if_changed_flag_set): Pass BBS.
10708 (execute_sm): Update.
10710 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10712 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
10715 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10717 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
10718 (peephole2): New peephole2 to emit the above.
10719 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
10721 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10723 * config/aarch64/aarch64.c (define_peephole2 above
10724 *sub_<shift>_<mode>): New peephole.
10726 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
10728 * config/i386/i386.c (make_resolver_func): Update.
10729 * Makefile.in: Add profile-count.h and profile-count.o
10730 * auto-profile.c (afdo_indirect_call): Update to new API.
10731 (afdo_set_bb_count): Update.
10732 (afdo_propagate_edge): Update.
10733 (afdo_propagate_circuit): Update.
10734 (afdo_calculate_branch_prob): Update.
10735 (afdo_annotate_cfg): Update.
10736 * basic-block.h: Include profile-count.h
10737 (struct edge_def): Turn count to profile_count.
10738 (struct basic_block_def): Likewie.
10739 (REG_BR_PROB_BASE): Move to profile-count.h
10740 (RDIV): Move to profile-count.h
10741 * bb-reorder.c (max_entry_count): Turn to profile_count.
10742 (find_traces): Update.
10743 (rotate_loop):Update.
10744 (connect_traces):Update.
10745 (sanitize_hot_paths):Update.
10746 * bt-load.c (migrate_btr_defs): Update.
10747 * cfg.c (RDIV): Remove.
10748 (init_flow): Use alloc_block.
10749 (alloc_block): Uninitialize count.
10750 (unchecked_make_edge): Uninitialize count.
10751 (check_bb_profile): Update.
10752 (dump_edge_info): Update.
10753 (dump_bb_info): Update.
10754 (update_bb_profile_for_threading): Update.
10755 (scale_bbs_frequencies_int): Update.
10756 (scale_bbs_frequencies_gcov_type): Update.
10757 (scale_bbs_frequencies_profile_count): New.
10758 * cfg.h (update_bb_profile_for_threading): Update.
10759 (scale_bbs_frequencies_profile_count): Declare.
10760 * cfgbuild.c (compute_outgoing_frequencies): Update.
10761 (find_many_sub_basic_blocks): Update.
10762 * cfgcleanup.c (try_forward_edges): Update.
10763 (try_crossjump_to_edge): Update.
10764 * cfgexpand.c (expand_gimple_tailcall): Update.
10765 (construct_exit_block): Update.
10766 * cfghooks.c (verify_flow_info): Update.
10767 (dump_bb_for_graph): Update.
10768 (split_edge): Update.
10769 (make_forwarder_block): Update.
10770 (duplicate_block): Update.
10771 (account_profile_record): Update.
10772 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
10773 (get_estimated_loop_iterations): Update.
10774 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
10775 (single_likely_exit): Update.
10776 * cfgloopmanip.c (scale_loop_profile): Update.
10778 (set_zero_probability): Update.
10779 (lv_adjust_loop_entry_edge): Update.
10780 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
10781 (purge_dead_edges): Update.
10782 (rtl_account_profile_record): Update.
10783 * cgraph.c (cgraph_node::create): Uninitialize count.
10784 (symbol_table::create_edge): Uninitialize count.
10785 (cgraph_update_edges_for_call_stmt_node): Update.
10786 (cgraph_edge::dump_edge_flags): Update.
10787 (cgraph_node::dump): Update.
10788 (cgraph_edge::maybe_hot_p): Update.
10789 * cgraph.h: Include profile-count.h
10790 (create_clone), create_edge, create_indirect_edge): Update.
10791 (cgraph_node): Turn count to profile_count.
10792 (cgraph_edge0: Likewise.
10793 (make_speculative, clone): Update.
10794 (create_edge): Update.
10795 (init_lowered_empty_function): Update.
10796 * cgraphclones.c (cgraph_edge::clone): Update.
10797 (duplicate_thunk_for_node): Update.
10798 (cgraph_node::create_clone): Update.
10799 * cgraphunit.c (cgraph_node::analyze): Update.
10800 (cgraph_node::expand_thunk): Update.
10801 * final.c (dump_basic_block_info): Update.
10802 * gimple-streamer-in.c (input_bb): Update.
10803 * gimple-streamer-out.c (output_bb): Update.
10804 * graphite.c (print_global_statistics): Update.
10805 (print_graphite_scop_statistics): Update.
10806 * hsa-brig.c: Include basic-block.h.
10807 * hsa-dump.c: Include basic-block.h.
10808 * hsa-gen.c (T sum_slice): Update.
10809 (convert_switch_statements):Update.
10810 * hsa-regalloc.c: Include basic-block.h.
10811 * ipa-chkp.c (chkp_produce_thunks): Update.
10812 * ipa-cp.c (struct caller_statistics): Update.
10813 (init_caller_stats): Update.
10814 (gather_caller_stats): Update.
10815 (ipcp_cloning_candidate_p): Update.
10816 (good_cloning_opportunity_p): Update.
10817 (get_info_about_necessary_edges): Update.
10818 (dump_profile_updates): Update.
10819 (update_profiling_info): Update.
10820 (update_specialized_profile): Update.
10821 (perhaps_add_new_callers): Update.
10822 (decide_about_value): Update.
10823 (ipa_cp_c_finalize): Update.
10824 * ipa-devirt.c (struct odr_type_warn_count): Update.
10825 (struct decl_warn_count): Update.
10826 (struct final_warning_record): Update.
10827 (possible_polymorphic_call_targets): Update.
10828 (ipa_devirt): Update.
10829 * ipa-fnsummary.c (redirect_to_unreachable): Update.
10830 * ipa-icf.c (sem_function::merge): Update.
10831 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
10832 * ipa-inline.c (compute_uninlined_call_time): Update.
10833 (compute_inlined_call_time): Update.
10834 (want_inline_small_function_p): Update.
10835 (want_inline_self_recursive_call_p): Update.
10836 (edge_badness): Update.
10837 (lookup_recursive_calls): Update.
10838 (recursive_inlining): Update.
10839 (inline_small_functions): Update.
10840 (dump_overall_stats): Update.
10841 (dump_inline_stats): Update.
10842 * ipa-profile.c (ipa_profile_generate_summary): Update.
10843 (ipa_propagate_frequency): Update.
10844 (ipa_profile): Update.
10845 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
10846 * ipa-utils.c (ipa_merge_profiles): Update.
10847 * loop-doloop.c (doloop_modify): Update.
10848 * loop-unroll.c (report_unroll): Update.
10849 (unroll_loop_runtime_iterations): Update.
10850 * lto-cgraph.c (lto_output_edge): Update.
10851 (lto_output_node): Update.
10852 (input_node): Update.
10853 (input_edge): Update.
10854 (merge_profile_summaries): Update.
10855 * lto-streamer-in.c (input_cfg): Update.
10856 * lto-streamer-out.c (output_cfg): Update.
10857 * mcf.c (create_fixup_graph): Update.
10858 (adjust_cfg_counts): Update.
10859 (sum_edge_counts): Update.
10860 * modulo-sched.c (sms_schedule): Update.
10861 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
10862 * predict.c (maybe_hot_count_p): Update.
10863 (probably_never_executed): Update.
10864 (dump_prediction): Update.
10865 (combine_predictions_for_bb): Update.
10866 (propagate_freq): Update.
10867 (handle_missing_profiles): Update.
10868 (counts_to_freqs): Update.
10869 (rebuild_frequencies): Update.
10870 (force_edge_cold): Update.
10871 * predict.h: Include profile-count.h
10872 (maybe_hot_count_p, counts_to_freqs): UPdate.
10873 * print-rtl-function.c: Do not include cfg.h
10874 * print-rtl.c: Include basic-block.h
10875 * profile-count.c: New file.
10876 * profile-count.h: New file.
10877 * profile.c (is_edge_inconsistent): Update.
10878 (correct_negative_edge_counts): Update.
10879 (is_inconsistent): Update.
10880 (set_bb_counts): Update.
10881 (read_profile_edge_counts): Update.
10882 (compute_frequency_overlap): Update.
10883 (compute_branch_probabilities): Update; Initialize and deinitialize
10885 (branch_prob): Update.
10886 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
10887 (edge_gcov_count): New.
10888 (bb_gcov_count): New.
10889 * shrink-wrap.c (try_shrink_wrapping): Update.
10890 * tracer.c (better_p): Update.
10891 * trans-mem.c (expand_transaction): Update.
10892 (ipa_tm_insert_irr_call): Update.
10893 (ipa_tm_insert_gettmclone_call): Update.
10894 * tree-call-cdce.c: Update.
10895 * tree-cfg.c (gimple_duplicate_sese_region): Update.
10896 (gimple_duplicate_sese_tail): Update.
10897 (gimple_account_profile_record): Update.
10898 (execute_fixup_cfg): Update.
10899 * tree-inline.c (copy_bb): Update.
10900 (copy_edges_for_bb): Update.
10901 (initialize_cfun): Update.
10902 (freqs_to_counts): Update.
10903 (copy_cfg_body): Update.
10904 (expand_call_inline): Update.
10905 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
10906 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
10907 (try_unroll_loop_completely): Update.
10908 (try_peel_loop): Update.
10909 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
10910 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
10911 * tree-ssa-loop-split.c (connect_loops): Update.
10912 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
10913 * tree-ssa-reassoc.c (branch_fixup): Update.
10914 * tree-ssa-tail-merge.c (replace_block_by): Update.
10915 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
10916 (compute_path_counts): Update.
10917 (update_profile): Update.
10918 (recompute_probabilities): Update.
10919 (update_joiner_offpath_counts): Update.
10920 (estimated_freqs_path): Update.
10921 (freqs_to_counts_path): Update.
10922 (clear_counts_path): Update.
10923 (ssa_fix_duplicate_block_edges): Update.
10924 (duplicate_thread_path): Update.
10925 * tree-switch-conversion.c (case_bit_test_cmp): Update.
10926 (struct switch_conv_info): Update.
10927 * tree-tailcall.c (decrease_profile): Update.
10928 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
10929 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
10930 * value-prof.c (check_counter): Update.
10931 (gimple_divmod_fixed_value): Update.
10932 (gimple_mod_pow2): Update.
10933 (gimple_mod_subtract): Update.
10934 (gimple_ic_transform): Update.
10935 (gimple_stringop_fixed_value): Update.
10936 * value-prof.h (gimple_ic): Update.
10938 2017-06-02 Carl Love <cel@us.ibm.com>
10940 * config/rs6000/rs6000-c: Add support for built-in functions
10941 vector double vec_doublee (vector signed int);
10942 vector double vec_doublee (vector unsigned int);
10943 vector double vec_doublee (vector float);
10944 vector double vec_doubleh (vector signed int);
10945 vector double vec_doubleh (vector unsigned int);
10946 vector double vec_doubleh (vector float);
10947 vector double vec_doublel (vector signed int);
10948 vector double vec_doublel (vector unsigned int);
10949 vector double vec_doublel (vector float);
10950 vector double vec_doubleo (vector signed int);
10951 vector double vec_doubleo (vector unsigned int);
10952 vector double vec_doubleo (vector float);.
10953 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
10954 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
10956 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
10957 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
10958 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
10960 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
10961 vec_doublel, vec_doubleh.
10962 * doc/extend.texi: Update the built-in documentation file for the
10963 new built-in functions.
10965 2017-06-02 David Malcolm <dmalcolm@redhat.com>
10968 * ipa-inline-analysis.c (free_growth_caches): Set
10969 edge_removal_hook_holder to NULL after removing it.
10971 2017-06-02 Sudakshina Das <sudi.das@arm.com>
10973 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
10974 comparision with zero.
10976 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
10977 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10978 for early expansion of vec_min and vec_max builtins.
10979 (builtin_function_type): Add min/max unsigned variants to those
10980 identified as having unsigned arguments.
10982 2017-06-02 Olivier Hainque <hainque@adacore.com>
10984 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
10986 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10988 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
10989 Use VALL_F16 iterator rather than VALL.
10991 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10993 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
10994 Emit CBNZ inside loop when doing a strong exchange and comparing
10995 against zero. Generate the CC flags after the loop.
10997 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
10999 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
11000 (dl_section_ref): New.
11001 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
11002 On AIX, append an expression to subtract the size of the
11003 section length to dl_section_ref.
11005 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
11007 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11008 for early expansion of vector absolute builtins.
11010 2017-06-02 Richard Biener <rguenther@suse.de>
11012 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
11013 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
11015 2017-06-02 Richard Biener <rguenther@suse.de>
11017 PR tree-optimization/80948
11018 * tree-tailcall.c (find_tail_calls): Track stmts to move in
11019 stmt order as well.
11021 2017-06-02 Richard Biener <rguenther@suse.de>
11023 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
11025 * tree-vect-stmts.c (process_use): Do not mark backedge defs
11026 for inductions as relevant.
11028 2017-06-02 Richard Biener <rguenther@suse.de>
11030 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
11031 (vectorizable_induction): ... this. Remove dead code.
11033 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
11035 * builtins. (expand_builtin_alloca): Remove second parameter and
11036 infer its value from the first parameter instead.
11037 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
11039 2017-06-02 Jakub Jelinek <jakub@redhat.com>
11041 PR rtl-optimization/80903
11042 * loop-doloop.c (add_test): Unshare sequence.
11044 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
11046 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
11048 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
11050 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
11052 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
11053 xlogue_layout::get_instance, logue_layout::xlogue_layout,
11054 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
11055 (xlogue_layout::get_stub_rtx): Make static.
11056 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
11057 (xlogue_layout::compute_stub_managed_regs): Rename to...
11058 (xlogue_layout::count_stub_managed_regs): ...this.
11059 (xlogue_layout::is_stub_managed_reg): New function.
11060 (xlogue_layout::m_stub_names): Rename to...
11061 (xlogue_layout::s_stub_names): ...this, make static.
11062 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
11063 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
11064 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
11065 xlogue_layout::s_stub_names): Instantiate statics.
11066 (stub_managed_regs): Remove.
11067 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
11068 (disable_call_ms2sysv_xlogues): Rename to...
11069 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
11070 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
11072 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
11073 change after reload_completed.
11074 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
11076 (ix86_expand_prologue): Likewise.
11077 (ix86_expand_epilogue): Likewise.
11078 (ix86_expand_split_stack_prologue): Likewise.
11079 (ix86_compute_frame_layout): Remove frame parameter ...
11080 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
11081 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
11083 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
11084 (ix86_frame): Move from here ...
11085 * config/i386/i386.h (ix86_frame): ... to here.
11086 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
11087 complete ix86_frame data structure instead. Remove some_ld_name.
11089 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
11091 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
11092 symbols that hold a DECL_VALUE_EXPR.
11094 2017-06-01 Martin Jambor <mjambor@suse.cz>
11096 PR tree-optimization/80898
11097 * tree-sra.c (process_subtree_disqualification): Removed.
11098 (disqualify_candidate): Do not acll
11099 process_subtree_disqualification.
11100 (subtree_mark_written_and_enqueue): New function.
11101 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
11102 RHS has been disqualified and re-queue LHS if necessary. Apart
11103 from that, ignore disqualified RHS.
11105 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11107 * config/s390/s390.c (s390_emit_epilogue): Disable early return
11108 address fetch for z10 or later.
11110 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11112 * config/arc/arc.md (tst_movb): Add guard when splitting.
11114 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11116 * config/arc/arc.c (arc_can_eliminate): Test against
11117 arc_frame_pointer_needed.
11119 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11121 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
11122 to prevent store reordering.
11123 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
11124 (type): Add block type.
11125 (stack_tie): Define special instruction to be used in
11128 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11130 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
11131 constraint. It is not valid for the pattern.
11132 (noncommutative_binary_comparison): Likewise.
11134 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11136 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
11139 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11141 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
11142 be used by the reg-alloc.
11144 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11146 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
11147 reg-alloc when having mul64 or mul32x16 instructions.
11148 (mulsidi3): Likewise.
11149 (umulsidi3): Likewise.
11150 (mulsi32x16): New pattern.
11151 (mulsi64): Likewise.
11152 (mulsidi64): Likewise.
11153 (umulsidi64): Likewise.
11154 (MUL32x16_REG): Define.
11155 (mul64_600): Use MUL32x16_REG.
11156 (mac64_600): Likewise.
11157 (umul64_600): Likewise.
11158 (umac64_600): Likewise.
11160 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11162 * config/arc/arc.md (mulsi3_700): Make it commutative.
11164 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
11166 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
11168 (*sign_extendsidi2_insn): Likewise for movstosw.
11170 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
11172 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
11173 the type of the input discriminant value. Convert the
11174 discriminant value of signedness vary.
11176 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
11178 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
11179 Add to -Wall section.
11181 2017-06-01 Richard Biener <rguenther@suse.de>
11183 PR middle-end/66313
11184 * fold-const.c (fold_plusminus_mult_expr): If the factored
11185 factor may be zero use a wrapping type for the inner operation.
11186 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
11187 and handle moved defs.
11188 (process_assignment): Properly guard the unary op case. Return a
11189 tri-state indicating that moving the stmt before the call may allow
11190 to continue. Pass through to_move.
11191 (find_tail_calls): Handle moving unrelated defs before
11194 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
11197 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
11198 splitter result in the canonical way.
11200 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
11202 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
11203 also for 32bit target. Update insn attributes.
11204 (zero-extendsidi2 splitter): Allow all registers for operand 1.
11206 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
11208 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
11209 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
11210 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
11211 (_mm_maskz_min_ss): New intrinsics.
11213 2017-05-31 Martin Liska <mliska@suse.cz>
11215 * tree-vect-loop.c (vect_create_epilog_for_reduction):
11216 Change comment style to one we normally use.
11217 (vectorizable_reduction): Likewise.
11218 (vectorizable_induction): Likewise.
11219 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11220 (vectorizable_call): Likewise.
11221 (vectorizable_simd_clone_call): Likewise.
11222 (vectorizable_conversion): Likewise.
11223 (vectorizable_assignment): Likewise.
11224 (vectorizable_shift): Likewise.
11225 (vectorizable_operation): Likewise.
11226 (vectorizable_store): Likewise.
11227 (vectorizable_load): Likewise.
11228 * tree-vectorizer.h: Likewise.
11230 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
11232 * passes.c (emergency_dump_function): New.
11233 * tree-pass.h (emergency_dump_function): Declare.
11234 * plugin.c (plugins_internal_error_function): Remove.
11235 * plugin.h (plugins_internal_error_function): Remove declaration.
11236 * toplev.c (internal_error_function): New static function. Use it...
11237 (general_init): ...here.
11239 2017-05-31 Graham Markall <graham.markall@embecosm.com>
11241 * config/arc/arc.c (arc_print_operand): Handle constant operands.
11242 (arc_rtx_costs): Add costs for new patterns.
11243 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
11244 * config/arc/predicates.md: Add _1_2_3_operand predicate.
11246 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11248 * tree-ssa-strlen.c (get_next_strinfo): New function.
11249 (get_stridx_plus_constant): Use it.
11250 (zero_length_string): Likewise.
11251 (adjust_related_strinfos): Likewise.
11252 (adjust_last_stmt): Likewise.
11254 2017-05-31 Richard Biener <rguenther@suse.de>
11257 * config/i386/i386.c (ix86_expand_builtin): Remove assert
11258 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
11260 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11262 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
11263 loop_vinfo argument and use of dependence distance vectors.
11264 Check instead whether the two references differ only in their
11265 initial value and assume that they have the same alignment if the
11266 difference is a multiple of the vector alignment.
11267 (vect_analyze_data_refs_alignment): Update call accordingly.
11269 2017-05-31 Martin Liska <mliska@suse.cz>
11272 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
11274 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11276 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
11277 (create_intersect_range_checks): Move from ...
11278 * tree-data-ref.c (create_intersect_range_checks_index)
11279 (create_intersect_range_checks): ... to here.
11280 (create_runtime_alias_checks): New function factored from ...
11281 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
11282 here. Call above function.
11283 * tree-data-ref.h (create_runtime_alias_checks): New function.
11285 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11287 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
11288 segment length for dr_b and compute it in wide_int.
11290 2017-05-31 Richard Biener <rguenther@suse.de>
11292 PR tree-optimization/80906
11293 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
11294 and pass through iv_map.
11295 (copy_bb_and_scalar_dependences): Adjust.
11296 (translate_pending_phi_nodes): Likewise.
11297 (copy_loop_close_phi_args): Handle code-generating IVs instead
11300 2017-05-30 David Malcolm <dmalcolm@redhat.com>
11302 * diagnostic-color.c (color_dict): Add "type-diff".
11303 (parse_gcc_colors): Update comment.
11304 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
11305 -fdiagnostics-show-template-tree and -fno-elide-type.
11306 (GCC_COLORS): Add type-diff to example.
11308 (-fdiagnostics-show-template-tree): New.
11309 (-fno-elide-type): New.
11310 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
11311 the pp_format_decoder callback. Call any m_format_postprocessor's
11313 (pretty_printer::pretty_printer): Initialize
11314 m_format_postprocessor.
11315 (pretty_printer::~pretty_printer): Delete any
11316 m_format_postprocessor.
11317 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
11318 (class format_postprocessor): New class.
11319 (struct pretty_printer::format_decoder): Document the new parameters.
11320 (struct pretty_printer::m_format_postprocessor): New field.
11321 * tree-diagnostic.c (default_tree_printer): Update for new
11322 bool and const char ** params.
11323 * tree-diagnostic.h (default_tree_printer): Likewise.
11325 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
11327 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
11328 (lwa_operand): Delete rs6000_gen_cell_microcode test.
11329 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
11330 rs6000_gen_cell_microcode code.
11331 (rs6000_final_prescan_insn): Delete.
11332 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
11333 "warn-cell-microcode" entries.
11334 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
11335 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
11336 throughout. Change cc_reg_not_micro_cr0_operand to
11337 cc_reg_not_cr0_operand throughout.
11338 (*extendhi<mode>2_noload): Delete.
11339 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
11340 (mwarn-cell-microcode): Delete.
11341 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
11342 -mgen-cell-microcode and -mwarn-cell-microcode.
11344 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
11347 * config/i386/constraints.md (Yd): New constraint.
11349 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
11350 and (?Yd, r) alternatives. Update insn attributes.
11351 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
11352 and (?*Yd, r) alternatives. Update insn attributes.
11353 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
11355 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
11357 * gimplify.c (gimplify_modify_expr): Don't create a
11358 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
11361 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
11363 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
11365 2017-05-30 Richard Biener <rguenther@suse.de>
11367 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
11368 and reduc_def fields.
11369 (STMT_VINFO_REDUC_TYPE): New define.
11370 (STMT_VINFO_REDUC_DEF): Likewise.
11371 (vect_force_simple_reduction): Adjust prototype.
11372 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
11373 (vect_is_simple_reduction): Remove check_reduction argument.
11374 (vect_force_simple_reduction): Adjust and set
11375 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11376 (vectorizable_reduction): Do not re-do reduction analysis
11377 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11378 * tree-parloops.c (gather_scalar_reductions): Adjust.
11380 2017-05-30 Richard Biener <rguenther@suse.de>
11382 PR middle-end/80901
11383 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
11386 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
11388 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
11389 Introduce unknown_misalignment parameter and remove vf.
11390 (vect_peeling_hash_get_lowest_cost):
11391 Pass unknown_misalignment parameter.
11392 (vect_enhance_data_refs_alignment):
11393 Fix unsupportable data ref treatment.
11395 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11397 * tree-vect-data-refs.c (vect_get_data_access_cost):
11398 Workaround for SLP handling.
11399 (vect_enhance_data_refs_alignment):
11400 Compute costs for doing no peeling at all, compare to the best
11401 peeling costs so far and avoid peeling if cheaper.
11403 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11405 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
11406 Return peeling info and set costs to zero for unlimited cost
11408 (vect_enhance_data_refs_alignment): Also inspect all datarefs
11409 with unknown misalignment. Compute and costs for unknown
11410 misalignment, compare them to the costs for known misalignment
11411 and choose the cheapest for peeling.
11413 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11415 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
11416 (vect_get_peeling_costs_all_drs): Create function.
11417 (vect_peeling_hash_get_lowest_cost):
11418 Use vect_get_peeling_costs_all_drs.
11419 (vect_peeling_supportable): Create function.
11420 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
11422 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11424 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
11425 DR_HAS_NEGATIVE_STEP.
11426 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
11427 (vect_enhance_data_refs_alignment): Use.
11428 (vect_duplicate_ssa_name_ptr_info): Use.
11429 * tree-vectorizer.h (dr_misalignment): Use.
11430 (known_alignment_for_access_p): Use.
11432 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11435 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
11437 (has_section_name): New function.
11439 2017-05-30 Martin Liska <mliska@suse.cz>
11442 * auto-profile.c (get_function_decl_from_block): Fix
11445 2017-05-30 Richard Biener <rguenther@suse.de>
11447 PR middle-end/80876
11448 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
11450 2017-05-30 Martin Liska <mliska@suse.cz>
11452 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
11453 * dumpfile.h (struct dump_file_info): Remove ctors.
11455 2017-05-30 Martin Liska <mliska@suse.cz>
11457 * predict.def: Fix GNU coding style.
11459 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11461 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
11462 Mark 'to' argument with ATTRIBUTE_UNUSED.
11464 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11466 * config/xtensa/xtensa.c (xtensa_emit_call): Use
11467 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
11468 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
11471 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
11473 * doc/install.texi (Options specification): Restore entry of
11474 --enable-sjlj-exceptions.
11476 2017-05-27 Michael Eager <eager@eagercon.com>
11479 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
11481 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
11483 * config/microblaze/microblaze.h
11484 (FIXED_REGISTERS): Update in macro.
11485 (CALL_USED_REGISTERS): Update in macro.
11487 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
11489 * doc/install.texi: Add links to macOS binary distributions.
11491 2017-05-27 Jakub Jelinek <jakub@redhat.com>
11495 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11497 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
11499 2017-05-26 Martin Liska <mliska@suse.cz>
11501 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
11503 2017-05-26 Martin Liska <mliska@suse.cz>
11505 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
11506 always leading ';; '.
11507 (dump_bb_info): Likewise.
11508 (brief_dump_cfg): Likewise.
11509 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
11510 * dumpfile.c: Remove usage of TDF_VERBOSE.
11511 * dumpfile.h (enum dump_kind): Likewise.
11512 (dump_gimple_bb_header): Do not use TDF_COMMENT.
11513 * print-tree.c (debug_verbose): Remove.
11514 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
11515 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
11516 * tree-diagnostic.c (default_tree_printer): Replace
11517 TDF_DIAGNOSTIC with TDF_SLIM.
11519 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11521 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
11522 in parameter loop, rather than loop_vinfo.
11523 (create_intersect_range_checks): Ditto.
11524 (vect_create_cond_for_alias_checks): Update call to above functions.
11526 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11528 PR tree-optimization/80815
11529 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
11530 for merging runtime alias checks. Handle negative DR_STEPs.
11532 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11534 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
11536 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
11537 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
11538 out code pruning runtime alias checks.
11539 * tree-data-ref.c (prune_runtime_alias_test_list): New function
11540 factored out from above.
11541 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
11543 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
11545 (prune_runtime_alias_test_list): New decalaration.
11547 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11549 * tree-vect-data-refs.c (compare_tree): Rename and move ...
11550 * tree-data-ref.c (data_ref_compare_tree): ... to here.
11551 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
11552 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
11553 (operator==, comp_dr_with_seg_len_pair): Ditto.
11554 (vect_prune_runtime_alias_test_list): Ditto.
11556 2017-05-26 Martin Liska <mliska@suse.cz>
11559 * params.def: Bound partial-inlining-entry-probability param.
11561 2017-05-26 Marek Polacek <polacek@redhat.com>
11564 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
11567 2017-05-26 Richard Biener <rguenther@suse.de>
11569 PR tree-optimization/80842
11570 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
11573 2017-05-26 Richard Biener <rguenther@suse.de>
11575 PR tree-optimization/80844
11576 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
11578 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
11580 * doc/md.texi (Machine Constraints): Update x86 family
11581 machine constraints section to match 'config/i386/constraints.md'.
11583 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
11585 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
11587 2017-05-25 Nathan Sidwell <nathan@acm.org>
11589 * doc/invoke.texi (--enable-languages): Update documentation.
11591 2017-05-25 Martin Liska <mliska@suse.cz>
11593 * dumpfile.c: Add TDF_FOLDING.
11594 * dumpfile.h (enum dump_kind): Likewise.
11595 * genmatch.c (dt_simplify::gen_1): Use it.
11597 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11599 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
11601 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11603 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
11604 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
11606 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11608 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
11609 * match.pd (X == C): Rewrite it here.
11610 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
11611 with_certain_nonzero_bits2): New predicates.
11612 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
11614 2017-05-24 Nathan Sidwell <nathan@acm.org>
11616 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
11619 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
11622 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
11624 * config/powerpcspe: New port. Files are copied from the rs6000
11625 port, with "rs6000" in filenames replaced by "powerpcspe".
11627 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
11629 PR rtl-optimization/80754
11630 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
11632 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
11634 * config/sparc/sparc.md (length): Return the correct value for -mflat
11635 sibcalls to match output_sibcall.
11637 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
11641 * config/rs6000/rs6000.c (struct machine_function): Add new field
11643 (rs6000_get_separate_components): Init that field, use it.
11644 (rs6000_components_for_bb): Use the field.
11646 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
11648 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
11650 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
11652 PR middle-end/80823
11653 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
11655 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11658 * config/s390/s390.c (s390_check_qrst_address): Check incoming
11659 address against address_operand predicate.
11660 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
11662 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
11664 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
11665 parameters passed indirectly.
11667 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
11669 * config/i386/i386.md (*movdi_internal): Remove SSE4
11670 alternative 18 (?r, *v). Update insn attributes.
11671 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
11672 Update insn attributes.
11673 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
11674 Update insn attributes.
11675 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
11676 alternative 1 (r, v). Remove isa attribute.
11677 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
11678 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
11679 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
11681 2017-05-23 Tom de Vries <tom@codesourcery.com>
11683 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
11686 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11688 * cgraphunit.c (symbol_table::process_new_functions): Update.
11689 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
11690 (inline_generate_summary): Rename to ...
11691 (ipa_fn_summary_generate): ... this one.
11692 (inline_read_summary): Rename to ...
11693 (ipa_fn_summary_read): ... this one.
11694 (inline_write_summary): Rename to ...
11695 (ipa_fn_summary_write): ... this one.
11696 (inline_free_summary): Rename to ...
11697 (ipa_free_fn_summary): ... this one.
11698 (pass_data_local_fn_summary, pass_local_fn_summary,
11699 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
11700 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
11701 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
11702 make_pass_ipa_fn_summary): New.
11703 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
11704 inline_write_summary, inline_free_summary): Remove.
11705 (ipa_free_fn_summary) : New.
11706 * ipa-inline.c (ipa_inline): Update.
11707 (pass_ipa_inline): Do not generate summaries.
11708 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
11710 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
11711 and add pass_ipa_fn_summary.
11712 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
11714 (make_pass_inline_parameters): Remove.
11716 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
11718 * omp-low.c (struct omp_context): Remove "default_kind" member.
11721 * omp-offload.c (execute_oacc_device_lower): Remove the
11722 parallelism dimensions function attributes for unparallelized
11723 OpenACC kernels constructs.
11725 2017-05-23 Martin Liska <mliska@suse.cz>
11727 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
11729 (cgraph_edge::make_speculative): Likewise.
11730 (cgraph_edge::resolve_speculation): Likewise.
11731 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
11732 (cgraph_node::dump): Likewise.
11733 * cgraph.h: Likewise.
11734 * cgraphunit.c (analyze_functions): Likewise.
11735 (symbol_table::compile): Likewise.
11736 * ipa-cp.c (print_all_lattices): Likewise.
11737 (determine_versionability): Likewise.
11738 (initialize_node_lattices): Likewise.
11739 (ipcp_verify_propagated_values): Likewise.
11740 (estimate_local_effects): Likewise.
11741 (update_profiling_info): Likewise.
11742 (create_specialized_node): Likewise.
11743 (perhaps_add_new_callers): Likewise.
11744 (decide_about_value): Likewise.
11745 (decide_whether_version_node): Likewise.
11746 (identify_dead_nodes): Likewise.
11747 (ipcp_store_bits_results): Likewise.
11748 * ipa-devirt.c (dump_targets): Likewise.
11749 (ipa_devirt): Likewise.
11750 * ipa-icf.c (sem_item::dump): Likewise.
11751 (sem_function::equals): Likewise.
11752 (sem_variable::equals): Likewise.
11753 (sem_item_optimizer::read_section): Likewise.
11754 (sem_item_optimizer::execute): Likewise.
11755 (congruence_class::dump): Likewise.
11756 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
11757 (dump_inline_summary): Likewise.
11758 (estimate_node_size_and_time): Likewise.
11759 (inline_analyze_function): Likewise.
11760 * ipa-inline-transform.c (inline_call): Likewise.
11761 * ipa-inline.c (report_inline_failed_reason): Likewise.
11762 (want_early_inline_function_p): Likewise.
11763 (edge_badness): Likewise.
11764 (update_edge_key): Likewise.
11765 (inline_small_functions): Likewise.
11766 * ipa-profile.c (ipa_profile): Likewise.
11767 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
11768 (ipa_make_edge_direct_to_target): Likewise.
11769 (remove_described_reference): Likewise.
11770 (ipa_impossible_devirt_target): Likewise.
11771 (propagate_controlled_uses): Likewise.
11772 (ipa_print_node_params): Likewise.
11773 (ipcp_transform_function): Likewise.
11774 * ipa-pure-const.c (pure_const_read_summary): Likewise.
11775 (propagate_pure_const): Likewise.
11776 * ipa-reference.c (generate_summary): Likewise.
11777 (read_write_all_from_decl): Likewise.
11778 (propagate): Likewise.
11779 (ipa_reference_read_optimization_summary): Likewise.
11780 * ipa-utils.c (ipa_merge_profiles): Likewise.
11781 * ipa.c (walk_polymorphic_call_targets): Likewise.
11782 (symbol_table::remove_unreachable_nodes): Likewise.
11783 (ipa_single_use): Likewise.
11784 * passes.c (execute_todo): Likewise.
11785 * predict.c (drop_profile): Likewise.
11786 * symtab.c (symtab_node::get_dump_name): New function.
11787 (symtab_node::dump_name): Likewise.
11788 (symtab_node::dump_asm_name): Likewise.
11789 (symtab_node::dump_references): Likewise.
11790 (symtab_node::dump_referring): Likewise.
11791 (symtab_node::dump_base): Likewise.
11792 (symtab_node::debug_symtab): Likewise.
11793 * tree-sra.c (convert_callers_for_node): Likewise.
11794 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
11795 * value-prof.c (init_node_map): Likewise.
11797 2017-05-23 Martin Liska <mliska@suse.cz>
11799 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
11800 and symtab_node::debug_symtab to symbol_table::debug.
11801 * cgraphunit.c (analyze_functions): Use the renamed function.
11802 (symbol_table::compile): Likewise.
11803 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
11804 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
11805 * passes.c (execute_todo): Likewise.
11806 * symtab.c (symbol_table::dump): New function.
11807 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
11809 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11811 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
11812 that nonconst implies exec.
11814 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11816 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
11817 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
11818 (inline_edge_summary_vec): Turn into ...
11819 (ipa_call_summaries): ... this one.
11820 (redirect_to_unreachable, edge_set_predicate,
11821 evaluate_properties_for_edge, inline_summary_alloc,
11822 reset_ipa_call_summary, reset_inline_summary,
11823 inline_summary_t::duplicate): Update.
11824 (inline_edge_duplication_hook): Turn to ...
11825 (ipa_call_summary_t::duplicate): ... this one.
11826 (inline_edge_removal_hook): Turn to ...
11827 (ipa_call_summary_t::remove): ... this one.
11828 (dump_inline_edge_summary): Turn to ...
11829 (dump_ipa_call_summary): ... this one.
11830 (estimate_function_body_sizes): Update.
11831 (inline_update_callee_summaries): Update.
11832 (remap_edge_change_prob): Update.
11833 (remap_edge_summaries): Update.
11834 (inline_merge_summary): Update.
11835 (do_estimate_edge_time): Update.
11836 (inline_generate_summary): Update.
11837 (inline_read_section): Update.
11838 (inline_read_summary): Update.
11839 (inline_free_summary): Update.
11840 * ipa-inline.c (can_inline_edge_p): Update.
11841 (compute_inlined_call_time): Update.
11842 (want_inline_small_function_p): Update.
11843 (edge_badness): Update.
11844 (early_inliner): Update.
11845 * ipa-inline.h (inline_edge_summary): Turn to ...
11846 (ipa_call_summary): ... this one.
11847 (ipa_call_summary_t): New class.
11848 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
11849 (ipa_call_summaries): New.
11850 (inline_edge_summary): Remove.
11851 (estimate_edge_growth): Update.
11852 * ipa-profile.c (ipa_propagate_frequency_1): Update.
11853 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11854 * ipa-split.c (execute_split_functions): Update.
11855 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
11857 2017-05-23 Tom de Vries <tom@codesourcery.com>
11859 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
11860 attributes): Document rdrand effective target.
11862 2017-05-23 Tom de Vries <tom@codesourcery.com>
11864 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
11865 attributes): Sort alphabetically.
11867 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
11869 * config/avr/genmultilib.awk: Use gsub instead of gensub.
11871 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
11874 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
11875 V2DF/V2DI splat into two separate patterns, one that handles
11876 registers, and the other that only handles memory. Drop support
11877 for splatting from a GPR on ISA 2.07 and then splitting the
11878 splat into direct move and splat.
11879 (vsx_splat_<mode>_reg): Likewise.
11880 (vsx_splat_<mode>_mem): Likewise.
11882 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
11884 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
11886 2017-05-22 Jakub Jelinek <jakub@redhat.com>
11888 PR middle-end/80809
11889 * omp-low.c (finish_taskreg_remap): New function.
11890 (finish_taskreg_scan): If unit size of ctx->record_type
11891 is non-constant, unshare the size expression and replace
11892 decls in it with possible outer var refs.
11894 PR middle-end/80809
11895 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
11896 GOVD_SHARED rather than GOVD_PRIVATE with it.
11897 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
11898 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
11900 PR middle-end/80853
11901 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
11902 as last argument to build_outer_var_ref for pointer bases of array
11903 section reductions.
11905 2017-05-19 Martin Sebor <msebor@redhat.com>
11907 * print-tree.c (print_node): Print DECL_READ_P flag.
11909 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11911 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
11912 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
11913 * cgraph.c: Likewise.
11914 * cgraphunit.c: Likewise.
11915 * gengtype.c: Likewise.
11916 * ipa-cp.c: Likewise.
11917 * ipa-devirt.c: Likewise.
11918 * ipa-icf.c: Likewise.
11919 * ipa-predicate.c: Likewise.
11920 * ipa-profile.c: Likewise.
11921 * ipa-prop.c: Likewise.
11922 * ipa-split.c: Likewise.
11924 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
11925 edge_predicate_pool, dump_inline_hints,
11926 inline_summary::account_size_time, redirect_to_unreachable,
11927 edge_set_predicate, set_hint_predicate,
11928 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
11929 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
11930 inline_summary_t::remove, remap_hint_predicate_after_duplication,
11931 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
11932 ipa_call_summary_t::remove, initialize_growth_caches,
11933 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
11934 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
11935 mark_modified, unmodified_parm_1, unmodified_parm,
11936 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
11937 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
11938 compute_bb_predicates, will_be_nonconstant_expr_predicate,
11939 will_be_nonconstant_predicate, record_modified_bb_info,
11940 get_minimal_bb, record_modified, param_change_prob,
11941 phi_result_unknown_predicate, predicate_for_phi_result,
11942 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
11943 estimate_function_body_sizes, compute_inline_parameters,
11944 compute_inline_parameters_for_curren, pass_data_inline_parameters,
11945 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
11946 inline_update_callee_summaries, remap_edge_change_prob,
11947 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
11948 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
11949 inline_analyze_function, inline_summary_t::insert,
11950 inline_generate_summary, read_ipa_call_summary, inline_read_section,
11951 inline_read_summary, write_ipa_call_summary, inline_write_summary,
11952 inline_free_summary): Move to ipa-fnsummary.h
11953 (predicate_t): Remove.
11954 * ipa-fnsummary.c: New file.
11955 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
11956 (enum inline_hints_vals, inline_hints, agg_position_info,
11957 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
11958 inline_summaries, ipa_call_summary, ipa_call_summary_t,
11959 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
11960 dump_inline_summary, dump_inline_hints, inline_generate_summary,
11961 inline_read_summary, inline_write_summary, inline_free_summary,
11962 inline_analyze_function, initialize_inline_failed,
11963 inline_merge_summary, inline_update_overall_summary,
11964 compute_inline_parameters): Move to ipa-fnsummary.h
11965 * ipa-fnsummary.h: New file.
11966 * ipa-inline-transform.h: Include ipa-inline.h.
11967 * ipa-inline.c: LIkewise.
11969 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11971 * ipa-inline.c (edge_badness): Use inlined_time instead of
11972 inline_summaries->get.
11974 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11976 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
11978 2017-05-22 Nathan Sidwell <nathan@acm.org>
11980 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
11981 (fdump-lang): Document 'raw' option.
11982 * dumpfile.h (TDI_tu): Delete.
11983 * dumpfile.c (dump_files): Remove translation-unit.
11984 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
11986 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
11988 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
11989 command option from $(AWK) call.
11990 * config/avr/genmultilib.awk: Simplify and rewrite so that it
11991 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
11992 [FORMAT]: Remove handling of variable.
11993 * config/avr/t-multilib: Regenerate.
11995 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
11997 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
11999 (dump_inline_summary): Do not print self_time.
12000 (estimate_function_body_sizes): Do not set self_time.
12001 (compute_inline_parameters): Likewise.
12002 (inline_read_section, inline_write_summary): Do not stream self_time.
12003 * ipa-inline.h (inline_summary): Drop self_time.
12005 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12007 * ipa-inline-analysis.c (account_size_time): Rename to ...
12008 (inline_summary::account_size_time): ... this one.
12009 (reset_ipa_call_summary): Turn to ...
12010 (ipa_call_summary::reset): ... this one.
12011 (reset_inline_summary): Turn to ...
12012 (inline_summary::reset): ... this one.
12013 (inline_summary_t::remove): Update.
12014 (inline_summary_t::duplicate): Update.
12015 (ipa_call_summary_t::remove): Update.
12016 (dump_inline_summary): Update.
12017 (estimate_function_body_sizes): Update.
12018 (compute_inline_parameters): Update.
12019 (estimate_node_size_and_time): Update.
12020 (inline_merge_summary): Update.
12021 (inline_update_overall_summary): Update.
12022 (inline_read_section): Update.
12023 (inline_write_summary): Update.
12024 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
12025 add account_size_time and reset member functions.
12026 (ipa_call_summary): Add reset function.
12027 * ipa-predicate.h (predicate::operator &): Constify.
12029 2017-05-22 Richard Biener <rguenther@suse.de>
12031 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
12033 2017-05-19 Jason Merrill <jason@redhat.com>
12035 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
12037 2017-05-19 Marek Polacek <polacek@redhat.com>
12040 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
12041 TYPE_OVERFLOW_WRAPS checks.
12043 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
12045 * tree-core.h (enum omp_clause_default_kind): Add
12046 "OMP_CLAUSE_DEFAULT_PRESENT".
12047 * tree-pretty-print.c (dump_omp_clause): Handle it.
12048 * gimplify.c (enum gimplify_omp_var_data): Add
12049 "GOVD_MAP_FORCE_PRESENT".
12050 (gimplify_adjust_omp_clauses_1): Map it to
12051 "GOMP_MAP_FORCE_PRESENT".
12052 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
12054 * gimplify.c (oacc_default_clause): Clarify.
12056 2017-05-19 Nathan Sidwell <nathan@acm.org>
12058 LANG_HOOK_REGISTER_DUMPS
12059 * toplev.c (general_init): Call register dump lang hook.
12060 * doc/invoke.texi: Document -fdump-lang option family.
12061 * dumpfile.c (dump_files): Remove class dump here.
12062 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
12063 * dumpfile.h (tree_dump_index): Remove TDI_class.
12064 * langhooks-def.h (lhd_register_dumps): Declare.
12065 (LANG_HOOKS_REGISTER_DUMPS): Define.
12066 (LANG_HOOKS_INITIALIZER): Add it.
12067 * langhooks.c (lhd_register_dumps): Define.
12068 * langhooks.h (struct lang_hooks): Add register_dumps.
12070 2017-05-19 Nathan Sidwell <nathan@acm.org>
12072 * context.h (context::set_passes): New.
12073 * context.c (context::context): Do not create pass manager.
12074 * toplev.c (general_init): Create pass manager here.
12076 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
12078 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
12079 use this splitter if two add or or instructions would also work for
12080 the constant we want to generate.
12082 2017-05-19 Richard Biener <rguenther@suse.de>
12085 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
12086 predicate evaluation.
12088 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12090 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
12092 * ipa-inline.c (want_inline_small_function_p): Do not cast to
12095 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12097 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
12098 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
12099 (inline_edge_summary_vec): Turn into ...
12100 (ipa_call_summaries): ... this one.
12101 (redirect_to_unreachable, edge_set_predicate,
12102 evaluate_properties_for_edge, inline_summary_alloc,
12103 reset_ipa_call_summary, reset_inline_summary,
12104 inline_summary_t::duplicate): Update.
12105 (inline_edge_duplication_hook): Turn to ...
12106 (ipa_call_summary_t::duplicate): ... this one.
12107 (inline_edge_removal_hook): Turn to ...
12108 (ipa_call_summary_t::remove): ... this one.
12109 (dump_inline_edge_summary): Turn to ...
12110 (dump_ipa_call_summary): ... this one.
12111 (estimate_function_body_sizes): Update.
12112 (inline_update_callee_summaries): Update.
12113 (remap_edge_change_prob): Update.
12114 (remap_edge_summaries): Update.
12115 (inline_merge_summary): Update.
12116 (do_estimate_edge_time): Update.
12117 (inline_generate_summary): Update.
12118 (inline_read_section): Update.
12119 (inline_read_summary): Update.
12120 (inline_free_summary): Update.
12121 * ipa-inline.c (can_inline_edge_p): Update.
12122 (compute_inlined_call_time): Update.
12123 (want_inline_small_function_p): Update.
12124 (edge_badness): Update.
12125 (early_inliner): Update.
12126 * ipa-inline.h (inline_edge_summary): Turn to ...
12127 (ipa_call_summary): ... this one.
12128 (ipa_call_summary_t): New class.
12129 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
12130 (ipa_call_summaries): New.
12131 (inline_edge_summary): Remove.
12132 (estimate_edge_growth): Update.
12133 * ipa-profile.c (ipa_propagate_frequency_1): Update.
12134 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12135 * ipa-split.c (execute_split_functions): Update.
12136 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
12138 2017-05-19 Richard Biener <rguenther@suse.de>
12140 PR middle-end/80764
12141 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
12143 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
12145 * config/rs6000/rs6000.c (struct machine_function): Add field
12146 fpr_is_wrapped_separately.
12147 (rs6000_get_separate_components): Use 64 components. Handle the
12148 new FPR components.
12149 (rs6000_components_for_bb): Handle the FPR components.
12150 (rs6000_emit_prologue_components): Handle the FPR components.
12151 (rs6000_emit_epilogue_components): Handle the FPR components.
12152 (rs6000_set_handled_components): Handle the FPR components.
12153 (rs6000_emit_prologue): Don't output prologue code for those FPRs
12154 that are already separately shrink-wrapped.
12155 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
12156 that are already separately shrink-wrapped.
12158 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
12161 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
12164 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
12165 (define_peephole2 for Altivec d-form load): Add peepholes to catch
12166 cases where the register allocator uses a move and an offsettable
12167 memory operation to/from a FPR register on ISA 2.06/2.07.
12168 (define_peephole2 for Altivec d-form store): Likewise.
12170 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
12173 * config/i386/mmx.md (*mov<mode>_internal): Enable
12174 alternatives 11, 12, 13 and 14 also for 32bit targets.
12175 Remove alternatives 15, 16, 17 and 18.
12176 * config/i386/sse.md (vec_concatv2di): Change
12177 alternative (!x, *y) to (x, ?!*Yn).
12179 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
12181 * dumpfile.h (enum dump_kind): Remove stray comma.
12183 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
12185 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
12186 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
12187 predicate::num_conditions
12188 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
12189 (CHANGED): turn into predicate::changed.
12190 (agg_position_info): Move to ipa-predicate.h
12191 (add_condition, predicate::add_clause, predicate::operator &=,
12192 predicate::or_with, predicate::evaluate, predicate::probability,
12193 dump_condition, dump_clause, predicate::dump,
12194 predicate::remap_after_duplication, predicate::remap_after_inlining,
12195 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
12196 (evaluate_conditions_for_known_args): Update.
12197 (set_cond_stmt_execution_predicate): Update.
12198 * ipa-inline.h: Include ipa-predicate.h
12199 (condition, inline_param_summary, conditions, agg_position_info,
12200 predicate): Move to ipa-predicate.h
12201 * ipa-predicate.c: New file.
12202 * ipa-predicate.h: New file.
12204 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
12206 * final.c (leaf_function_p): Check we are not in a sequence.
12208 2017-05-18 Martin Liska <mliska@suse.cz>
12210 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
12211 * dumpfile.c (dump_register): Use new enum dump_kind.
12212 (get_dump_file_name): Likewise.
12213 (dump_enable_all): Likewise.
12214 (dump_switch_p_1): Likewise.
12215 (enable_rtl_dump_file): Remove usage of TDF_RTL.
12216 * dumpfile.h (enum dump_kind): New enum type.
12217 (struct dump_file_info): Create constructor and
12218 format fields and comments.
12219 * passes.c (pass_manager::register_one_dump_file):
12221 * statistics.c (statistics_early_init): Likewise.
12222 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
12223 TDF_TREE with TDF_SLIM.
12224 (gather_memory_references_ref): Likewise.
12226 2017-05-18 Martin Liska <mliska@suse.cz>
12228 * vec.h (struct vnull): Use it.
12230 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
12232 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
12233 (true_predicate, false_predicate, true_predicate_p,
12234 false_predicate_p): Remove.
12235 (single_cond_predicate, not_inlined_predicate): Turn to member function
12237 (add_condition): Update.
12238 (add_clause): Turn to...
12239 (predicate::add_clause): ... this one; update; allow passing NULL
12241 (and_predicates): Turn to ...
12242 (predicate::operator &=): ... this one.
12243 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
12244 (or_predicates): Turn to ...
12245 (predicate::or_with): ... this one.
12246 (evaluate_predicate): Turn to ...
12247 (predicate::evaluate): ... this one.
12248 (predicate_probability): Turn to ...
12249 (predicate::probability): ... this one.
12250 (dump_condition): Update.
12251 (dump_predicate): Turn to ...
12252 (predicate::dump): ... this one.
12253 (account_size_time): Update.
12254 (edge_set_predicate): Update.
12255 (set_hint_predicate): UPdate.
12256 (evaluate_conditions_for_known_args): Update.
12257 (evaluate_properties_for_edge): Update.
12258 (remap_predicate_after_duplication): Turn to...
12259 (predicate::remap_after_duplication): ... this one.
12260 (remap_hint_predicate_after_duplication): Update.
12261 (inline_summary_t::duplicate): UPdate.
12262 (dump_inline_edge_summary): Update.
12263 (dump_inline_summary): Update.
12264 (set_cond_stmt_execution_predicate): Update.
12265 (set_switch_stmt_execution_predicate): Update.
12266 (compute_bb_predicates): Update.
12267 (will_be_nonconstant_expr_predicate): Update.
12268 (will_be_nonconstant_predicate): Update.
12269 (phi_result_unknown_predicate): Update.
12270 (predicate_for_phi_result): Update.
12271 (array_index_predicate): Update.
12272 (estimate_function_body_sizes): Update.
12273 (estimate_node_size_and_time): Update.
12274 (estimate_ipcp_clone_size_and_time): Update.
12275 (remap_predicate): Rename to ...
12276 (predicate::remap_after_inlining): ... this one.
12277 (remap_hint_predicate): Update.
12278 (inline_merge_summary): Update.
12279 (inline_update_overall_summary): Update.
12280 (estimate_size_after_inlining): Update.
12281 (read_predicate): Rename to ...
12282 (predicate::stream_in): ... this one.
12283 (read_inline_edge_summary): Update.
12284 (write_predicate): Rename to ...
12285 (predicate::stream_out): ... this one.
12286 (write_inline_edge_summary): Update.
12287 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
12288 (clause_t): Turn to uint32_t
12289 (predicate): Turn to class; implement constructor and operators
12291 (size_time_entry): Update.
12292 (inline_summary): Update.
12293 (inline_edge_summary): Update.
12295 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
12297 * fold-const.c (fold_binary_loc): Move transformation...
12298 * match.pd (C - X CMP X): ... here.
12300 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
12302 * config/sparc/sparc.c (sparc_option_override): Set function
12303 alignment for -mcpu=niagara7 to 64 to match the I$ line.
12304 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
12306 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
12308 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
12310 2017-05-18 Marek Polacek <polacek@redhat.com>
12313 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
12314 (pass_ubsan::execute): Call gimple_assign_single_p instead of
12315 gimple_assign_load_p.
12317 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
12319 PR middle-end/80692
12320 * real.c (do_compare): Give decimal_do_compare preference over
12321 comparing just the signs.
12323 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
12325 * doc/md.texi (Canonicalization of Instructions): Describe the
12326 canonical form of instructions that inherently set a condition
12329 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
12331 PR middle-end/80775
12332 * tree-cfg.c: Move deletion of unreachable case statements to after
12333 the merging of consecutive case labels.
12335 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
12337 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
12338 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
12339 restoring of callee-saved registers.
12341 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
12343 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
12344 * config/visium/visium.c (single_set_and_flags): Likewise.
12345 * config/visium/visium.md (Substitutions): Likewise.
12347 2017-05-17 Martin Liska <mliska@suse.cz>
12349 * cfg.c: Introduce dump_flags_t type and
12350 use it instead of int type.
12352 * cfghooks.c: Likewise.
12353 * cfghooks.h (struct cfg_hooks): Likewise.
12354 * cfgrtl.c: Likewise.
12355 * cfgrtl.h: Likewise.
12356 * cgraph.c (cgraph_node::get_body): Likewise.
12357 * coretypes.h: Likewise.
12358 * domwalk.c: Likewise.
12359 * domwalk.h: Likewise.
12360 * dumpfile.c (struct dump_option_value_info): Likewise.
12361 (dump_enable_all): Likewise.
12362 (dump_switch_p_1): Likewise.
12363 (opt_info_switch_p): Likewise.
12364 * dumpfile.h (enum tree_dump_index): Likewise.
12365 (struct dump_file_info): Likewise.
12366 * genemit.c: Likewise.
12367 * generic-match-head.c: Likewise.
12368 * gengtype.c (open_base_files): Likewise.
12369 * gimple-pretty-print.c: Likewise.
12370 * gimple-pretty-print.h: Likewise.
12371 * graph.c (print_graph_cfg): Likewise.
12372 * graphite-scop-detection.c (dot_all_sese): Likewise.
12373 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
12374 * loop-unroll.c (report_unroll): Likewise.
12375 * passes.c (pass_manager::register_one_dump_file): Likewise.
12376 * print-tree.c: Likewise.
12377 * statistics.c: Likewise.
12378 * tree-cfg.c: Likewise.
12379 * tree-cfg.h: Likewise.
12380 * tree-dfa.c: Likewise.
12381 * tree-dfa.h: Likewise.
12382 * tree-dump.c (dump_function): Likewise.
12383 * tree-dump.h (struct dump_info): Likewise.
12384 * tree-pretty-print.c: Likewise.
12385 * tree-pretty-print.h: Likewise.
12386 * tree-ssa-live.c: Likewise.
12387 * tree-ssa-live.h: Likewise.
12388 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
12389 * tree-vect-loop.c: Likewise.
12390 * tree-vect-slp.c: Likewise.
12392 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
12393 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12395 PR tree-optimization/80457
12396 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
12397 of all arguments to a statement as scalar_to_vec operations.
12398 (vectorizable_call): Adjust call to vect_model_simple_cost for
12400 (vectorizable_conversion): Likewise.
12401 (vectorizable_assignment): Likewise.
12402 (vectorizable_shift): Likewise.
12403 (vectorizable_operation): Likewise.
12404 (vectorizable_comparison): Likewise.
12405 (vect_is_simple_cond): Record the def types for operands.
12406 (vectorizable_condition): Likewise, call vect_model_simple_cost.
12407 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
12408 for statement argument count.
12410 2017-05-16 Carl Love <cel@us.ibm.com>
12412 * config/rs6000/rs6000-c: Add support for built-in functions
12413 vector unsigned long long vec_bperm (vector unsigned long long,
12414 vector unsigned char)
12415 vector signed long long vec_mule (vector signed int,
12417 vector unsigned long long vec_mule (vector unsigned int,
12418 vector unsigned int)
12419 vector signed long long vec_mulo (vector signed int,
12421 vector unsigned long long vec_mulo (vector unsigned int,
12422 vector unsigned int)
12423 vector signed char vec_sldw (vector signed char,
12424 vector signed char,
12426 vector unsigned char vec_sldw (vector unsigned char,
12427 vector unsigned char,
12429 vector signed short vec_sldw (vector signed short,
12430 vector signed short,
12432 vector unsigned short vec_sldw (vector unsigned short,
12433 vector unsigned short,
12435 vector signed int vec_sldw (vector signed int,
12438 vector unsigned int vec_sldw (vector unsigned int,
12439 vector unsigned int,
12441 vector signed long long vec_sldw (vector signed long long,
12442 vector signed long long,
12444 vector unsigned long long vec_sldw (vector unsigned long long,
12445 vector unsigned long long,
12447 * config/rs6000/rs6000-c: Add support for built-in functions
12448 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
12449 * config/rs6000/altivec.h: Add defintion for vec_sldw.
12450 * doc/extend.texi: Update the built-in documentation for the
12451 new built-in functions.
12453 2017-05-16 Marek Polacek <polacek@redhat.com>
12457 * tree.c (save_expr): Don't fold the expression.
12459 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
12461 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
12462 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
12463 and (?*y,m). Update insn attributes.
12465 2017-05-16 Martin Liska <mliska@suse.cz>
12467 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
12468 flags argument of print_gimple_stmt, print_gimple_expr,
12469 print_generic_stmt and print_generic_expr.
12470 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
12471 * coretypes.h: Likewise.
12472 * except.c (dump_eh_tree): Likewise.
12473 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12474 * gimple-pretty-print.h: Likewise.
12475 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
12476 (backprop::push_to_worklist): Likewise.
12477 (backprop::pop_from_worklist): Likewise.
12478 (backprop::process_use): Likewise.
12479 (backprop::intersect_uses): Likewise.
12480 (note_replacement): Likewise.
12481 * gimple-ssa-store-merging.c
12482 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
12483 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
12484 (pass_store_merging::execute): Likewise.
12485 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
12486 (ssa_base_cand_dump_callback): Likewise.
12487 (dump_incr_vec): Likewise.
12488 (replace_refs): Likewise.
12489 (replace_mult_candidate): Likewise.
12490 (create_add_on_incoming_edge): Likewise.
12491 (create_phi_basis): Likewise.
12492 (insert_initializers): Likewise.
12493 (all_phi_incrs_profitable): Likewise.
12494 (introduce_cast_before_cand): Likewise.
12495 (replace_one_candidate): Likewise.
12496 * gimplify.c (gimplify_expr): Likewise.
12497 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
12498 (set_rename): Likewise.
12499 (rename_uses): Likewise.
12500 (copy_loop_phi_nodes): Likewise.
12501 (add_close_phis_to_merge_points): Likewise.
12502 (copy_loop_close_phi_args): Likewise.
12503 (copy_cond_phi_args): Likewise.
12504 (graphite_copy_stmts_from_block): Likewise.
12505 (translate_pending_phi_nodes): Likewise.
12506 * graphite-poly.c (print_pdr): Likewise.
12507 (dump_gbb_cases): Likewise.
12508 (dump_gbb_conditions): Likewise.
12509 (print_scop_params): Likewise.
12510 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
12511 (build_cross_bb_scalars_use): Likewise.
12512 (gather_bbs::before_dom_children): Likewise.
12513 * hsa-dump.c (dump_hsa_immed): Likewise.
12514 * ipa-cp.c (print_ipcp_constant_value): Likewise.
12515 (get_replacement_map): Likewise.
12516 * ipa-inline-analysis.c (dump_condition): Likewise.
12517 (estimate_function_body_sizes): Likewise.
12518 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
12519 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
12520 * ipa-prop.c (ipa_dump_param): Likewise.
12521 (ipa_print_node_jump_functions_for_edge): Likewise.
12522 (ipa_modify_call_arguments): Likewise.
12523 (ipa_modify_expr): Likewise.
12524 (ipa_dump_param_adjustments): Likewise.
12525 (ipa_dump_agg_replacement_values): Likewise.
12526 (ipcp_modif_dom_walker::before_dom_children): Likewise.
12527 * ipa-pure-const.c (check_stmt): Likewise.
12528 (pass_nothrow::execute): Likewise.
12529 * ipa-split.c (execute_split_functions): Likewise.
12530 * omp-offload.c (dump_oacc_loop_part): Likewise.
12531 (dump_oacc_loop): Likewise.
12532 * trans-mem.c (tm_log_emit): Likewise.
12533 (tm_memopt_accumulate_memops): Likewise.
12534 (dump_tm_memopt_set): Likewise.
12535 (dump_tm_memopt_transform): Likewise.
12536 * tree-cfg.c (gimple_verify_flow_info): Likewise.
12537 (print_loop): Likewise.
12538 * tree-chkp-opt.c (chkp_print_addr): Likewise.
12539 (chkp_gather_checks_info): Likewise.
12540 (chkp_get_check_result): Likewise.
12541 (chkp_remove_check_if_pass): Likewise.
12542 (chkp_use_outer_bounds_if_possible): Likewise.
12543 (chkp_reduce_bounds_lifetime): Likewise.
12544 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
12545 (chkp_mark_completed_bounds): Likewise.
12546 (chkp_register_incomplete_bounds): Likewise.
12547 (chkp_mark_invalid_bounds): Likewise.
12548 (chkp_maybe_copy_and_register_bounds): Likewise.
12549 (chkp_build_returned_bound): Likewise.
12550 (chkp_get_bound_for_parm): Likewise.
12551 (chkp_build_bndldx): Likewise.
12552 (chkp_get_bounds_by_definition): Likewise.
12553 (chkp_generate_extern_var_bounds): Likewise.
12554 (chkp_get_bounds_for_decl_addr): Likewise.
12555 * tree-chrec.c (chrec_apply): Likewise.
12556 * tree-data-ref.c (dump_data_reference): Likewise.
12557 (dump_subscript): Likewise.
12558 (dump_data_dependence_relation): Likewise.
12559 (analyze_overlapping_iterations): Likewise.
12560 * tree-inline.c (expand_call_inline): Likewise.
12561 (tree_function_versioning): Likewise.
12562 * tree-into-ssa.c (dump_defs_stack): Likewise.
12563 (dump_currdefs): Likewise.
12564 (dump_names_replaced_by): Likewise.
12565 (dump_update_ssa): Likewise.
12566 (update_ssa): Likewise.
12567 * tree-object-size.c (pass_object_sizes::execute): Likewise.
12568 * tree-parloops.c (build_new_reduction): Likewise.
12569 (try_create_reduction_list): Likewise.
12570 (ref_conflicts_with_region): Likewise.
12571 (oacc_entry_exit_ok_1): Likewise.
12572 (oacc_entry_exit_single_gang): Likewise.
12573 * tree-pretty-print.h: Likewise.
12574 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
12575 (get_scalar_evolution): Likewise.
12576 (add_to_evolution): Likewise.
12577 (get_loop_exit_condition): Likewise.
12578 (analyze_evolution_in_loop): Likewise.
12579 (analyze_initial_condition): Likewise.
12580 (analyze_scalar_evolution): Likewise.
12581 (instantiate_scev): Likewise.
12582 (number_of_latch_executions): Likewise.
12583 (gather_chrec_stats): Likewise.
12584 (final_value_replacement_loop): Likewise.
12585 (scev_const_prop): Likewise.
12586 * tree-sra.c (dump_access): Likewise.
12587 (disqualify_candidate): Likewise.
12588 (create_access): Likewise.
12589 (reject): Likewise.
12590 (maybe_add_sra_candidate): Likewise.
12591 (create_access_replacement): Likewise.
12592 (analyze_access_subtree): Likewise.
12593 (analyze_all_variable_accesses): Likewise.
12594 (sra_modify_assign): Likewise.
12595 (initialize_constant_pool_replacements): Likewise.
12596 (find_param_candidates): Likewise.
12597 (decide_one_param_reduction): Likewise.
12598 (replace_removed_params_ssa_names): Likewise.
12599 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
12600 * tree-ssa-copy.c (dump_copy_of): Likewise.
12601 (copy_prop_visit_cond_stmt): Likewise.
12602 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
12603 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
12604 (record_equivalences_from_stmt): Likewise.
12605 * tree-ssa-dse.c (compute_trims): Likewise.
12606 (delete_dead_call): Likewise.
12607 (delete_dead_assignment): Likewise.
12608 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
12609 (forward_propagate_into_cond): Likewise.
12610 (pass_forwprop::execute): Likewise.
12611 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
12612 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
12614 (move_computations_worker): Likewise.
12615 (execute_sm): Likewise.
12616 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
12617 (remove_exits_and_undefined_stmts): Likewise.
12618 (remove_redundant_iv_tests): Likewise.
12619 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
12620 (adjust_iv_update_pos): Likewise.
12621 * tree-ssa-math-opts.c (bswap_replace): Likewise.
12622 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
12623 (value_replacement): Likewise.
12624 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
12625 * tree-ssa-pre.c (print_pre_expr): Likewise.
12626 (get_representative_for): Likewise.
12627 (create_expression_by_pieces): Likewise.
12628 (insert_into_preds_of_block): Likewise.
12629 (eliminate_insert): Likewise.
12630 (eliminate_dom_walker::before_dom_children): Likewise.
12631 (eliminate): Likewise.
12632 (remove_dead_inserted_code): Likewise.
12633 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
12634 * tree-ssa-reassoc.c (get_rank): Likewise.
12635 (eliminate_duplicate_pair): Likewise.
12636 (eliminate_plus_minus_pair): Likewise.
12637 (eliminate_not_pairs): Likewise.
12638 (undistribute_ops_list): Likewise.
12639 (eliminate_redundant_comparison): Likewise.
12640 (update_range_test): Likewise.
12641 (optimize_range_tests_var_bound): Likewise.
12642 (optimize_vec_cond_expr): Likewise.
12643 (rewrite_expr_tree): Likewise.
12644 (rewrite_expr_tree_parallel): Likewise.
12645 (linearize_expr): Likewise.
12646 (break_up_subtract): Likewise.
12647 (linearize_expr_tree): Likewise.
12648 (attempt_builtin_powi): Likewise.
12649 (attempt_builtin_copysign): Likewise.
12650 (transform_stmt_to_copy): Likewise.
12651 (transform_stmt_to_multiply): Likewise.
12652 (dump_ops_vector): Likewise.
12653 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
12654 (print_scc): Likewise.
12655 (set_ssa_val_to): Likewise.
12656 (visit_reference_op_store): Likewise.
12657 (visit_use): Likewise.
12658 (sccvn_dom_walker::before_dom_children): Likewise.
12659 (run_scc_vn): Likewise.
12660 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
12662 (expr_hash_elt::print): Likewise.
12663 (const_and_copies::pop_to_marker): Likewise.
12664 (const_and_copies::record_const_or_copy_raw): Likewise.
12665 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
12666 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
12667 (dump_predicates): Likewise.
12668 (find_uninit_use): Likewise.
12669 (warn_uninitialized_phi): Likewise.
12670 (pass_late_warn_uninitialized::execute): Likewise.
12671 * tree-ssa.c (verify_vssa): Likewise.
12672 (verify_ssa): Likewise.
12673 (maybe_optimize_var): Likewise.
12674 * tree-vrp.c (dump_value_range): Likewise.
12675 (dump_all_value_ranges): Likewise.
12676 (dump_asserts_for): Likewise.
12677 (register_edge_assert_for_2): Likewise.
12678 (vrp_visit_cond_stmt): Likewise.
12679 (vrp_visit_switch_stmt): Likewise.
12680 (vrp_visit_stmt): Likewise.
12681 (vrp_visit_phi_node): Likewise.
12682 (simplify_cond_using_ranges_1): Likewise.
12683 (fold_predicate_in): Likewise.
12684 (evrp_dom_walker::before_dom_children): Likewise.
12685 (evrp_dom_walker::push_value_range): Likewise.
12686 (evrp_dom_walker::pop_value_range): Likewise.
12687 (execute_early_vrp): Likewise.
12689 2017-05-16 Richard Biener <rguenther@suse.de>
12691 * dwarf2out.c (loc_list_from_tree_1): Do not create
12692 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
12694 2017-05-16 Richard Biener <rguenther@suse.de>
12696 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
12698 (note_variable_value_in_expr): If we resolved the decl ref
12699 do not push to the stack.
12701 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
12703 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
12704 operations in fast-math mode.
12705 (vaddq_f16): Likewise.
12706 (vmul_f16): Likewise.
12707 (vmulq_f16): Likewise.
12708 (vsub_f16): Likewise.
12709 (vsubq_f16): Likewise.
12710 * config/arm/neon.md (add<mode>3): New.
12712 (fma:<VH:mode>3): New. Also remove outdated comment.
12715 2017-05-16 Martin Liska <mliska@suse.cz>
12719 * ipa-devirt.c (warn_types_mismatch): Fix typo.
12720 (odr_types_equivalent_p): Likewise.
12722 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
12724 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
12726 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
12729 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
12730 non-interunit SSE move alternatives with '?'.
12731 (zero-extendsidi peephole2): New peephole to skip intermediate
12732 general register in SSE zero-extend sequence.
12734 2017-05-15 Jeff Law <law@redhat.com>
12736 * reorg.c (relax_delay_slots): Create a new variable to hold
12737 the temporary target rather than clobbering TARGET_LABEL.
12739 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
12740 missing argument to extract_bit_field call.
12741 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
12743 2017-05-15 Martin Liska <mliska@suse.cz>
12746 * gcc.c (process_command): Do not allow empty argument of -o option.
12748 2017-05-15 Renlin Li <renlin.li@arm.com>
12750 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
12751 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
12752 * config/aarch64/constraints.md (Usf): Add long call check.
12753 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
12754 (call_value): Likewise.
12755 (sibcall): Likewise.
12756 (sibcall_value): Likewise.
12758 (call_value_insn): New.
12759 (sibcall_insn): Update rtx pattern.
12760 (sibcall_value_insn): Likewise.
12761 (call_internal): Remove.
12762 (call_value_internal): Likewise.
12763 (sibcall_internal): Likewise.
12764 (sibcall_value_internal): Likewise.
12765 (call_reg): Likewise.
12766 (call_symbol): Likewise.
12767 (call_value_reg): Likewise.
12768 (call_value_symbol): Likewise.
12770 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
12773 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
12775 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
12777 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
12778 compatible with CCGOCmode and with CCZmode.
12780 2017-05-14 Martin Sebor <msebor@redhat.com>
12782 PR middle-end/77671
12783 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
12784 (gimple_fold_builtin_snprintf): Same.
12785 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
12786 (gimple_fold_builtin_snprintf): Same.
12787 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
12788 of character types.
12789 (is_call_safe): New function.
12790 (try_substitute_return_value): Call it.
12791 (try_simplify_call): New function.
12792 (pass_sprintf_length::handle_gimple_call): Call it.
12794 2017-05-14 Martin Sebor <msebor@redhat.com>
12796 PR middle-end/80669
12797 * builtins.c (expand_builtin_stpncpy): Simplify.
12799 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
12801 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
12802 * config/i386/i386.h
12803 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
12804 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
12805 (struct machine_function): Add new members call_ms2sysv,
12806 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
12807 (struct machine_frame_state): New fields sp_realigned and
12808 sp_realigned_offset.
12809 * config/i386/i386.c
12810 (enum xlogue_stub): New enum.
12811 (enum xlogue_stub_sets): New enum.
12812 (class xlogue_layout): New class.
12813 (struct ix86_frame): New fields stack_realign_allocate_offset,
12814 stack_realign_offset and outlined_save_offset. Modify comments to
12815 detail stack layout when using out-of-line stubs.
12816 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
12817 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
12818 -mcall-ms2sysv-xlogues.
12819 (stub_managed_regs): New static variable.
12820 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
12821 registers managed by out-of-line stub.
12822 (disable_call_ms2sysv_xlogues): New function.
12823 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
12824 m->call_ms2sysv when appropriate and compute frame layout for
12826 (sp_valid_at, fp_valid_at): New inline functions.
12827 (choose_basereg): New function.
12828 (choose_baseaddr): Add align parameter, use choose_basereg and modify
12830 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
12831 Use align parameter of choose_baseaddr to generated aligned SSE movs
12833 (pro_epilogue_adjust_stack): Modify to track
12834 machine_frame_state::sp_realigned.
12835 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
12836 (ix86_nsaved_sseregs): Likewise.
12837 (ix86_emit_save_regs): Likewise.
12838 (ix86_emit_save_regs_using_mov): Likewise.
12839 (ix86_emit_save_sse_regs_using_mov): Likewise.
12840 (get_scratch_register_on_entry): Likewise.
12841 (gen_frame_set): New function.
12842 (gen_frame_load): Likewise.
12843 (gen_frame_store): Likewise.
12844 (emit_outlined_ms2sysv_save): Likewise.
12845 (emit_outlined_ms2sysv_restore): Likewise.
12846 (ix86_expand_prologue): Modify stack re-alignment code and call
12847 emit_outlined_ms2sysv_save when appropriate.
12848 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
12849 parameter rtx_insn *insn, which allows the function to be used to only
12850 generate the notes.
12851 (ix86_expand_epilogue): Modify validity checks of frame and stack
12852 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
12853 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
12854 * config/i386/predicates.md
12855 (save_multiple): New predicate.
12856 (restore_multiple): Likewise.
12857 * config/i386/sse.md
12858 (save_multiple<mode>): New pattern.
12859 (save_multiple_realign<mode>): Likewise.
12860 (restore_multiple<mode>): Likewise.
12861 (restore_multiple_and_return<mode>): Likewise.
12862 (restore_multiple_leave_return<mode>): Likewise.
12863 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
12865 2017-05-14 Julia Koval <julia.koval@intel.com>
12867 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
12868 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
12869 (__builtin_ia32_xsetbv): New builtins.
12870 * config/i386/i386.c (ix86_expand_special_args_builtin):
12872 (ix86_expand_builtin): Special expand for new intrinsics.
12873 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
12874 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
12875 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
12877 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12879 * cfganal.c (inverted_post_order_compute): Change argument type
12881 * cfganal.h (inverted_post_order_compute): Adjust prototype.
12882 * df-core.c (rest_of_handle_df_initialize): Adjust.
12883 (rest_of_handle_df_finish): Likewise.
12884 (df_analyze_1): Likewise.
12885 (df_analyze): Likewise.
12886 (loop_inverted_post_order_compute): Change argument to be a vec *.
12887 (df_analyze_loop): Adjust.
12888 (df_get_n_blocks): Likewise.
12889 (df_get_postorder): Likewise.
12890 * df.h (struct df_d): Change field to be a vec.
12891 * lcm.c (compute_laterin): Adjust.
12892 (compute_available): Likewise.
12893 * lra-lives.c (lra_create_live_ranges_1): Likewise.
12894 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
12895 * tree-ssa-pre.c (compute_antic): Likewise.
12897 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12899 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
12900 (depth_first_search::depth_first_search): Change structure init
12901 function to this constructor.
12902 (depth_first_search::add_bb): Rename function to this member.
12903 (depth_first_search::execute): Likewise.
12904 (flow_dfs_compute_reverse_finish): Adjust.
12906 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12908 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
12909 (longest_simple_path): Likewise.
12910 * shrink-wrap.c (spread_components): Likewise.
12911 (disqualify_problematic_components): Likewise.
12912 (emit_common_heads_for_components): Likewise.
12913 (emit_common_tails_for_components): Likewise.
12914 (insert_prologue_epilogue_for_components): Likewise.
12916 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12918 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
12921 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12923 * df-core.c (df_set_blocks): Start using auto_bitmap.
12924 (df_compact_blocks): Likewise.
12925 * df-problems.c (df_rd_confluence_n): Likewise.
12926 * df-scan.c (df_insn_rescan_all): Likewise.
12927 (df_process_deferred_rescans): Likewise.
12928 (df_update_entry_block_defs): Likewise.
12929 (df_update_exit_block_uses): Likewise.
12930 (df_entry_block_bitmap_verify): Likewise.
12931 (df_exit_block_bitmap_verify): Likewise.
12932 (df_scan_verify): Likewise.
12933 * lra-constraints.c (lra_constraints): Likewise.
12934 (undo_optional_reloads): Likewise.
12935 (lra_undo_inheritance): Likewise.
12936 * lra-remat.c (calculate_gen_cands): Likewise.
12937 (do_remat): Likewise.
12938 * lra-spills.c (assign_spill_hard_regs): Likewise.
12939 (spill_pseudos): Likewise.
12940 * tree-ssa-pre.c (bitmap_set_and): Likewise.
12941 (bitmap_set_subtract_values): Likewise.
12943 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12945 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
12946 management with auto_bitmap.
12947 (fix_inter_tick): Likewise.
12948 (fix_recovery_deps): Likewise.
12949 * ira.c (add_store_equivs): Likewise.
12950 (find_moveable_pseudos): Likewise.
12951 (split_live_ranges_for_shrink_wrap): Likewise.
12952 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
12953 (rtx_reuse_manager::seen_def_p): Likewise.
12954 (rtx_reuse_manager::set_seen_def): Likewise.
12955 * print-rtl.h (class rtx_reuse_manager): Likewise.
12957 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12959 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
12961 (migrate_btr_def): Likewise.
12962 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
12963 * df-core.c (loop_post_order_compute): Likewise.
12964 (loop_inverted_post_order_compute): Likewise.
12965 * hsa-common.h: Likewise.
12966 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
12967 * init-regs.c (initialize_uninitialized_regs): Likewise.
12968 * ipa-inline.c (resolve_noninline_speculation): Likewise.
12969 (inline_small_functions): Likewise.
12970 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
12971 * ira.c (combine_and_move_insns): Likewise.
12972 (build_insn_chain): Likewise.
12973 * loop-invariant.c (find_invariants): Likewise.
12974 * lower-subreg.c (propagate_pseudo_copies): Likewise.
12975 * predict.c (tree_predict_by_opcode): Likewise.
12976 (predict_paths_leading_to): Likewise.
12977 (predict_paths_leading_to_edge): Likewise.
12978 (estimate_loops_at_level): Likewise.
12979 (estimate_loops): Likewise.
12980 * shrink-wrap.c (try_shrink_wrapping): Likewise.
12981 (spread_components): Likewise.
12982 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
12983 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
12984 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
12985 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
12986 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
12987 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
12988 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
12989 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
12990 (mark_threaded_blocks): Likewise.
12991 (thread_through_all_blocks): Likewise.
12992 * tree-ssa.c (verify_ssa): Likewise.
12993 (execute_update_addresses_taken): Likewise.
12994 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
12996 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
12998 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
13000 (post_order_compute): Likewise.
13001 (inverted_post_order_compute): Likewise.
13002 (pre_and_rev_post_order_compute_fn): Likewise.
13004 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13006 * genrecog.c (int_set::int_set): Explicitly construct our
13007 auto_vec base class.
13008 * vec.h (auto_vec::auto_vec): New constructor.
13010 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13012 * bitmap.h (class auto_bitmap): New constructor taking
13013 bitmap_obstack * argument.
13015 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13017 * bitmap.h (class auto_bitmap): Change type of m_bits to
13018 bitmap_head, and adjust ctor / dtor and member operators.
13020 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
13022 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
13023 when returned register mode doesn't match original mode.
13025 2017-05-12 Jeff Law <law@redhat.com>
13026 Jakub Jelinek <jakub@redhat.com>
13028 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
13029 we look for cc setter after the compare-elim changes.
13030 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
13031 within the vector to match what compare-elim now expects.
13032 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
13033 (xorsi3_flags, one_cmplsi2_flags): Likewise.
13035 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
13036 after the compare-elim changes.
13037 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
13038 the vector to match what compare-elim now expects.
13039 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
13040 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
13041 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
13042 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
13043 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
13045 * config/visium/visium.c (single_set_and_flags): Fix where
13046 we look for cc setter after the compare-elim changes.
13047 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
13048 with the vector to match what compare-elim now expects.
13049 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
13050 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
13051 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
13052 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
13053 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
13054 (neg<mode>2_insn_set_overflow): Likewise.
13056 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
13058 PR middle-end/79794
13059 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
13060 maybe_expand_insn call, set ops[0].target. If still set after call,
13061 set alt_rtl. Add extra arg to recursive calls.
13062 (extract_bit_field): Add alt_rtl argument. Pass to
13064 * expmed.h (extract_bit_field): Fix prototype.
13065 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
13066 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
13067 to extract_bit_field_calls.
13068 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
13069 Pass alt_rtl to extract_bit_field calls.
13070 * calls.c (store_unaligned_arguments_into_psuedos)
13071 load_register_parameters): Pass extra NULL to extract_bit_field calls.
13072 * optabs.c (maybe_legitimize_operand): Clear op->target when call
13074 * optabs.h (struct expand_operand): Add target bitfield.
13076 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13078 * compare-elim.c (try_eliminate_compare): Canonicalize
13079 operation with embedded compare to
13080 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
13081 (set (reg) (operation)].
13083 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
13085 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13088 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
13089 cost of adding a carry flag for ADC instruction.
13090 [case MINUS]: Ignore the cost of subtracting a carry flag
13091 for SBB instruction.
13093 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
13095 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
13097 * config/rs6000/bmiintrin.h: New file.
13098 * config/rs6000/bmi2intrin.h: New file.
13099 * config/rs6000/x86intrin.h: New file.
13101 2017-05-12 Jeff Law <law@redhat.com>
13103 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
13106 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
13108 PR middle-end/80707
13109 * tree-cfg.c: Remove cfg edges of unreachable case statements.
13111 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13113 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
13114 early expansion of vector divide builtins.
13115 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
13116 builtins identified as having unsigned arguments.
13118 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13120 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
13121 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
13122 expansion of vector logical operations (and, andc, or, xor,
13125 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13127 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
13128 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
13130 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13132 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
13133 early GIMPLE expansion of vector multiplies.
13135 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13137 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
13138 TARGET_HAVE_MOVT conditional.
13139 (movt splitter): Likewise.
13141 2017-05-12 Richard Biener <rguenther@suse.de>
13143 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
13144 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13145 Fold all stmts not inplace.
13147 2017-05-12 Richard Biener <rguenther@suse.de>
13149 PR tree-optimization/80713
13150 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
13151 inserted_exprs bit for not removed stmts.
13153 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
13155 PR middle-end/69921
13156 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
13157 parallelized" attribute for parallelized OpenACC kernels.
13158 * omp-offload.c (execute_oacc_device_lower): Use it.
13160 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
13161 Set "oacc kernels" attribute.
13162 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
13163 parameter. Adjust all users.
13164 (oacc_fn_attrib_kernels_p): Remove function.
13165 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
13166 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
13167 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
13168 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
13169 assert "oacc kernels" attribute is set.
13171 2017-05-11 Carl Love <cel@us.ibm.com>
13173 * config/rs6000/rs6000-c: Add support for built-in functions
13174 vector unsigned char vec_popcnt (vector signed char)
13175 vector unsigned char vec_popcnt (vector unsigned char)
13176 vector unsigned short vec_popcnt (vector signed short)
13177 vector unsigned short vec_popcnt (vector unsigned short)
13178 vector unsigned int vec_popcnt (vector signed int)
13179 vector unsigned int vec_popcnt (vector unsigned int)
13180 vector unsigned long long vec_popcnt (vector signed long long)
13181 vector unsigned long long vec_popcnt (vector unsigned long long)
13182 vector signed long long vec_slo (vector signed long long,
13183 vector signed char)
13184 vector signed long long vec_slo (vector signed long long,
13185 vector unsigned char)
13186 vector unsigned long long vec_slo (vector unsigned long long,
13187 vector signed char)
13188 vector unsigned long long vec_slo (vector unsigned long long,
13189 vector unsigned char)
13190 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
13191 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
13192 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
13193 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
13194 * doc/extend.texi: Update the built-in documentation file for the
13195 new built-in functions.
13197 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
13199 * attribs.h (sorted_attr_string): Move machine independent
13200 functions for target clone support from the i386 port to common
13201 code. Rename ix86_function_versions to common_function_versions.
13202 Rename make_name to make_unique_name.
13203 (common_function_versions): Likewise.
13204 (make_unique_name): Likewise.
13205 (make_dispatcher_decl): Likewise.
13206 (is_function_default_version): Likewise.
13207 * attribs.c (attr_strcmp): Likewise.
13208 (sorted_attr_string): Likewise.
13209 (common_function_versions): Likewise.
13210 (make_unique_name): Likewise.
13211 (make_dispatcher_decl): Likewise.
13212 (is_function_default_version): Likewise.
13213 * config/i386/i386.c (attr_strcmp): Likewise.
13214 (sorted_attr_string): Likewise.
13215 (ix86_function_versions): Likewise.
13216 (make_name): Likewise.
13217 (make_dispatcher_decl): Likewise.
13218 (is_function_default_version): Likewise.
13219 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
13221 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13224 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
13225 Account for direct move costs for vec_construct of integer
13228 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
13231 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
13232 (UNSPEC_STX_ATOMIC): Ditto.
13233 (loaddi_via_sse): New insn.
13234 (storedi_via_sse): Ditto.
13235 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
13236 Update corresponding peephole2 patterns.
13237 (atomic_storedi_fpu): Ditto.
13239 2017-05-11 Julia Koval <julia.koval@intel.com>
13241 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
13242 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
13244 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
13245 (__builtin_ia32_rsqrt14ss_mask): New builtins.
13246 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
13248 2017-05-11 Nathan Sidwell <nathan@acm.org>
13250 * graphite-poly.c: Include dumpfile.h.
13252 * dumpfle.h (dump_function): Declare here ...
13253 * tree-dump.h (dump_function): ... not here.
13254 * dumpfile.c: #include tree-cfg.h.
13255 (dump_function): Move here from ...
13256 * tree-dump.c (dump_function): ... here.
13257 * gimplify.c: #include splay-tree.h, not tree-dump.h.
13258 * graphite-poly.c: Don't include tree-dump.h.
13259 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
13260 * print-tree.c: Likewise.
13261 * stor-layout.c: Likewise.
13262 * tree-nested.c: Likewise.
13264 * dumpfile.c (dump_start): Use TDF_FLAGS.
13265 (dump_enable_all): Fix TDF_KIND check thinko.
13267 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
13269 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13270 array entries to represent two legal parameterizations of the
13271 overloaded __builtin_cmpb function, as represented by the
13272 P6_OV_BUILTIN_CMPB constant.
13273 (altivec_resolve_overloaded_builtin): Add special case handling
13274 for the __builtin_cmpb function, as represented by the
13275 P6_OV_BUILTIN_CMPB constant.
13276 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
13277 (BU_P6_64BIT_2): New macro.
13278 (BU_P6_OVERLOAD_2): New macro
13279 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
13280 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
13281 (CMPB): Add overload support to represent both 32-bit and 64-bit
13282 compare-bytes function.
13283 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13284 support for TARGET_CMPB.
13285 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
13286 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13287 documentation of the __builtin_cmpb overloaded built-in function.
13289 2017-05-11 Richard Biener <rguenther@suse.de>
13291 PR tree-optimization/80705
13292 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
13293 bases are not vectorizable.
13295 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13297 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
13298 when counting register pressure.
13300 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13302 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
13303 (iv_ca_more_deps): Renamed to ...
13304 (iv_ca_compare_deps): ... this.
13305 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
13307 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13309 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
13311 (determine_group_iv_costs): ... here.
13312 (find_inv_vars_cb): Record inv var if it's not recorded before.
13314 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13316 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
13317 (get_shiftadd_cost): Ditto.
13319 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13321 * tree-ssa-address.c: Include header file.
13322 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
13324 (add_to_parts): Refactor.
13325 (addr_to_parts): New parameter. Update use of move_hint_to_base.
13326 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
13329 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13331 PR tree-optimization/53090
13332 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
13334 (extract_cond_operands): Detect condition with IV on both sides
13335 and return COMP_IV_EXPR_2.
13336 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
13337 (rewrite_use_compare): Simplify by removing call to function
13338 extract_cond_operands.
13340 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13342 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
13343 (extract_cond_operands): Detect condition comparing against non-
13344 invariant bound and return appropriate enum value.
13345 (find_interesting_uses_cond): Update use of extract_cond_operands.
13346 Handle its return value accordingly.
13347 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
13349 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13351 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
13352 nonlinear iv_use computation in loop invariant sensitive way.
13354 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13356 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
13357 (find_iv_candidates): Call relate_compare_use_with_all_cands.
13359 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13361 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
13362 (dump_cand): Support iv_cand.inv_exprs.
13363 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
13365 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
13368 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13370 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
13372 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
13373 as local function. Include necessary header files.
13374 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
13376 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13378 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
13380 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13382 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
13383 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
13384 RSHIFT_EXPR and BIT_NOT_EXPR.
13386 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13388 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
13389 (adjust_setup_cost): New parameter supporting round up adjustment.
13390 (struct address_cost_data): Delete.
13391 (force_expr_to_var_cost): Don't bound cost with spill_cost.
13392 (split_address_cost, ptr_difference_cost): Delete.
13393 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
13394 (struct ainc_cost_data): New struct.
13395 (get_address_cost_ainc): New function.
13396 (get_address_cost, get_computation_cost): Reimplement.
13397 (determine_group_iv_cost_address): Record inv_expr for all uses of
13399 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
13400 (iv_ca_has_deps): Reimplemented to ...
13401 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
13403 (iv_ca_extend): Call iv_ca_more_deps.
13405 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13407 * tree-ssa-address.c (struct mem_address): Move to header file.
13408 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
13409 * tree-ssa-address.h (struct mem_address): Move from C file.
13410 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
13412 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13414 * tree-affine.h (aff_combination_type): New interface.
13415 (aff_combination_zero_p): Remove static.
13416 (aff_combination_const_p): New interface.
13417 (aff_combination_singleton_var_p): New interfaces.
13419 2017-05-11 Richard Biener <rguenther@suse.de>
13421 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13422 Skip unreachable blocks and destinations.
13423 (eliminate): Move stmt removal and fixup ...
13424 (fini_eliminate): ... here. Skip inserted exprs.
13425 (pass_pre::execute): Move fini_pre after fini_eliminate.
13426 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
13427 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
13428 PRE to get rid of dead code that has invalid SSA form and
13429 split critical edges again.
13431 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13433 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
13435 2017-05-11 Richard Biener <rguenther@suse.de>
13437 * passes.c (execute_function_todo): Verify loops if they are
13438 said to be up-to-date.
13439 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
13440 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
13442 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
13445 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
13446 handle calling assemble_external ourself.
13449 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
13450 modes with zero size. Enhance comment.
13452 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13454 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
13455 built-ins for vec_xl and vec_xst with short and char pointer
13458 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13460 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
13461 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
13462 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
13463 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
13464 (_mm_maskz_min_round_ss): New intrinsics.
13465 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
13466 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
13467 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
13468 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
13469 (__builtin_ia32_minss_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<code><mode>3<round_saeonly_name>):
13474 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
13475 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
13477 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
13480 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13482 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
13483 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
13484 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
13485 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
13486 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
13487 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
13488 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
13489 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13490 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
13491 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
13492 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
13493 (__builtin_ia32_mulss_mask_round): New builtins.
13494 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13495 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13496 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
13498 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
13499 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
13501 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
13504 2017-05-10 Julia Koval <julia.koval@intel.com>
13506 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
13507 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
13508 (_mm256_setr_m128i): New intrinsics.
13510 2017-05-10 Julia Koval <julia.koval@intel.com>
13512 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
13513 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
13514 (_mm_maskz_rcp14_ss): New intrinsics.
13515 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
13516 (__builtin_ia32_rcp14ss_mask): New builtins.
13517 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
13519 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
13521 PR tree-optimization/51513
13522 * tree-cfg.c (gimple_seq_unreachable_p): New function.
13523 (assert_unreachable_fallthru_edge_p): Use it.
13524 (group_case_labels_stmt): Likewise.
13525 * tree-cfg.h: Prototype it.
13526 * stmt.c: Include cfghooks.h and tree-cfg.h.
13527 (emit_case_dispatch_table) <gap_label>: New local variable.
13528 Use it to fill dispatch table gaps.
13529 Test for default_label before updating probabilities.
13530 (expand_case) <default_label>: Remove unneeded initialization.
13531 Test for unreachable default case statement and remove its edge.
13532 Set default_label accordingly.
13533 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
13535 2017-05-10 Carl Love <cel@us.ibm.com>
13537 * config/rs6000/rs6000-c: Add support for built-in functions
13538 vector signed char vec_neg (vector signed char)
13539 vector signed short int vec_neg (vector short int)
13540 vector signed int vec_neg (vector signed int)
13541 vector signed long long vec_neg (vector signed long long)
13542 vector float vec_neg (vector float)
13543 vector double vec_neg (vector double)
13544 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
13546 * config/rs6000/altivec.h: Add define for vec_neg
13547 * doc/extend.texi: Update the built-in documentation for the
13548 new built-in functions.
13550 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13552 PR tree-optimization/77644
13553 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
13555 2017-05-10 Nathan Sidwell <nathan@acm.org>
13557 * dumpfile.h (TDI_lang_all): New.
13558 (TDF_KIND): New. Renumber others
13559 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
13561 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
13563 (get_dump_file_name): Adjust suffix generation.
13564 (dump_enable_all): Use TDF_KIND.
13565 * doc/invoke.texi (-fdump-lang-all): Document.
13567 * dumpfile.h: Tabify.
13569 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
13572 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
13573 Move member access before delete.
13575 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
13577 * tree-inline.c (expand_call_inline): Split block at stmt
13580 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
13583 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
13584 are now unused after splitting mov{sf,sd}_hardfloat.
13585 (f32_lr2): Likewise.
13586 (f32_lm): Likewise.
13587 (f32_lm2): Likewise.
13588 (f32_li): Likewise.
13589 (f32_li2): Likewise.
13590 (f32_lv): Likewise.
13591 (f32_sr): Likewise.
13592 (f32_sr2): Likewise.
13593 (f32_sm): Likewise.
13594 (f32_sm2): Likewise.
13595 (f32_si): Likewise.
13596 (f32_si2): Likewise.
13597 (f32_sv): Likewise.
13598 (f32_dm): Likewise.
13599 (f32_vsx): Likewise.
13600 (f32_av): Likewise.
13601 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
13602 For movsf, order stores so the VSX stores occur before the GPR
13603 store which encourages the register allocator to use a traditional
13604 FPR instead of a GPR. For movsd, order the stores so that the GPR
13605 store comes before the VSX stores to allow the power6 to work.
13606 This is due to the power6 not having a 32-bit integer store
13607 instruction from a FPR.
13608 (movsf_hardfloat): Likewise.
13609 (movsd_hardfloat): Likewise.
13611 2017-05-09 Martin Sebor <msebor@redhat.com>
13613 PR translation/80280
13614 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
13617 PR translation/80280
13618 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
13619 data member added in r247778.
13620 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
13622 2017-05-09 Nathan Sidwell <nathan@acm.org>
13624 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
13626 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
13629 2017-05-09 Marek Polacek <polacek@redhat.com>
13631 * doc/invoke.texi: Fix typo.
13633 2017-05-09 Richard Biener <rguenther@suse.de>
13635 * tree-vrp.c (vrp_val_is_max): Adjust comment.
13636 (vrp_val_is_min): Likewise.
13637 (set_value_range_to_value): Likewise.
13638 (set_value_range_to_nonnegative): Likewise.
13639 (gimple_assign_nonzero_p): Likewise.
13640 (gimple_stmt_nonzero_p): Likewise.
13641 (vrp_int_const_binop): Likewise. Remove unreachable case.
13642 (adjust_range_with_scev): Adjust comments.
13643 (compare_range_with_value): Likewise.
13644 (extract_range_from_phi_node): Likewise.
13645 (test_for_singularity): Likewise.
13647 2017-05-09 Richard Biener <rguenther@suse.de>
13649 * tree-vrp.c (get_single_symbol): Add assert that we don't
13650 get overflowed constants as invariant part.
13651 (compare_values_warnv): Add comment before the TREE_NO_WARNING
13652 checks. Use wi::cmp instead of recursing for integer constants.
13653 (compare_values): Just ignore whether we assumed undefined
13654 overflow instead of failing the compare.
13655 (extract_range_for_var_from_comparison_expr): Add comment before the
13656 TREE_NO_WARNING sets.
13657 (test_for_singularity): Likewise.
13658 (extract_range_from_comparison): Do not disable optimization
13659 when we assumed undefined overflow.
13660 (extract_range_basic): Remove init of unused var.
13662 2017-05-09 Richard Biener <rguenther@suse.de>
13664 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
13665 (extract_range_from_multiplicative_op_1): Adjust.
13666 (extract_range_from_binary_expr_1): Use int_const_binop.
13668 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13671 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
13672 rs6000_store_data_bypass_p in seven define_bypass directives and
13673 in several comments.
13674 * config/rs6000/rs6000-protos.h: Add prototype for
13675 rs6000_store_data_bypass_p function.
13676 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
13677 function implements slightly different (rs6000-specific) semantics
13678 than store_data_bypass_p, returning false rather than aborting
13679 with assertion error when arguments do not satisfy the
13680 requirements of store data bypass.
13681 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
13682 rs6000_store_data_bypass_p.
13684 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
13686 * config/xtensa/xtensa-protos.h
13687 (xtensa_initial_elimination_offset): New declaration.
13688 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
13689 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
13690 macro definition, add case for FRAME_POINTER_REGNUM when
13691 FRAME_GROWS_DOWNWARD.
13692 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
13693 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
13694 xtensa_initial_elimination_offset.
13696 2017-05-08 Nathan Sidwell <nathan@acm.org>
13698 * doc/invoke.texi: Alphabetize -fdump options.
13700 2017-05-08 Martin Sebor <msebor@redhat.com>
13702 PR translation/80280
13703 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
13705 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
13707 * target.def (compute_frame_layout): New optional target hook.
13708 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
13709 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
13710 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
13712 * reload1.c (verify_initial_elim_offsets): Likewise.
13713 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
13714 (use_simple_return_p): Call arm_compute_frame_layout if needed.
13715 (arm_get_frame_offsets): Split up into this ...
13716 (arm_compute_frame_layout): ... and this function.
13718 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
13720 * config/aarch64/constraints.md (Usa): New constraint.
13721 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
13723 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
13725 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
13726 with_multilib_list after it has been checked.
13728 2017-05-08 Richard Biener <rguenther@suse.de>
13730 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
13731 (bitmap_set_subtract_values): Likewise.
13733 2017-05-08 Richard Biener <rguenther@suse.de>
13735 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
13736 (gimple_assign_nonzero): ... this and remove strict_overflow_p
13738 (gimple_stmt_nonzero_warnv_p): Rename to ...
13739 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
13741 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
13742 (extract_range_basic): Adjust, do not disable propagation on
13743 strict overflow sensitive simplification.
13744 (vrp_visit_cond_stmt): Likewise.
13746 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
13748 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
13749 body size unconditionally.
13751 2017-05-07 Jeff Law <law@redhat.com>
13754 2017-05-06 Jeff Law <law@redhat.com>
13755 PR tree-optimization/78496
13756 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
13759 PR tree-optimization/78496
13760 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
13761 (simplify_stmt_using_ranges): Call it.
13762 (vrp_dom_walker::before_dom_children): Extract equivalences
13763 from an ASSERT_EXPR with an equality comparison against a
13766 2017-05-06 Jeff Law <law@redhat.com>
13768 PR tree-optimization/78496
13769 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
13772 PR tree-optimization/78496
13773 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
13774 (simplify_stmt_using_ranges): Call it.
13775 (vrp_dom_walker::before_dom_children): Extract equivalences
13776 from an ASSERT_EXPR with an equality comparison against a
13779 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
13781 * lra-constraints.c (lra_copy_reg_equiv): New function.
13782 (split_reg): Use it to copy equivalence information from the
13783 original register to the spill register.
13785 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
13787 PR rtl-optimization/75964
13788 * simplify-rtx.c (simplify_const_relational_operation): Remove
13789 invalid handling of comparisons of integer ABS.
13791 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
13793 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
13794 initialize to zero.
13795 (init_regs): Remove declaration.
13796 (function_arg_advance_32): Initialize error_p as boolean variable.
13798 2017-05-05 Nathan Sidwell <nathan@acm.org>
13800 * store-motion.c (remove_reachable_equiv_notes): Reformat long
13801 lines. Use for (;;).
13803 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13805 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
13806 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
13807 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
13808 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
13809 VF=2 that require versioning.
13811 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13813 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
13816 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13818 * diagnostic.h (diagnostic_override_option_index): Convert from
13819 macro to inline function.
13821 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13823 * diagnostic.c (last_module_changed_p): New function.
13824 (set_last_module): New function.
13825 (diagnostic_report_current_module): Convert macro usage to
13826 the above functions.
13827 * diagnostic.h (diagnostic_context::last_module): Strengthen
13828 from const line_map * to const line_map_ordinary *.
13829 (diagnostic_last_module_changed): Delete macro.
13830 (diagnostic_set_last_module): Delete macro.
13832 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13834 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
13835 with diagnostic_report_diagnostic.
13836 (diagnostic_n_impl_richloc): Likewise.
13837 * diagnostic.h (report_diagnostic): Delete macro.
13838 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
13839 with diagnostic_report_diagnostic.
13840 * substring-locations.c (format_warning_va): Likewise.
13842 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13844 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
13845 save/restor of format_spec. Move option-printing code to...
13846 (print_option_information): ...this new function, and
13847 reimplement by simply printing to the pretty_printer,
13848 rather than appending to the format string.
13850 2017-05-05 David Malcolm <dmalcolm@redhat.com>
13852 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
13853 handling logic into...
13854 (update_effective_level_from_pragmas): ...this new function.
13856 2017-05-04 Andrew Waterman <andrew@sifive.com>
13858 * config/riscv/riscv.opt (mstrict-align): New option.
13859 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
13860 (SLOW_UNALIGNED_ACCESS): Define.
13861 (riscv_slow_unaligned_access): Declare.
13862 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
13864 (riscv_slow_unaligned_access): New variable.
13865 (rocket_tune_info): Set slow_unaligned_access to true.
13866 (optimize_size_tune_info): Set slow_unaligned_access to false.
13867 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
13868 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
13869 (riscv_option_override): Set riscv_slow_unaligned_access.
13870 * doc/invoke.texi: Add -mstrict-align to RISC-V.
13872 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
13874 * config/riscv/riscv.md: Unify indentation.
13876 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
13881 * config/rs6000/rs6000.md (u code attribute): Add FIX and
13883 (extendsi<mode>2): Add support for doing sign extension via
13884 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
13885 don't have ISA 3.0 instructions.
13886 (extendsi<mode>2 splitter): Likewise.
13887 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
13888 generate the normal insns since SImode can now go in vector
13889 registers. Disallow the special UNSPECs needed for previous
13890 machines to hide SImode being used. Add new insns
13891 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
13892 (fix_trunc<mode>si2_stfiwx): Likewise.
13893 (fix_trunc<mode>si2_internal): Likewise.
13894 (fixuns_trunc<mode>si2): Likewise.
13895 (fixuns_trunc<mode>si2_stfiwx): Likewise.
13896 (fctiw<u>z_<mode>_smallint): Likewise.
13897 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
13898 of floating point to 32-bit integer from doing a direct move to
13899 the GPR registers to do a store.
13900 (fctiwz_<mode>): Break long line.
13902 2017-05-05 Bin Cheng <bin.cheng@arm.com>
13904 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
13905 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
13906 (addr_list, addr_offset_valid_p): New.
13907 (split_address_groups): Check offset validity with above function.
13908 (gt-tree-ssa-loop-ivopts.h): Include header file.
13910 2017-05-05 Nathan Sidwell <nathan@acm.org>
13912 * config.gcc (arm*-*-*): Add missing 'fi'.
13914 2017-05-05 Steve Ellcey <sellcey@cavium.com>
13916 * doc/invoke.texi (-fopt-info): Explicitly say order of options
13917 included in -fopt-info does not matter.
13918 * doc/optinfo.texi (-fopt-info): Fix description of default
13919 behavour. Explicitly say order of options included in -fopt-info
13922 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
13924 * config.gcc: Allow combinations of aprofile and rmprofile values for
13925 --with-multilib-list.
13926 * config/arm/t-multilib: New file.
13927 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
13928 variables. Remove setting of ISA and floating-point ABI in
13929 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
13930 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
13931 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
13932 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
13934 * config/arm/t-rmprofile: Likewise except for the matches changes.
13935 * doc/install.texi (--with-multilib-list): Document the combination of
13936 aprofile and rmprofile values and warn about pitfalls in doing that.
13938 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
13940 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
13941 (movdi_aarch64): Likewise.
13943 2017-05-05 Jakub Jelinek <jakub@redhat.com>
13945 PR tree-optimization/80632
13946 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
13948 (build_arrays): Initialize it for virtual phis.
13949 (fix_phi_nodes): Use it for virtual phis.
13951 PR tree-optimization/80558
13952 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
13953 [x, y] op z into [x op, y op z] for op & or | if conditions
13956 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
13957 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13960 * config/arm/arm.md (use_literal_pool): Remove.
13961 (64-bit immediate split): No longer takes cost into consideration
13962 if arm_disable_literal_pool is enabled.
13963 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
13964 used when arm_disable_literal_pool is enabled.
13965 (arm_max_const_double_inline_cost): Remove use of
13966 arm_disable_literal_pool.
13967 (push_minipool_fix): Add assert.
13968 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
13969 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
13970 (no_literal_pool_sf_immediate): New.
13972 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13974 PR tree-optimization/80613
13975 * tree-ssa-dce.c (propagate_necessity): Remove cases for
13976 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
13978 2017-05-05 Richard Biener <rguenther@suse.de>
13980 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
13982 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
13984 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
13985 of this flag from insn conditions due to removal from r247495.
13987 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
13989 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
13991 (arm_early_store_addr_dep_ptr): Likewise.
13992 * config/arm/aarch-common-protos.h
13993 (arm_early_load_addr_dep_ptr): Add prototype.
13994 (arm_early_store_addr_dep_ptr): Likewise.
13995 * config/arm/cortex-a53.md: Add new bypasses.
13997 2017-05-05 Jakub Jelinek <jakub@redhat.com>
13999 * tree.c (next_type_uid): Change type to unsigned.
14000 (type_hash_canon): Decrement back next_type_uid if
14001 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
14002 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
14005 2017-05-04 Martin Sebor <msebor@redhat.com>
14007 * builtins.c: Fix a trivial typo in a comment.
14009 PR middle-end/79234
14010 * builtins.c (check_sizes): Adjust to handle reading past the end.
14011 Avoid printing excessive upper bound of ranges. Use %E to print
14012 tree nodes instead of converting them to %wu.
14013 (expand_builtin_memchr): New function.
14014 (compute_dest_size): Rename...
14015 (compute_objsize): ...to this.
14016 (expand_builtin_memcpy): Adjust.
14017 (expand_builtin_mempcpy): Adjust.
14018 (expand_builtin_strcat): Adjust.
14019 (expand_builtin_strcpy): Adjust.
14020 (check_strncat_sizes): Adjust.
14021 (expand_builtin_strncat): Adjust.
14022 (expand_builtin_strncpy): Adjust and simplify.
14023 (expand_builtin_memset): Adjust.
14024 (expand_builtin_bzero): Adjust.
14025 (expand_builtin_memcmp): Adjust.
14026 (expand_builtin): Handle memcmp.
14027 (maybe_emit_chk_warning): Check strncat just once.
14029 2017-05-04 Martin Sebor <msebor@redhat.com>
14031 PR preprocessor/79214
14032 PR middle-end/79222
14033 PR middle-end/79223
14034 * builtins.c (check_sizes): Add inlining context and issue
14035 warnings even when -Wno-system-headers is set.
14036 (check_strncat_sizes): Same.
14037 (expand_builtin_strncat): Same.
14038 (expand_builtin_memmove): New function.
14039 (expand_builtin_stpncpy): Same.
14040 (expand_builtin): Handle memmove and stpncpy.
14042 2017-05-04 Bin Cheng <bin.cheng@arm.com>
14044 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
14045 which is not used any more.
14047 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14049 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
14051 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14053 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
14054 (cortexa53_tunings): Likewise.
14055 (cortexa57_tunings): Likewise.
14056 (cortexa72_tunings): Likewise.
14057 (cortexa73_tunings): Likewise.
14059 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14061 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
14062 Set loop alignment to 8.
14064 2017-05-04 Martin Sebor <msebor@redhat.com>
14066 PR translation/80280
14067 * builtins.c (expand_builtin_object_size): Add missing quoting to
14068 %D and like directives.
14069 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
14070 (hsa_type_for_tree_type): Same.
14071 (verify_function_arguments): Same.
14072 * symtab.c (symbol_table::change_decl_assembler_name): Same.
14073 * varasm.c (get_section): Same.
14076 2017-05-04 Martin Sebor <msebor@redhat.com>
14078 PR translation/80280
14079 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
14081 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14083 * config/aarch64/aarch64.c (generic_addrcost_table):
14084 Change HI/TI mode setting.
14086 2017-05-04 Martin Jambor <mjambor@suse.cz>
14088 PR tree-optimization/80622
14089 * tree-sra.c (comes_initialized_p): New function.
14090 (build_accesses_from_assign): Only set write lazily when
14091 comes_initialized_p is false.
14092 (analyze_access_subtree): Use comes_initialized_p.
14093 (propagate_subaccesses_across_link): Assert !comes_initialized_p
14094 instead of testing for PARM_DECL.
14096 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14098 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
14099 constraint on operand 0 to allow more general addressing modes.
14100 Adjust output template.
14101 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
14103 * config/aarch64/aarch64-protos.h
14104 (aarch64_address_valid_for_prefetch_p): Declare prototype.
14105 * config/aarch64/constraints.md (Dp): New address constraint.
14106 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
14109 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
14111 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
14112 update use of estimate_ipcp_clone_size_and_time.
14113 (estimate_local_effects): Update use of
14114 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
14115 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
14116 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
14117 Return nonspecialized time.
14119 2017-05-04 Richard Biener <rguenther@suse.de>
14121 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
14122 for the last VUSE which def dominates the PHI. Directly call
14124 (get_continuation_for_phi_1): Remove.
14126 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
14128 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
14129 to explain the use of truncating division. Cap the number of
14130 iterations to the maximum given by nb_iterations_upper_bound,
14133 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14135 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
14136 * configure: Regenerate.
14137 * config.in: Regenerate.
14138 * config/i386/driver-mingw32.c: new file.
14139 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
14140 * config.host: Link driver-mingw32.o on MinGW host.
14141 * doc/install.texi: Document new --enable-mingw-wildcard configure
14144 2017-05-04 Marek Polacek <polacek@redhat.com>
14146 PR tree-optimization/80612
14147 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
14149 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14150 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
14152 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
14153 (movt splitter): Likewise.
14154 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
14155 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
14156 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
14157 block for Thumb-1 with MOVT.
14158 (thumb2_legitimate_address_p): Move code block ...
14159 (can_avoid_literal_pool_for_label_p): ... into this new function.
14160 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
14162 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
14163 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
14164 "M-profile targets with the MOVT instruction".
14166 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14168 * config/arm/arm-builtins.c (arm_init_builtins): Rename
14169 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
14170 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
14172 2017-05-04 Martin Liska <mliska@suse.cz>
14174 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
14175 variable cond_code.
14177 2017-05-04 Richard Biener <rguenther@suse.de>
14179 * tree.c (array_at_struct_end_p): Handle arrays at struct
14180 end with flexarrays more conservatively. Refactor and treat
14181 arrays of arrays or aggregates more strict. Fix
14182 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
14183 * tree.c (array_at_struct_end_p): Adjust prototype.
14184 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
14185 * gimple-fold.c (get_range_strlen): Likewise.
14186 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
14188 2017-05-04 Richard Biener <rguenther@suse.de>
14190 PR tree-optimization/31130
14191 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
14193 (supports_overflow_infinity): Likewise.
14194 (is_negative_overflow_infinity): Likewise.
14195 (is_positive_overflow_infinity): Likewise.
14196 (is_overflow_infinity): Likewise.
14197 (stmt_overflow_infinity): Likewise.
14198 (overflow_infinity_range_p): Likewise.
14199 (usable_range_p): Remove as always returning true.
14200 (make_overflow_infinity): Remove.
14201 (negative_overflow_infinity): Likewise.
14202 (positive_overflow_infinity): Likewise.
14203 (avoid_overflow_infinity): Likewise.
14204 (set_value_range): Adjust accordingly.
14205 (set_value_range_to_nonnegative): Likewise, remove now unused
14206 overflow_infinity arg.
14207 (vrp_operand_equal_p): Adjust.
14208 (update_value_range): Likewise.
14209 (range_int_cst_singleton_p): Likewise.
14210 (operand_less_p): Likewise.
14211 (compare_values_warnv): Likewise.
14212 (extract_range_for_var_from_comparison_expr): Likewise.
14213 (vrp_int_const_binop): Likewise.
14214 (zero_nonzero_bits_from_vr): Likewise.
14215 (extract_range_from_multiplicative_op_1): Likewise.
14216 (extract_range_from_binary_expr_1): Likewise.
14217 (extract_range_from_unary_expr): Likewise.
14218 (extract_range_from_comparison): Likewise.
14219 (extract_range_basic): Likewise.
14220 (adjust_range_with_scev): Likewise.
14221 (compare_ranges): Likewise.
14222 (compare_range_with_value): Likewise.
14223 (dump_value_range): Likewise.
14224 (test_for_singularity): Likewise, remove strict_overflow_p parameter
14226 (simplify_cond_using_ranges): Adjust.
14228 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
14230 * brig-builtins.def: Added a builtin for class_f64.
14231 * builtin-types.def: Added a builtin type needed by class_f64.
14233 2017-05-03 Jason Merrill <jason@redhat.com>
14235 * timevar.def: Add TV_CONSTEXPR.
14237 2017-05-03 David Malcolm <dmalcolm@redhat.com>
14239 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
14241 2017-05-03 Martin Jambor <mjambor@suse.cz>
14243 * ipa-prop.c (ipa_update_after_lto_read): Removed.
14244 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
14245 * ipa-cp.c (ipcp_propagate_stage): Do not call
14246 ipa_update_after_lto_read.
14247 * ipa-inline.c (ipa_inline): Likewise.
14249 2017-05-03 Martin Jambor <mjambor@suse.cz>
14251 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
14252 tag. Added a default constructor and a destructor.
14253 (ipa_edge_args_sum_t): New class;
14254 (ipa_edge_args_sum): Declare.
14255 (ipa_edge_args_vector): Remove declaration.
14256 (IPA_EDGE_REF): Use ipa_edge_args_sum.
14257 (ipa_free_edge_args_substructures): Remove declaration.
14258 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
14259 (ipa_edge_args_info_available_for_edge_p): Likewise.
14260 * ipa-prop.c (ipa_edge_args_vector): Removed.
14261 (edge_removal_hook_holder): Likewise.
14262 (edge_duplication_hook_holder): Likewise.
14263 (ipa_edge_args_sum): New variable.
14264 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
14265 ipa_edge_args_vector.
14266 (ipa_free_edge_args_substructures): Likewise.
14267 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
14268 ipa_edge_args_vector.
14269 (ipa_edge_removal_hook): Turned into method
14270 ipa_edge_args_sum_t::remove.
14271 (ipa_edge_duplication_hook): Turned into method
14272 ipa_edge_args_sum_t::duplicate.
14273 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
14274 registering edge hooks.
14275 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
14276 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
14277 ipa_edge_args_sum instead of ipa_edge_args_vector.
14278 * ipa-profile.c (ipa_profile): Likewise.
14280 2017-05-03 Martin Jambor <mjambor@suse.cz>
14282 * symbol-summary.h (function_summary): New method exists.
14283 (function_summary::symtab_removal): Deallocate through release.
14284 (call_summary): New class.
14285 (gt_ggc_mx): New overload.
14286 (gt_pch_nx): Likewise.
14287 (gt_pch_nx): Likewise.
14289 2017-05-03 Jeff Law <law@redhat.com>
14291 PR tree-optimization/78496
14292 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
14293 from simplify_cond_using_ranges. Split off code to walk
14294 backwards through casts into ...
14295 (simplify_cond_using_ranges_2): New function.
14296 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
14297 (execute_vrp): After identifying jump threads, call
14298 simplify_cond_using_ranges_2.
14300 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
14303 * ipa-inline.h (inline_summary): Add ctor.
14304 (create_ggc): Do not use ggc_cleared_alloc.
14306 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
14307 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14309 * gcc.c (handle_braces): Support escaping in switch matching
14311 * doc/invoke.texi (Spec Files): Document it.
14312 Remove superfluous @code markup in items.
14314 2017-05-03 David Malcolm <dmalcolm@redhat.com>
14316 * diagnostic-show-locus.c (struct column_range): New struct.
14317 (get_affected_columns): New function.
14318 (get_printed_columns): New function.
14319 (struct correction): New struct.
14320 (correction::ensure_capacity): New function.
14321 (correction::ensure_terminated): New function.
14322 (struct line_corrections): New struct.
14323 (line_corrections::~line_corrections): New dtor.
14324 (line_corrections::add_hint): New function.
14325 (layout::print_trailing_fixits): Reimplement in terms of the new
14327 (selftest::test_overlapped_fixit_printing): New function.
14328 (selftest::diagnostic_show_locus_c_tests): Call it.
14330 2017-05-03 Nathan Sidwell <nathan@acm.org>
14332 Canonicalize canonical type hashing
14333 * tree.h (type_hash_canon_hash): Declare.
14334 * tree.c (type_hash_list, attribute_hash_list): Move into
14335 type_hash_canon_hash.
14336 (build_type_attribute_qual_variant): Break out hash code calc into
14337 type_hash_canon_hash.
14338 (type_hash_canon_hash): New. Generic type hash computation.
14339 (build_range_type_1, build_array_type_1, build_function_type,
14340 build_method_type_directly, build_offset_type, build_complex_type,
14341 make_vector_type): Call it.
14343 2017-05-03 Richard Biener <rguenther@suse.de>
14345 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
14346 When all DRs have unknown misaligned do not always peel
14347 when there is a store but apply the same costing model as if
14348 there were only loads.
14350 2017-05-03 Richard Biener <rguenther@suse.de>
14353 PR tree-optimization/80492
14354 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
14355 compare_base_decls returning dont-know properly.
14357 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
14359 * config/arm/iterators.md (CCSI): New mode iterator.
14360 (arch): New mode attribute.
14361 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
14362 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
14363 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
14364 code iterator for success result mode.
14365 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
14366 the corresponding new insn generators.
14368 2017-05-03 Bin Cheng <bin.cheng@arm.com>
14371 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14372 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14374 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
14376 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
14377 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
14378 (DDR_A): Wrap DDR argument in brackets.
14379 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
14380 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
14381 (DDR_REVERSED_P): Likewise.
14383 2017-05-03 Jakub Jelinek <jakub@redhat.com>
14385 PR tree-optimization/79472
14386 * tree-switch-conversion.c (struct switch_conv_info): Add
14387 contiguous_range and default_case_nonstandard fields.
14388 (collect_switch_conv_info): Compute contiguous_range and
14389 default_case_nonstandard fields, don't clear final_bb if
14390 contiguous_range and only the default case doesn't have the required
14392 (check_all_empty_except_final): Set default_case_nonstandard instead
14393 of failing if contiguous_range and the default case doesn't have empty
14395 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
14396 and only the default case doesn't have the required constants. Skip
14398 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
14399 if default_case_nonstandard.
14400 (build_constructors): Build constant 1 just once. Assert that default
14401 values aren't inserted in between cases if contiguous_range. Skip
14403 (build_arrays): Skip virtual phis.
14404 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
14405 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
14406 Handle virtual phis.
14407 (gen_inbound_check): Handle default_case_nonstandard case.
14408 (process_switch): Adjust check_final_bb caller. Call
14409 gather_default_values with the first non-default case instead of
14410 default case if default_case_nonstandard.
14412 2017-05-02 Nathan Sidwell <nathan@acm.org>
14414 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
14415 check. Fix formatting.
14417 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
14419 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
14420 errors when comparing specialized and unspecialized times.
14422 2017-05-02 David Malcolm <dmalcolm@redhat.com>
14424 * diagnostic-show-locus.c
14425 (layout::should_print_annotation_line_p): Make private.
14426 (layout::print_annotation_line): Make private.
14427 (layout::annotation_line_showed_range_p): Make private.
14428 (layout::show_ruler): Make private.
14429 (layout::print_source_line): Make private. Pass in line and
14430 line_width, rather than calling location_get_source_line. Drop
14432 (layout::print_leading_fixits): New method.
14433 (layout::print_any_fixits): Rename to...
14434 (layout::print_trailing_fixits): ...this, and make private.
14435 Don't print newline fixits.
14436 (diagnostic_show_locus): Move logic for printing one row into...
14437 (layout::print_line): ...this new function. Move the
14438 location_get_source_line call and error-handling from
14439 print_source_line to here. Call print_leading_fixits, and rename
14440 print_any_fixits to print_trailing_fixits.
14441 (selftest::test_fixit_insert_containing_newline): Update now that
14442 newlines are partially supported.
14443 (selftest::test_fixit_insert_containing_newline_2): New test.
14444 (selftest::test_fixit_replace_containing_newline): Update comments.
14445 (selftest::diagnostic_show_locus_c_tests): Call the new test.
14446 * edit-context.c (class added_line): New class.
14447 (class edited_line): Describe newline handling in comment.
14448 (edited_line::actually_edited_p): New method.
14449 (edited_line::print_content): Delete redundant decl.
14450 (edited_line::m_predecessors): New field.
14451 (edited_file::print_content): Call edited_line::print_content.
14452 (edited_file::print_diff): Update to support newlines.
14453 (edited_file::print_diff_hunk): Likewise.
14454 (edited_file::print_run_of_changed_lines): New function.
14455 (edited_file::print_diff_line): Convert to...
14456 (print_diff_line): ...this.
14457 (edited_file::get_effective_line_count): New function.
14458 (edited_line::edited_line): Initialize new field m_predecessors.
14459 (edited_line::~edited_line): Clean up m_predecessors.
14460 (edited_line::apply_fixit): Handle newlines.
14461 (edited_line::get_effective_line_count): New function.
14462 (edited_line::print_content): New function.
14463 (edited_line::print_diff_lines): New function.
14464 (selftest::test_applying_fixits_insert_containing_newline): New
14466 (selftest::test_applying_fixits_replace_containing_newline): New
14468 (selftest::insert_line): New function.
14469 (selftest::test_applying_fixits_multiple_lines): Add example of
14471 (selftest::edit_context_c_tests): Call the new tests.
14473 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14475 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
14476 parameter cand. Update dump information.
14477 (get_computation_cost): Update uses.
14479 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14481 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
14482 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
14483 (get_computation_at, rewrite_use_address): Update use of
14484 get_computation_aff.
14486 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14488 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
14489 (get_computation): Delete.
14490 (get_computation_cost): Implement like get_computation_cost_at.
14491 Use get_computation_at.
14492 (get_computation_cost_at): Delete.
14493 (rewrite_use_nonlinear_expr): Use get_computation_at.
14494 (rewrite_use_compare, remove_unused_ivs): Ditto.
14496 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14498 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
14500 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14502 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
14503 (ivopts_global_cost_for_size): Rename parameter and update uses.
14504 (iv_ca_recount_cost): Update uses.
14505 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
14506 candidates seperately in n_invs and n_cands.
14507 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
14509 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14511 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
14512 (find_inv_vars_cb): New.
14513 (find_depends): Renamed to ...
14514 (find_inv_vars): ... this.
14515 (add_candidate_1, force_var_cost): Call find_inv_vars.
14516 (split_address_cost, determine_group_iv_cost_cond): Ditto.
14518 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14520 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
14521 inv_vars. Add inv_exprs.
14522 (struct iv_cand): Rename depends_on to inv_vars.
14523 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
14524 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
14525 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
14526 (dump_cand): Dump inv_vars.
14527 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
14528 (record_invariant, find_depends, add_candidate_1): Ditto.
14529 (set_group_iv_cost, force_var_cost): Ditto.
14530 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
14531 (get_computation_cost_at, get_computation_cost): Ditto.
14532 (determine_group_iv_cost_generic): Ditto.
14533 (determine_group_iv_cost_address): Ditto.
14534 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
14535 (determine_group_iv_costs): Ditto.
14536 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
14537 (iv_ca_set_remove_invariants): Renamed to ...
14538 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
14539 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
14540 (iv_ca_set_add_invariants): Renamed to ...
14541 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
14542 (iv_ca_set_cp): Use iv_ca_set_add_invs.
14543 (iv_ca_has_deps): Support inv_vars and inv_exprs.
14544 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
14545 (create_new_ivs): Remove useless dump.
14547 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14549 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
14551 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
14552 (iv_ca_set_no_cp, create_new_iv): Ditto.
14554 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14556 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14558 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
14560 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
14561 function tree_check2.
14563 2017-05-02 Martin Liska <mliska@suse.cz>
14565 * doc/gcov.texi: Add missing preposition.
14566 * gcov.c (function_info::function_info): Properly fill up
14567 all member variables.
14569 2017-05-02 Tamar Christina <tamar.christina@arm.com>
14571 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
14573 2017-05-02 Tamar Christina <tamar.christina@arm.com>
14575 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
14577 2017-05-02 Martin Liska <mliska@suse.cz>
14580 * lto-streamer-in.c (lto_read_tree_1): Remove
14581 LTO_STREAMER_DEBUG.
14582 * lto-streamer.c (struct tree_hash_entry): Likewise.
14583 (struct tree_entry_hasher): Likewise.
14584 (tree_entry_hasher::hash): Likewise.
14585 (tree_entry_hasher::equal): Likewise.
14586 (lto_streamer_init): Likewise.
14587 (lto_orig_address_map): Likewise.
14588 (lto_orig_address_get): Likewise.
14589 (lto_orig_address_remove): Likewise.
14590 * lto-streamer.h: Likewise.
14591 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
14592 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
14594 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
14596 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
14597 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
14598 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
14599 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
14600 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
14601 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
14602 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
14603 (mm_maskz_sub_ss): New intrinsics.
14604 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14605 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
14606 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
14607 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
14608 (__builtin_ia32_subss_mask_round): New builtins.
14609 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
14610 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
14611 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
14613 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
14614 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
14616 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
14619 2017-05-02 Martin Jambor <mjambor@suse.cz>
14621 PR tree-optimization/78687
14622 * tree-sra.c (access): New field parent.
14623 (process_subtree_disqualification): New function.
14624 (disqualify_candidate): Call it.
14625 (build_accesses_from_assign): Reset write flag if creating an
14627 (build_access_subtree): Fill in parent field and also prpagate
14628 down grp_write flag.
14629 (create_artificial_child_access): New parameter set_grp_write, set
14630 grp_write to its value.
14631 (propagate_subaccesses_across_link): Also propagate grp_write flag
14633 (propagate_all_subaccesses): Push the closest parent back to work
14634 queue if add_access_to_work_queue returned true.
14636 2017-05-02 Richard Biener <rguenther@suse.de>
14638 * common.opt (fstrict-overflow): Alias negative to fwrapv.
14639 * doc/invoke.texi (fstrict-overflow): Remove all traces of
14640 -fstrict-overflow documentation.
14641 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
14642 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
14643 flag_strict_overflow.
14644 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
14645 * lto-opts.c (lto_write_options): Do not stream it.
14646 * lto-wrapper.c (merge_and_complain): Do not handle it.
14647 * opts.c (default_options_table): Do not set -fstrict-overflow.
14648 (finish_options): Likewise do not clear it when sanitizing.
14649 * simplify-rtx.c (simplify_const_relational_operation): Do not
14650 test flag_strict_overflow.
14652 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
14654 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
14655 using enabled attribute.
14656 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
14657 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
14658 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
14659 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
14660 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
14661 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
14662 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
14663 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
14664 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
14665 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
14667 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
14669 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
14671 2017-05-02 Richard Biener <rguenther@suse.de>
14673 PR tree-optimization/80591
14675 2017-04-10 Richard Biener <rguenther@suse.de>
14677 * tree-ssa-structalias.c (find_func_aliases): Properly handle
14680 2017-05-02 Richard Biener <rguenther@suse.de>
14682 PR tree-optimization/80549
14683 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
14684 (cleanup_tree_cfg_noloop): Create forwarders to known loop
14685 headers if they do not have a preheader.
14687 2017-05-02 Martin Liska <mliska@suse.cz>
14690 * common.opt: Fix typo.
14691 * doc/invoke.texi: Likewise.
14693 2017-05-01 Jan Beulich <jbeulich@suse.com>
14695 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
14696 swapping, add (x,x,m,x,n) alternative.
14698 2017-05-01 Nathan Sidwell <nathan@acm.org>
14700 * calls.c (combine_pending_stack_adjustment_and_call): Remove
14701 unnecessary unadjusted_alignment check.
14703 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
14706 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
14708 * gimplify.c (gimplify_cilk_detach): New function.
14709 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
14710 * tree-core.h: Document EXPR_CILK_SPAWN.
14711 * tree.h (EXPR_CILK_SPAWN): Define.
14713 2017-05-01 David Malcolm <dmalcolm@redhat.com>
14715 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
14716 to use new fixit_hint representation, using the "replace" logic.
14717 (get_line_span_for_fixit_hint): Likewise.
14718 (layout::print_any_fixits): Likewise.
14719 (selftest::test_one_liner_many_fixits): Rename to...
14720 (selftest::test_one_liner_many_fixits_1): ...this, and update
14721 comment and expected output to reflect that the multiple fix-it
14722 hints are now consolidated into one insertion.
14723 (selftest::test_one_liner_many_fixits_2): New test.
14724 (selftest::test_diagnostic_show_locus_one_liner): Update for
14726 (selftest::test_fixit_consolidation): Update for fix-it API
14728 * diagnostic.c (print_parseable_fixits): Likewise.
14729 * edit-context.c (edited_line::m_line_events): Convert from
14730 auto_vec <line_event *> to auto_vec <line_event>.
14731 (class line_event): Convert from abstract base class to a concrete
14732 class, taking over the role of replace_event.
14733 (class insert_event): Delete.
14734 (class replace_event): Rename to class line_event. Convert to
14736 (edit_context::add_fixits): Reimplement.
14737 (edit_context::apply_insert): Delete.
14738 (edit_context::apply_replace): Rename to...
14739 (edit_context::apply_fixit): ...this. Convert to half-open range.
14740 (edited_file::apply_insert): Delete.
14741 (edited_file::apply_replace): Rename to...
14742 (edited_file::apply_fixit): ...this.
14743 (edited_line::~edited_line): Drop deletion of events.
14744 (edited_line::apply_insert): Delete.
14745 (edited_line::apply_replace): Rename to...
14746 (edited_line::apply_fixit): ...this. Convert to half-open range.
14747 Update for change to type of m_line_events.
14748 * edit-context.h (edit_context::apply_insert): Delete.
14749 (edit_context::apply_replace): Rename to...
14750 (edit_context::apply_fixit): ...this.
14752 2017-05-01 Martin Sebor <msebor@redhat.com>
14754 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
14757 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
14760 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
14761 __get_cpuid_max returns 0.
14762 (__get_cpuid_count): Ditto.
14764 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
14766 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
14767 replacement expression is another instance of one of its arguments.
14769 2017-05-01 Jakub Jelinek <jakub@redhat.com>
14772 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
14773 check for stack push/pop autoinc.
14774 * config/i386/i386.c (ix86_agi_dependent): Return false
14775 if the only reason why modified_in_p returned true is that
14776 addr is SP based and set_insn is a push or pop.
14778 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
14780 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
14783 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
14786 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
14787 (account_size_time): Use two predicates - exec_pred and
14789 (evaluate_conditions_for_known_args): Compute both clause and
14791 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
14792 (inline_summary_t::duplicate): Update.
14793 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
14795 (compute_inline_parameters): Likewise.
14796 (estimate_edge_size_and_time): Update caluclation of time.
14797 (estimate_node_size_and_time): Compute both time and nonspecialized
14799 (estimate_ipcp_clone_size_and_time): Update.
14800 (inline_merge_summary): Update.
14801 (do_estimate_edge_time): Update.
14802 (do_estimate_edge_size): Update.
14803 (do_estimate_edge_hints): Update.
14804 (inline_read_section, inline_write_summary): Stream both new predicates.
14805 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
14807 (compute_inlined_call_time): Cleanup.
14808 (big_speedup_p): Update.
14809 (edge_badness): Update.
14810 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
14811 (size_time_entry): Replace predicate by exec_predicate and
14812 nonconst_predicate.
14813 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
14814 (estimate_edge_time): Return also nonspec_time.
14815 (reset_edge_growth_cache): Update.
14817 2017-04-29 Jakub Jelinek <jakub@redhat.com>
14819 PR rtl-optimization/80491
14820 * ifcvt.c (noce_process_if_block): When looking for x setter
14821 with missing else_bb, don't check only the insn right before
14822 cond_earliest, but look for the last insn that x is modified in
14823 within the same bb.
14825 PR rtl-optimization/80491
14826 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
14828 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
14830 PR tree-optimization/80487
14831 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
14833 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14835 PR tree-optimization/79697
14836 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
14837 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
14838 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
14840 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
14841 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
14843 2017-04-28 Martin Sebor <msebor@redhat.com>
14845 PR tree-optimization/80523
14846 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
14847 (init_target_to_host_charmap, target_to_host, target_strtol10): New
14849 (maybe_warn, format_directive, parse_directive): Use new functions.
14850 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
14852 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
14854 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
14856 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
14858 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
14859 target_header_dir): Set correctly.
14860 * configure: Regenerated.
14861 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
14862 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
14863 instead of SYSTEM_HEADER_DIR.
14865 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
14867 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
14868 (estimate_local_effects): Likewise.
14869 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
14870 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
14871 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
14872 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
14873 do_estimate_edge_time, estimate_edge_time): Likewise.
14874 * ipa-inline-analysis.c (estimate_node_size_and_time,
14875 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
14876 (estimate_time_after_inlining): Remove.
14878 2017-04-28 Martin Liska <mliska@suse.cz>
14880 * doc/gcov.texi: Enhance documentation of gcov.
14882 2017-04-28 Martin Liska <mliska@suse.cz>
14884 * doc/gcov.texi: Sort options in alphabetic order.
14885 * doc/gcov-dump.texi: Likewise.
14886 * doc/gcov-tool.texi: Likewise.
14887 * gcov.c (print_usage): Likewise.
14888 * gcov-dump.c (print_usage): Likewise.
14889 * gcov-tool.c (print_merge_usage_message): Likewise.
14890 (print_rewrite_usage_message): Likewise.
14891 (print_overlap_usage_message): Likewise.
14893 2017-04-28 Martin Liska <mliska@suse.cz>
14895 PR gcov-profile/53915
14896 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
14898 2017-04-28 Martin Liska <mliska@suse.cz>
14900 PR gcov-profile/79891
14901 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
14902 is marked by compiler as living on a line.
14903 (get_cycles_count): Remove usage of the union.
14904 (output_intermediate_file): Likewise.
14905 (find_source): Fix GNU coding style.
14906 (accumulate_line_counts): Remove old non-all block mode.
14907 (output_lines): Remove usage of the union.
14908 * profile.c (output_location): Include all BBs, even if
14909 belonging to a same line (and file) as a previous BB.
14911 2017-04-28 Martin Liska <mliska@suse.cz>
14913 * gcov.c (process_args): Handle new argument 'w'.
14914 (read_graph_file): Assign ID to BBs.
14915 (output_branch_count): Display BB # if verbose flag is set.
14916 (output_lines): Likewise for arcs.
14917 (print_usage): Add '--verbose' option help.
14918 * doc/gcov.texi: Document --verbose (-w) option.
14920 2017-04-28 Martin Liska <mliska@suse.cz>
14922 * gcov.c (struct block_location_info): New struct.
14923 (process_file): Fill up the new structure.
14924 (read_graph_file): Replace usage of encoding by the newly added
14926 (add_line_counts): Likewise.
14927 (accumulate_line_counts): Remove usage of the union.
14928 (function_info::function_info): New function.
14929 (function_info::~function_info): Likewise.
14930 (process_file): Call delete instead of release_function.
14931 (release_function): Release the function.
14932 (release_structures): Call delete instead of release_function.
14933 (solve_flow_graph): Replace usage of num_blocks.
14934 (find_exception_blocks): Likewise.
14935 (output_lines): Fix GNU coding style.
14937 2017-04-28 Martin Liska <mliska@suse.cz>
14940 * coverage.c (coverage_remove_note_file): New function.
14941 * coverage.h: Declare the function.
14942 * toplev.c (finalize): Clean if an error has been seen.
14944 2017-04-28 Martin Liska <mliska@suse.cz>
14946 PR gcov-profile/80031
14947 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
14948 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
14949 * gcov.c (read_graph_file): Read just number of blocks.
14950 * profile.c (branch_prob): Do not stream 0 flags per a basic
14953 2017-04-28 Martin Liska <mliska@suse.cz>
14955 * gcov-dump.c (tag_*): Add new argument to declarations.
14956 (dump_gcov_file): Likewise.
14957 (tag_blocks): Add and use new argument depth.
14958 (tag_arcs): Likewise.
14959 (tag_lines): Likewise.
14960 (tag_counters): Likewise.
14961 (tag_summary): Likewise.
14962 (dump_working_sets): Use depth to do a proper indentation.
14964 2017-04-28 Jakub Jelinek <jakub@redhat.com>
14967 * cgraph.h (symtab_node::debug_symtab): No longer inline.
14968 * symtab.c (symtab_node::debug_symtab): Move definition here.
14970 2017-04-28 Richard Biener <rguenther@suse.de>
14972 * lto-streamer.h (LTO_major_version): Bump to 7.
14974 2017-04-28 Richard Biener <rguenther@suse.de>
14976 * tree-vrp.c (assert_info): New struct.
14977 (add_assert_info): New helper.
14978 (register_edge_assert_for_2): Refactor to add asserts to a vector
14980 (register_edge_assert_for_1): Likewise.
14981 (register_edge_assert_for): Likewise.
14982 (finish_register_edge_assert_for): New helper actually registering
14983 asserts where live on edge.
14984 (find_conditional_asserts): Adjust.
14985 (find_switch_asserts): Likewise.
14986 (evrp_dom_walker::try_find_new_range): Generalize.
14987 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
14989 2017-04-27 Marek Polacek <polacek@redhat.com>
14992 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
14993 arg10 and arg11 to itype.
14995 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
14997 * doc/extend.texi (Object Size Checking): Improve grammar.
14999 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
15002 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
15003 that the logic for permitting reciprocal estimates matches that
15006 2017-04-27 Jakub Jelinek <jakub@redhat.com>
15009 * tree.c (type_cache_hasher::equal): Only compare
15010 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
15011 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
15012 non-aggregate element types.
15013 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
15014 about the flag on ARRAY_TYPEs in the comment, formatting fix.
15016 2017-04-27 Richard Biener <rguenther@suse.de>
15018 PR middle-end/80533
15019 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
15020 stripping ARRAY_REFs from MEM_EXPR make sure we're not
15021 keeping a reference to a trailing array.
15023 2017-04-27 Richard Biener <rguenther@suse.de>
15025 PR middle-end/80539
15026 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
15027 being in loop-closed SSA form conservatively.
15028 (chrec_fold_multiply_poly_poly): Likewise.
15030 2017-04-27 Tamar Christina <tamar.christina@arm.com>
15032 PR middle-end/79665
15033 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
15034 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
15036 2017-04-27 Jakub Jelinek <jakub@redhat.com>
15039 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
15040 (aarch64_function_arg_alignment): Return unsigned int again, but still
15041 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
15042 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
15043 Don't emit -Wpsabi note.
15044 (aarch64_function_arg_boundary): Likewise.
15045 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
15048 2017-04-26 Nathan Sidwell <nathan@acm.org>
15050 * tree.h (crc32_unsigned_n): Declare.
15051 (crc32_unsigned, crc32_unsigned): Make inline.
15052 * tree.c (crc32_unsigned_bits): Replace with ...
15053 (crc32_unsigned_n): ... this.
15054 (crc32_unsigned, crc32_byte): Remove.
15055 (crc32_string): Remove unnecessary braces.
15057 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15059 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
15060 * ipa-inline-analysis.c (MAX_TIME): Remove.
15061 (account_size_time): Use sreal for time.
15062 (dump_inline_summary): Update.
15063 (estimate_function_body_sizes): Update.
15064 (estimate_edge_size_and_time): Update.
15065 (estimate_calls_size_and_time): Update.
15066 (estimate_node_size_and_time): Update.
15067 (inline_merge_summary): Update.
15068 (inline_update_overall_summary): Update.
15069 (estimate_time_after_inlining): Update.
15070 (inline_read_section): Update.
15071 (inline_write_summary): Update.
15072 * ipa-inline.c (compute_uninlined_call_time): Update.
15073 (compute_inlined_call_time): Update.
15074 (recursive_inlining): Update.
15075 (inline_small_functions): Update.
15076 (dump_overall_stats): Update.
15077 * ipa-inline.h: Include sreal.h.
15078 (size_time_entry): Turn time to sreal.
15079 (inline_summary): Turn self_time nad time to sreal.
15081 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15083 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
15085 (sreal::stream_out, sreal::stream_in): New.
15086 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
15088 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15090 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
15093 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
15096 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
15098 (dimode_scalar_chain::compute_convert_gain): Ditto.
15099 (dimode_scalar_chain::make_vector_copies): Ditto.
15100 (dimode_scalar_chain::convert_reg): Ditto.
15101 (dimode_scalar_chain::convert_insn): Ditto.
15102 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
15103 (VI248_AVX512BW_1): New mode iterator.
15104 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
15105 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
15108 2017-04-25 Martin Sebor <msebor@redhat.com>
15110 PR tree-optimization/80497
15111 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
15112 constants are representable in HOST_WIDE_INT.
15113 (parse_directive): Ditto.
15115 2017-04-25 Martin Sebor <msebor@redhat.com>
15118 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
15119 (new_zero_array): Adjust signature.
15120 (dom_info::dom_init): Used unsigned rather that size_t.
15121 (dom_info::dom_info): Same.
15123 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15124 Jakub Jelinek <jakub@redhat.com>
15127 * config/arm/arm.c: Include gimple.h.
15128 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
15129 returns negative, increment ncrn only if it returned positive.
15130 (arm_needs_doubleword_align): Return int instead of bool,
15131 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
15132 members, but if there is any such non-FIELD_DECL
15133 > PARM_BOUNDARY aligned decl, return -1 instead of false.
15134 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
15135 returns negative, increment nregs only if it returned positive.
15136 (arm_setup_incoming_varargs): Likewise.
15137 (arm_function_arg_boundary): Emit -Wpsabi note if
15138 arm_needs_doubleword_align returns negative, return
15139 DOUBLEWORD_ALIGNMENT only if it returned positive.
15141 2017-04-25 Marek Polacek <polacek@redhat.com>
15144 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
15145 first argument to type.
15147 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
15150 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
15151 type checks to test for compatibility instead of equality.
15153 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15154 Jakub Jelinek <jakub@redhat.com>
15157 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
15159 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
15160 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
15161 the alignment computation, but return their maximum in warn_alignment.
15162 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
15163 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
15165 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
15166 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
15169 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15171 * config/arc/simdext.md (dmpyh): Fix typo.
15173 2017-04-25 Richard Biener <rguenther@suse.de>
15175 PR tree-optimization/80492
15176 * alias.c (compare_base_decls): Handle registers with asm
15177 specification conservatively.
15178 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
15179 compare_base_decls returning dont-know properly.
15181 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15183 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
15184 (legitimate_offset_address_p): New function.
15185 (arc_legitimate_address_p): Use above function.
15187 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15189 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
15191 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15193 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
15194 ACCH registers whenever they are available.
15196 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15198 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
15199 double regs fix when not used.
15201 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15203 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
15205 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
15206 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
15208 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15210 * config/arc/arc.c (arc_output_addsi): Check for h-register class
15211 when emitting short ADD instructions.
15213 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15215 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
15217 (cmpsi_cc_c_insn): Likewise.
15218 (cbranchsi4_scratch): Compute proper instruction length using
15219 compact_hreg_operand.
15220 * config/arc/predicates.md (compact_hreg_operand): New predicate.
15222 2017-04-25 Richard Biener <rguenther@suse.de>
15224 PR middle-end/80509
15225 * passes.c (pass_manager::pass_manager): Initialize
15226 m_name_to_pass_map.
15228 2017-04-25 Richard Biener <rguenther@suse.de>
15230 PR tree-optimization/79201
15231 * tree-ssa-sink.c (statement_sink_location): Handle calls.
15233 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15236 * config/s390/vector.md: Split MEM->GPR vector moves for
15237 non-s_operand addresses.
15239 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15242 * config/s390/predicates.md (reload_const_wide_int_operand): New
15244 * config/s390/s390.md ("movti"): Remove d/P alternative.
15245 ("movti_bigconst"): New pattern definition.
15247 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15250 * s390-protos.h (s390_expand_cs_hqi): Removed.
15251 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
15252 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
15253 modes as well as CCZ1mode and CCZmode.
15254 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
15255 signature of s390_emit_compare_and_swap.
15256 (s390_expand_cs_hqi): Likewise, make static.
15257 (s390_expand_cs_tdsi): Generate an explicit compare before trying
15258 compare-and-swap, in some cases.
15259 (s390_expand_cs): Wrapper function.
15260 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
15262 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
15263 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
15264 patterns for small and large integers. Forbid symref memory operands.
15265 Move expander to s390.c. Require cc register.
15266 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
15267 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
15268 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
15269 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
15270 symref memory operands. Remove CC mode and call s390_match_ccmode
15272 ("atomic_exchange<mode>"): Allow and implement all integer modes.
15274 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15276 * config/s390/s390.md (define_peephole2): New peephole to help
15277 combining the load-and-test pattern with volatile memory.
15279 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15281 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
15282 with CCZmode for TARGET_Z196.
15284 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15286 PR rtl-optimization/80501
15287 * combine.c (make_compound_operation_int): Set subreg_code to SET
15288 even for AND with mask of the sign bit of mode.
15290 PR rtl-optimization/80500
15291 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
15292 sum's initial value.
15294 2017-04-25 Julian Brown <julian@codesourcery.com>
15295 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15297 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
15299 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
15301 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
15303 2017-04-25 Julian Brown <julian@codesourcery.com>
15304 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15306 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
15307 (thunderx2t99_sha): New Reservation.
15309 2017-04-25 Julian Brown <julian@codesourcery.com>
15310 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15312 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
15313 type for 1-element load.
15315 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
15317 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
15319 2017-04-24 Martin Jambor <mjambor@suse.cz>
15321 PR tree-optimization/80293
15322 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
15323 char arrays not totally scalarizable if it is false.
15324 (analyze_all_variable_accesses): Pass correct value in the new
15325 parameter. Add a statistics counter.
15327 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
15329 PR middle-end/79931
15330 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
15332 2017-04-24 Richard Biener <rguenther@suse.de>
15334 PR tree-optimization/80494
15335 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
15336 out for complex types.
15338 2017-04-24 Richard Biener <rguenther@suse.de>
15340 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
15341 * tree-ssa-sccvn.c (print_scc): Print SCC size.
15342 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
15343 (DFS): Adjust and never fail.
15344 (sccvn_dom_walker::fail): Remove.
15345 (sccvn_dom_walker::before_dom_children): Adjust.
15346 (run_scc_vn): Likewise and never fail.
15347 * tree-ssa-pre.c (pass_pre::execute): Adjust.
15348 (pass_fre::execute): Likewise.
15350 2017-04-24 Richard Biener <rguenther@suse.de>
15352 PR tree-optimization/79725
15353 * tree-ssa-sink.c (statement_sink_location): Return whether
15354 failure reason was zero uses. Move that check later.
15355 (sink_code_in_bb): Deal with zero uses by removing the stmt
15358 2017-04-24 Richard Biener <rguenther@suse.de>
15361 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
15362 pointer-based references.
15364 2017-04-24 Richard Biener <rguenther@suse.de>
15367 * pass_manager.h (pass_manager::operator new): Remove.
15368 (pass_manager::operator delete): Likewise.
15369 * passes.c (pass_manager::operator new): Remove.
15370 (pass_manager::operator delete): Likewise.
15371 (pass_manager::pass_manager): Zero individual pass members.
15373 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
15376 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
15377 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
15378 Check "XEXP (src, 1)" operand here.
15379 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
15380 Check "XEXP (src, 1)" operand here.
15381 (dimode_scalar_chain::make_vector_copies): Detect count register
15382 of a shift instruction. Zero extend count register from QImode
15383 to DImode to satisfy vector shift pattern count operand predicate.
15384 Substitute vector shift count operand with a DImode copy.
15385 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
15388 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
15390 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
15391 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15392 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
15393 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15394 (UNSPEC_NOREX_MEM): Remove definition.
15396 2017-04-21 Richard Biener <rguenther@suse.de>
15398 PR tree-optimization/79547
15399 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15400 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
15401 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
15402 without any constraints.
15404 2017-04-21 Richard Biener <rguenther@suse.de>
15406 PR tree-optimization/78847
15407 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
15409 2017-04-21 Richard Biener <rguenther@suse.de>
15411 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
15412 (build_distinct_type_copy): Likewise.
15413 (build_variant_type_copy): Likewise.
15414 * tree.c (build_qualified_type): Pass down mem-stat info.
15415 (build_distinct_type_copy): Likewise.
15416 (build_variant_type_copy): Likewise.
15418 2017-04-21 Richard Biener <rguenther@suse.de>
15420 PR tree-optimization/80237
15421 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
15423 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
15424 for a simplified result.
15426 2016-04-21 Richard Biener <rguenther@suse.de>
15428 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
15429 sth as strict as a simple_iv but a chrec without symbols and an
15430 operand defined in the loop we are peeling (and not some subloop).
15431 (propagate_constants_for_unrolling): Propagate all constants.
15433 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15436 * config/i386/i386.c (print_reg): Remove assert for disalowed
15437 regno values, call output_operand_lossage instead.
15439 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15442 * config/i386/constraints.md (Yc): New register constraint.
15443 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
15444 Use Yc constraint for alternative 2 of operand 0. Remove
15445 preferred_for_speed attribute.
15447 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15449 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
15450 lastprivate clauses in SIMT case.
15452 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
15454 * doc/invoke.texi (-Wextra-semi): Document new warning option.
15456 2017-04-20 Richard Biener <rguenther@suse.de>
15458 PR tree-optimization/57796
15459 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
15460 as N scalar stores.
15461 (vect_model_load_cost): Cost gathers as N scalar loads.
15463 2017-04-20 Richard Biener <rguenther@suse.de>
15465 * ggc-page.c (ggc_allocated_p): Rename to ...
15466 (safe_lookup_page_table_entry): ... this and return the lookup
15468 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
15470 2017-04-20 Richard Biener <rguenther@suse.de>
15472 PR tree-optimization/80453
15473 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
15474 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
15475 from the conditions.
15476 (vn_phi_eq): Pass them down.
15477 (vn_phi_lookup): Record them.
15478 (vn_phi_insert): Likewise.
15480 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
15482 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
15483 uninitialized variable warning to avoid buffer overrun.
15485 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15488 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
15489 is suppressed for '{ 0 }' in C.
15491 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15493 * BASE-VER: Set to 8.0.0.
15495 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
15497 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
15498 priority .init_array and .fini_array section with SECTION_NOTYPE
15501 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15503 PR middle-end/80423
15504 * tree.h (build_array_type): Add typeless_storage default argument.
15505 * tree.c (type_cache_hasher::equal): Also compare
15506 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
15507 (build_array_type): Add typeless_storage argument, set
15508 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
15510 (build_nonshared_array_type): Adjust build_array_type_1 caller.
15511 (build_array_type): Likewise. Add typeless_storage argument.
15513 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
15514 Jakub Jelinek <jakub@redhat.com>
15516 PR tree-optimization/80426
15517 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
15518 operation on symbolic operands, also compute the overflow for the
15519 invariant part when the operation degenerates into a negation.
15521 2017-04-19 Jakub Jelinek <jakub@redhat.com>
15524 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
15525 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
15528 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
15530 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
15533 * config/avr/avr.c (tree.h): Include it.
15534 (cgraph.h): Include it.
15535 (avr_encode_section_info): Don't warn for uninitialized progmem
15536 variable if it's just an alias.
15538 2017-04-19 Richard Biener <rguenther@suse.de>
15541 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
15542 when needed by AutoPGO.
15544 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
15547 * doc/lto.texi: Remove an extra 'that'.
15549 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
15551 PR rtl-optimization/80429
15552 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
15553 are only used in debug insns.
15555 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
15556 Vladimir Makarov <vmakarov@redhat.com>
15558 * config/sparc/predicates.md (input_operand): Add comment. Return
15559 true for any memory operand when LRA is in progress.
15560 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
15562 2017-04-18 Jeff Law <law@redhat.com>
15565 * mips.md ({return,simple_return}_internal): Do not overwrite
15568 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15570 PR tree-optimization/80443
15571 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
15572 instead of adding 1, subtract -1 and similarly instead of subtracting
15575 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
15577 PR rtl-optimization/80357
15578 * haifa-sched.c (tmp_bitmap): New variable.
15579 (model_recompute): Handle duplicate use records.
15580 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
15581 (free_global_sched_pressure_data): Free it.
15583 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
15586 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
15587 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15588 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15589 instead of SYSTEM_HEADER_DIR.
15591 2017-04-18 Jeff Law <law@redhat.com>
15593 PR middle-end/80422
15594 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
15595 predecessors after walking up the insn chain.
15597 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15600 * dwarf2out.c (modified_type_die): Try harder not to emit internal
15601 sizetype type into debug info.
15603 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15606 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
15607 unneeded test for TARGET_UPPER_REGS_SF.
15608 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
15610 2017-04-18 Jakub Jelinek <jakub@redhat.com>
15613 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
15614 instead of gsi_after_labels.
15616 2017-04-18 Jeff Law <law@redhat.com>
15618 * regcprop.c (maybe_mode_change): Avoid creating copies of the
15622 2017-04-13 Jeff Law <law@redhat.com>
15623 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
15624 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
15626 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
15629 * config/avr/avr.c (intl.h): Include it.
15630 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
15632 2017-04-18 Martin Liska <mliska@suse.cz>
15634 PR gcov-profile/78783
15635 * gcov-tool.c (gcov_output_files): Validate that destination
15636 file is either removed by the tool or by a user.
15638 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
15639 Guy Benyei <guybe@mellanox.com>
15641 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
15642 block, and do not negate it, the stored id is already negative.
15644 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
15646 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
15648 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
15651 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
15652 masks of options that should be turned off if the VSX vector
15653 options are turned off.
15654 (OTHER_P8_VECTOR_MASKS): Likewise.
15655 (OTHER_VSX_VECTOR_MASKS): Likewise.
15656 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
15657 rs6000_disable_incompatible_switches to validate no type switches
15659 (rs6000_incompatible_switch): New function to disallow turning on
15660 other vector options if -mno-vsx, -mno-power8-vector, or
15661 -mno-power9-vector are specified.
15663 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15665 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
15667 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15669 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
15670 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
15671 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
15672 (ARG_POINTER_CFA_OFFSET): Likewise.
15674 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
15676 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
15677 conditions to take advantage of various optimizations.
15679 2017-04-13 Jeff Law <law@redhat.com>
15681 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
15682 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
15683 (zero_extendsidi2_dext): Likewise.
15685 2017-04-13 Jakub Jelinek <jakub@redhat.com>
15688 * fold-const.c (fold_ternary_loc): Revert
15689 use op0 instead of fold_convert_loc (loc, type, arg0) part of
15692 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
15694 PR rtl-optimization/80343
15695 * lra-remat.c (update_scratch_ops): Assign original hard reg to
15696 new scratch pseudo.
15698 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
15701 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
15702 to ubsan_encode_value.
15704 2017-04-13 Jeff Law <law@redhat.com>
15706 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
15707 appearing in DEBUG_INSNs.
15709 2017-04-13 Martin Liska <mliska@suse.cz>
15711 PR gcov-profile/80413
15712 * gcov-io.c (gcov_write_string): Copy to buffer just when
15713 allocated size is greater than zero.
15715 2017-04-13 Jakub Jelinek <jakub@redhat.com>
15718 * dwarf2out.c (decls_for_scope): Ignore declarations of
15719 current_function_decl in BLOCK_NONLOCALIZED_VARS.
15721 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
15724 * ipa-visibility.c (non_local_p): Fix typos.
15725 (localize_node): When localizing symbol in same comdat group,
15726 dissolve the group only when we know external symbols are going
15728 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
15730 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15732 PR tree-optimization/79390
15733 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
15734 order does not result in usable sequence, retry with reversed operand
15740 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
15741 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
15742 op0 instead of fold_convert_loc (loc, type, arg0).
15744 2017-04-12 Jeff Law <law@redhat.com>
15746 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
15747 has a delay slot in the generated code.
15749 * config/cris/cris.md (cris_preferred_reload_class): Return
15750 GENNONACR_REGS rather than GENERAL_REGS.
15752 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15755 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
15756 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
15757 signedness of the result type.
15759 2017-04-12 Richard Biener <rguenther@suse.de>
15760 Jeff Law <law@redhat.com>
15762 PR tree-optimization/80359
15763 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
15764 trim stores to TARGET_MEM_REFs.
15766 2017-04-12 Richard Biener <rguenther@suse.de>
15768 PR tree-optimization/79390
15769 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
15770 threading case even more.
15772 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
15775 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
15776 for quad_address_p for TImode, instead of just not indexed_address.
15778 2017-04-12 Richard Biener <rguenther@suse.de>
15779 Bernd Edlinger <bernd.edlinger@hotmail.de>
15781 PR middle-end/79671
15782 * alias.c (component_uses_parent_alias_set_from): Handle
15783 TYPE_TYPELESS_STORAGE.
15784 (get_alias_set): Likewise.
15785 * tree-core.h (tree_type_common): Add typeless_storage flag.
15786 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
15787 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
15788 for types containing members with TYPE_TYPELESS_STORAGE.
15789 (place_field): Likewise.
15790 (layout_type): Likewise for ARRAY_TYPE.
15791 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
15792 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
15793 TYPE_TYPELESS_STORAGE.
15794 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
15796 2017-04-12 Jakub Jelinek <jakub@redhat.com>
15799 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
15800 first argument to type.
15802 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15806 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
15807 CONST0_RTX (mode) rather than const0_rtx where appropriate.
15808 (rs6000_expand_binop_builtin): Likewise.
15809 (rs6000_expand_ternop_builtin): Likewise; also add missing
15810 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
15811 vshasigma built-ins.
15812 * doc/extend.texi: Document that vec_xxpermdi's third argument
15813 must be a constant.
15815 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
15817 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
15818 Use shift_const cost parameter when calculating gain of STV shifts.
15820 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
15822 PR rtl-optimization/70478
15823 * lra-constraints.c (process_alt_operands): Check memory for
15824 disfavoring memory insn operand.
15826 2017-04-11 Jakub Jelinek <jakub@redhat.com>
15828 PR middle-end/80100
15829 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
15830 left shift in unsigned HOST_WIDE_INT type.
15832 PR rtl-optimization/80385
15833 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
15834 (not (neg X)) into (plus X -1) for complex or non-integral modes.
15837 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
15838 if they have any depend clauses.
15840 2017-04-11 Martin Liska <mliska@suse.cz>
15843 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
15844 * ipa-split.c (split_function): Create a local comdat symbol
15845 if caller is in a comdat group.
15847 2017-04-11 Martin Liska <mliska@suse.cz>
15850 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
15853 2017-04-11 Martin Sebor <msebor@redhat.com>
15855 PR middle-end/80364
15856 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
15857 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
15859 (directive::set_width, directive::set_precision, format_character):
15861 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
15864 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
15867 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
15868 conflict, set target->arch_name instead of target->cpu_name.
15870 2017-04-11 Richard Biener <rguenther@suse.de>
15872 PR tree-optimization/80374
15873 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
15874 build_zero_cst, remove fold_convertible_p check again.
15876 2017-04-11 Martin Liska <mliska@suse.cz>
15879 * ubsan.c (instrument_object_size): Do not instrument register
15882 2017-04-11 Jakub Jelinek <jakub@redhat.com>
15885 * config/i386/i386-builtin-types.def
15886 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
15887 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
15888 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
15889 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
15890 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
15891 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
15892 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
15893 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
15894 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
15895 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
15896 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
15897 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
15898 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
15899 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
15900 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
15901 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
15902 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
15903 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
15904 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
15905 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
15906 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
15907 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
15908 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
15909 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
15910 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
15911 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
15912 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
15913 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
15914 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
15915 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
15916 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
15917 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
15918 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
15919 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
15920 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
15921 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
15922 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
15923 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
15924 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
15925 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
15926 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
15927 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
15928 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
15929 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
15930 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
15931 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
15933 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
15934 flag to second_arg_count, handle 4 argument function type _COUNT
15935 aliases, handle second_arg_count on second argument rather than last.
15937 2017-04-10 Jeff Law <law@redhat.com>
15939 PR tree-optimization/80374
15940 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
15941 record anything if we can not convert integer_zero_node to the
15944 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
15947 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15948 Enhance special handling given to the TARGET_P9_MINMAX option in
15949 relation to certain other options.
15951 2017-04-10 Bin Cheng <bin.cheng@arm.com>
15953 PR tree-optimization/80153
15954 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
15955 remove POINTER_PLUS_EXPR's base part directly, rather than through
15958 2017-04-10 Richard Biener <rguenther@suse.de>
15959 Bin Cheng <bin.cheng@arm.com>
15961 PR tree-optimization/80153
15962 * tree-affine.c (aff_combination_to_tree): Get base pointer from
15963 the first element of pointer type aff_tree. Build result expr in
15965 (add_elt_to_tree): Convert to type unconditionally. Remove other
15966 fold_convert calls.
15967 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
15968 (rewrite_use_nonlinear_expr): Check invariant using iv information.
15970 2017-04-10 Richard Biener <rguenther@suse.de>
15972 * tree-ssa-structalias.c (find_func_aliases): Properly handle
15975 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
15977 PR rtl-optimization/70478
15978 * lra-constraints.c (curr_small_class_check): New.
15979 (update_and_check_small_class_inputs): New.
15980 (process_alt_operands): Update curr_small_class_check. Disfavor
15981 alternative insn memory operands. Check available regs for small
15984 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
15987 * config/mips/mips.opt (-mvirt): Update description.
15988 * doc/invoke.texi (-mvirt): Likewise.
15990 2017-04-10 Richard Biener <rguenther@suse.de>
15992 PR middle-end/80362
15993 * fold-const.c (fold_binary_loc): Look at unstripped ops when
15994 looking for NEGATE_EXPR in -A / -B to A / B folding.
15996 2017-04-10 Martin Liska <mliska@suse.cz>
15998 PR gcov-profile/80224
15999 * gcov.c (print_usage): Fix usage string.
16000 (get_gcov_intermediate_filename): Remove.
16001 (output_gcov_file): Use both for normal and intermediate format.
16002 (generate_results): Do not initialize special file for
16003 intermediate format.
16005 2017-04-10 Richard Biener <rguenther@suse.de>
16007 PR tree-optimization/80304
16008 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
16011 2017-04-10 Nathan Sidwell <nathan@acm.org>
16014 * config/rs6000/rs6000.c (rs6000_vector_type): New.
16015 (rs6000_init_builtins): Use it.
16017 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16019 * config/arm/arm.md (<mrc>): Add mode to SET source.
16020 (<mrrc>): Likewise.
16022 2017-04-10 Richard Biener <rguenther@suse.de>
16024 PR middle-end/80344
16025 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
16027 2017-04-10 Jakub Jelinek <jakub@redhat.com>
16030 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
16031 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
16032 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
16033 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
16034 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
16035 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
16036 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
16037 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
16038 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
16039 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
16040 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
16041 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
16042 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
16043 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
16044 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
16045 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
16046 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
16047 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
16048 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
16049 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
16050 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
16051 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
16052 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
16054 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
16056 PR rtl-optimization/70478
16057 * lra-constraints.c: Reverse the last patch.
16059 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
16061 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
16062 Add comment for WCHAR_T.
16064 2017-04-08 Martin Liska <mliska@suse.cz>
16067 2017-04-07 Martin Liska <mliska@suse.cz>
16070 * ipa-split.c (split_function): Add function part to a same comdat
16073 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16076 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
16078 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
16080 * rs6000/rs6000.c (vec_load_pendulum): Rename...
16081 (vec_pairing): ...to this.
16082 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
16083 (rs6000_sched_init): Adjust for name change.
16084 (struct rs6000_sched_context): Likewise.
16085 (rs6000_init_sched_context): Likewise.
16086 (rs6000_set_sched_context): Likewise.
16088 2017-04-07 Jakub Jelinek <jakub@redhat.com>
16094 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
16096 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
16097 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
16098 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
16100 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
16102 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
16104 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
16106 PR rtl-optimization/70703
16107 * ira-color.c (update_conflict_hard_regno_costs): Use
16108 int64_t instead of HOST_WIDE_INT.
16110 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
16112 PR rtl-optimization/70478
16113 * lra-constraints.c (process_alt_operands): Disfavor alternative
16114 insn memory operands.
16116 2017-04-07 Jeff Law <law@redhat.com>
16118 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
16119 CALL and NOTE_INSN_CALL_ARG_LOCATION.
16121 2017-04-07 Martin Liska <mliska@suse.cz>
16124 * config/aarch64/aarch64.c (aarch64_process_target_attr):
16125 Show error message instead of an ICE.
16127 2017-04-07 Martin Liska <mliska@suse.cz>
16130 * ipa-split.c (split_function): Add function part to a same comdat
16133 2017-04-07 Richard Biener <rguenther@suse.de>
16135 PR middle-end/80341
16136 * tree.c (get_unwidened): Also handle ! for_type case for
16138 * convert.c (do_narrow): Split out from ...
16139 (convert_to_integer_1): ... here. Do not pass final truncation
16140 type to get_unwidened for TRUNC_DIV_EXPR.
16142 2017-04-07 Richard Biener <rguenther@suse.de>
16144 * tree-affine.c (wide_int_ext_for_comb): Take type rather
16146 (aff_combination_const): Adjust.
16147 (aff_combination_scale): Likewise.
16148 (aff_combination_add_elt): Likewise.
16149 (aff_combination_add_cst): Likewise.
16150 (aff_combination_convert): Likewise.
16151 (add_elt_to_tree): Likewise. Remove unused argument.
16152 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
16154 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
16156 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
16158 * config/arm/arm.c (arm_default_short_enums): Use
16159 ARM_DEFAULT_SHORT_ENUMS.
16160 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
16162 2017-04-06 Jakub Jelinek <jakub@redhat.com>
16165 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
16166 members with redundant out-of-class redeclaration.
16168 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
16171 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
16172 * config/i386/i386.md (*zero_extendsidi2):
16173 Add (?*x,*x) and (?*v,*v) alternatives.
16175 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
16178 * config/i386/i386.c (ix86_expand_builtin)
16179 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
16180 mode from insn data. Convert operands to insn operand mode.
16181 Copy operands that don't satisfy insn predicate to a register.
16183 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
16185 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
16188 2017-04-06 Richard Biener <rguenther@suse.de>
16190 PR tree-optimization/80334
16191 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
16192 preserve alignment of accesses.
16194 2017-04-06 Richard Biener <rguenther@suse.de>
16196 PR tree-optimization/80262
16197 * tree-sra.c (build_ref_for_offset): Preserve address-space
16199 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
16200 Drop useless address-space information on MEM_REF offsets.
16202 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
16204 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
16206 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
16208 PR rtl-optimization/70703
16209 * ira-color.c (update_conflict_hard_regno_costs): Use
16210 HOST_WIDE_INT instead of long.
16212 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
16215 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
16216 not defined for x86_64 target. Add -mmmx target option when __SSE2__
16218 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
16219 for x86_64 target. Handle -m3dnowa option.
16221 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
16223 PR rtl-optimization/70703
16224 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
16225 (update_conflict_hard_regno_costs): Use long instead of unsigned
16226 arithmetic for cost calculation.
16228 2017-04-05 Jakub Jelinek <jakub@redhat.com>
16229 Bernd Edlinger <bernd.edlinger@hotmail.de>
16232 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
16235 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
16238 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
16239 ptr_mode with Pmode throughout.
16240 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
16241 into probe_stack_range and use DImode.
16243 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
16246 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
16247 call_eh_return is true.
16249 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16251 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
16252 Initialize last_match_fntype_index.
16254 2017-04-05 Jakub Jelinek <jakub@redhat.com>
16257 * tree-nvr.c: Include internal-fn.h.
16258 (pass_return_slot::execute): Ignore internal calls without
16261 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16262 Richard Biener <rguenther@suse.de>
16265 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
16266 captures used multiple times, except for the last use.
16267 * generic-match-head.c: Include gimplify.h.
16269 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16271 PR tree-optimization/79390
16272 * target.h (struct noce_if_info): Declare.
16273 * targhooks.h (default_noce_conversion_profitable_p): Declare.
16274 * target.def (noce_conversion_profitable_p): New target hook.
16275 * ifcvt.h (struct noce_if_info): New type, moved from ...
16276 * ifcvt.c (struct noce_if_info): ... here.
16277 (noce_conversion_profitable_p): Renamed to ...
16278 (default_noce_conversion_profitable_p): ... this. No longer
16280 (noce_try_store_flag_constants, noce_try_addcc,
16281 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
16282 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
16283 instead of noce_conversion_profitable_p.
16284 * config/i386/i386.c: Include ifcvt.h.
16285 (ix86_option_override_internal): Don't override
16286 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
16287 (ix86_noce_conversion_profitable_p): New function.
16288 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
16289 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
16290 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
16291 * doc/tm.texi: Regenerated.
16293 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16295 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
16298 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16301 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
16302 instructions for small multiply cores.
16304 2017-04-04 Jeff Law <law@redhat.com>
16306 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
16308 (mips_expand_vec_perm_const): Initialize elements in orig_perm
16309 that are not set by the loop over the elements.
16311 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16314 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
16315 int mode, convert_modes it to mode as unsigned, otherwise use
16316 lowpart_subreg to mode rather than SImode.
16317 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
16318 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
16319 Use DImode instead of SImode for the shift count operand.
16320 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
16323 2017-04-04 Richard Biener <rguenther@suse.de>
16325 PR middle-end/80281
16326 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
16327 arithmetic done for the negate or the plus. Simplify.
16328 (A - (-B) -> A + B): Likewise.
16329 * fold-const.c (split_tree): Make sure to not negate pointers.
16331 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
16333 PR rtl-optimization/60818
16334 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
16335 a compare of comparisons with the thing compared if this results
16336 in a different machine mode.
16338 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
16340 * alias.c (base_alias_check): Fix typo in comment.
16341 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
16342 * cgraphunit.c (symbol_table::compile): Likewise.
16343 * collect2.c (maybe_run_lto_and_relink): Likewise.
16344 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
16345 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
16346 * config/avr/avr.c (avr_map_op_t): Likewise.
16347 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
16348 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
16349 * config/epiphany/epiphany.md (movcc): Likewise.
16350 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
16351 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
16353 * config/mips/mips.c (mips_save_restore_reg): Likewise.
16354 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
16355 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
16356 * config/sh/sh.c (sh_rtx_costs): Likewise.
16357 * fold-const.c (fold_truth_andor): Likewise.
16358 * genautomata.c (collapse_flag): Likewise.
16359 * gengtype.h (struct type::u::s): Likewise.
16360 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
16361 * input.c (FORMAT_AMOUNT): Likewise.
16362 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
16363 (known_aggs_to_agg_replacement_list): Likewise.
16364 * ipa-inline-analysis.c: Likewise.
16365 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
16366 * ipa-polymorphic-call.c
16367 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
16368 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
16369 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
16371 * modulo-sched.c (apply_reg_moves): Likewise.
16372 * omp-expand.c (build_omp_regions_1): Likewise.
16373 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
16374 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
16375 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
16376 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
16377 * value-prof.c: Likewise.
16378 * var-tracking.c (val_reset): Likewise.
16380 2017-04-03 Richard Biener <rguenther@suse.de>
16382 PR tree-optimization/80275
16383 * fold-const.c (split_address_to_core_and_offset): Handle
16386 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
16388 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
16389 descriptors is at least equal to that of functions.
16391 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16393 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
16395 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16398 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
16399 (mov<IMOD4:mode>): New expander.
16400 (*mov<IMOD4:mode>_internal): New insn and split pattern.
16402 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
16404 PR rtl-optimization/79405
16405 * fwprop.c (propagations_left): New variable.
16406 (forward_propagate_into): Decrement it.
16407 (fwprop_init): Initialize it.
16408 (fw_prop): If the variable has reached zero, stop propagating.
16409 (fwprop_addr): Ditto.
16411 2017-03-31 Jakub Jelinek <jakub@redhat.com>
16414 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
16415 a FUNCTION_DECL, pass it as decl instead of origin to
16418 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
16420 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
16423 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
16426 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
16427 TARGET_VSX_SMALL_INTEGER.
16429 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16431 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16432 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
16434 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
16436 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
16437 extraction from odd-numbered MSA register.
16439 2017-03-31 Jakub Jelinek <jakub@redhat.com>
16441 PR middle-end/80173
16442 * expmed.c (store_bit_field_1): Don't attempt to create
16443 a word subreg out of hard registers wider than word if they
16444 have HARD_REGNO_NREGS of 1 for their mode.
16446 PR middle-end/80163
16447 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
16448 conversions to integer types wider than word and pointer.
16451 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
16452 (rtx_equal_for_cselib_p): Pass 0 to it.
16453 * cselib.c (cselib_hasher::equal): Likewise.
16454 (rtx_equal_for_cselib_1): Add depth argument. If depth
16455 is 128, don't look up VALUE locs and punt. Increment
16456 depth in recursive calls when walking VALUE locs.
16458 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
16460 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
16461 (make_gcov_file_name): Use the canonical path name for generating
16463 (read_line): Fix handling of files with ascii null bytes.
16465 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
16467 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
16468 to initialise a vector register instead
16469 of using a const_int.
16471 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16473 PR translation/80189
16474 * gimplify.c (omp_default_clause): Use %qs instead of %s in
16475 diagnostic messages.
16477 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
16480 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
16481 (dfp_diex_<mode>): Update mode of operand 1.
16482 * doc/extend.texi (dxex, dxexq): Document change to return type.
16483 (diex, diexq): Document change to argument type.
16485 2017-03-30 Martin Jambor <mjambor@suse.cz>
16488 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
16489 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
16490 it reflects the signature changes performed at the callee side.
16491 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
16492 to cgraph_build_function_type_skip_args.
16493 (build_function_decl_skip_args): Adjust call to the above function.
16495 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16498 * config/i386/sse.md
16499 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
16500 register as dest whenever it is a MEM not rtx_equal_p to the
16501 corresponding dup operand, and when forcing into reg move the
16502 reg into the memory afterwards.
16503 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
16504 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
16505 for the force_reg mode.
16506 (avx512vl_vextractf128<mode>): Use register as dest either
16507 always when a MEM, or when it is a MEM not rtx_equal_p to the
16508 corresponding dup operand, or even not when it is a CONST_VECTOR
16509 depending on the mode and lo vs. hi.
16510 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
16512 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
16513 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
16514 Likewise. Require that operands[2] is even.
16515 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
16516 Remove extraneous parens. Require that operands[2] is a multiple
16518 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
16519 operands[0] is a MEM if <mask_applied>, the predicates/constraints
16520 disallow memory then.
16522 2017-03-30 Richard Biener <rguenther@suse.de>
16524 PR tree-optimization/77498
16525 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
16526 to non-constants over backedges.
16528 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
16530 PR rtl-optimization/80233
16531 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
16532 as last_combined_insn. Do not test for BARRIER_P separately.
16534 2017-03-29 Andreas Schwab <schwab@suse.de>
16537 * calls.c (prepare_call_address): Convert funexp to Pmode before
16538 copying to temp reg.
16540 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16542 PR tree-optimization/80158
16543 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
16544 Handle possible future case of more than one alternate
16546 (replace_rhs_if_not_dup): Likewise.
16547 (replace_one_candidate): Likewise.
16549 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
16551 PR rtl-optimization/80193
16552 * ira.c (ira): Do not check allocation for LRA.
16554 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
16556 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
16557 (nvptx_output_simt_exit): Declare.
16558 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
16559 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
16560 (init_softstack_frame): Move initialization of crtl->is_leaf to...
16561 (nvptx_declare_function_name): ...here. Emit declaration of local
16562 memory space buffer for omp_simt_enter insn.
16563 (nvptx_output_unisimt_switch): New.
16564 (nvptx_output_softstack_switch): New.
16565 (nvptx_output_simt_enter): New.
16566 (nvptx_output_simt_exit): New.
16567 * config/nvptx/nvptx.h (struct machine_function): New fields
16568 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
16569 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
16570 (UNSPECV_SIMT_EXIT): Ditto.
16571 (omp_simt_enter_insn): New insn.
16572 (omp_simt_enter): New expansion.
16573 (omp_simt_exit): New insn.
16574 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
16576 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
16577 (expand_GOMP_SIMT_ENTER_ALLOC): New.
16578 (expand_GOMP_SIMT_EXIT): New.
16579 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
16580 (GOMP_SIMT_ENTER_ALLOC): Ditto.
16581 (GOMP_SIMT_EXIT): Ditto.
16582 * target-insns.def (omp_simt_enter): New insn.
16583 (omp_simt_exit): Ditto.
16584 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
16586 (lower_rec_simd_input_clauses): Implement SIMT privatization.
16587 (lower_rec_input_clauses): Likewise.
16588 (lower_lastprivate_clauses): Handle SIMT privatization.
16590 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
16591 (ompdevlow_adjust_simt_enter): New.
16592 (find_simtpriv_var_op): New.
16593 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
16594 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
16596 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
16597 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
16598 (copy_decl_for_dup_finish): Ditto.
16600 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
16602 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
16605 * config/i386/i386.c (ix86_option_override_internal): Always
16606 allow -mpreferred-stack-boundary=3 for 64-bit targets.
16608 2017-03-28 Bin Cheng <bin.cheng@arm.com>
16610 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
16612 2017-03-28 Bin Cheng <bin.cheng@arm.com>
16614 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
16615 mark new edge's irreducible flag accordign to it.
16616 (vect_do_peeling): Check loop preheader edge's irreducible flag
16617 and pass it to function slpeel_add_loop_guard.
16619 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
16621 PR tree-optimization/80218
16622 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
16623 Update block frequencies and counts.
16625 2017-03-28 Richard Biener <rguenther@suse.de>
16627 PR tree-optimization/78644
16628 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
16629 of a simplification result we may not use it at all.
16631 2017-03-28 Richard Biener <rguenther@suse.de>
16634 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
16635 without arguments, generate default definition of a SSA name.
16637 2017-03-28 Richard Biener <rguenther@suse.de>
16639 PR middle-end/80222
16640 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
16641 TYPE_REF_CAN_ALIAS_ALL references.
16642 * fold-const.c (fold_indirect_ref_1): Likewise.
16644 2017-03-28 Martin Liska <mliska@suse.cz>
16647 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
16648 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
16650 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
16651 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
16653 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
16654 (EXTRA_SPECS): Define.
16655 (SUBTARGET_EXTRA_SPECS): Likewise.
16656 (SUBTARGET_CPP_SPEC): Likewise.
16657 * config/arc/elf.h (EXTRA_SPECS): Renamed to
16658 SUBTARGET_EXTRA_SPECS.
16659 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
16661 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
16663 * config/arc/simdext.md (vst64_insn): Update pattern.
16664 (vld32wh_insn): Likewise.
16665 (vld32wl_insn): Likewise.
16666 (vld64_insn): Likewise.
16667 (vld32_insn): Likewise.
16669 2017-03-28 Marek Polacek <polacek@redhat.com>
16672 * fold-const.c (fold_comparison): Use protected_set_expr_location
16673 instead of SET_EXPR_LOCATION.
16675 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
16677 * tree.c (add_expr): Avoid name lookup warning.
16679 2017-03-27 Jeff Law <law@redhat.com>
16681 PR tree-optimization/80216
16682 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
16683 function name. Limit recursion depth.
16684 (record_temporary_equivalences): Corresponding changes.
16686 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
16688 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
16691 2017-03-27 Jakub Jelinek <jakub@redhat.com>
16694 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
16696 * cfgcleanup.c (reg_note_cfa_p): New array.
16697 (insns_have_identical_cfa_notes): New function.
16698 (old_insns_match_p): Don't cross-jump in between /f
16699 and non-/f instructions. If both i1 and i2 are frame related,
16700 verify all CFA notes, their order and content.
16702 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
16705 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
16706 HImode and SImode with zero extend to DImode to one insn.
16707 (bswap<mode>2_extenddi): Likewise.
16708 (bswapsi2_extenddi): Likewise.
16709 (bswaphi2_extendsi): Likewise.
16710 (bswaphi2): Combine bswap HImode and SImode into one insn.
16711 Separate memory insns from swapping register.
16712 (bswapsi2): Likewise.
16713 (bswap<mode>2): Likewise.
16714 (bswaphi2_internal): Delete, no longer used.
16715 (bswapsi2_internal): Likewise.
16716 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
16717 store, and gpr<-gpr swap insns.
16718 (bswap<mode>2_store): Likewise.
16719 (bswaphi2_reg): Register only splitter, combine with the splitter.
16720 (bswaphi2 splitter): Likewise.
16721 (bswapsi2_reg): Likewise.
16722 (bswapsi2 splitter): Likewise.
16723 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
16724 the insns into load, store, and register/register insns.
16725 (bswapdi2_ldbrx): Likewise.
16726 (bswapdi2_load): Likewise.
16727 (bswapdi2_store): Likewise.
16728 (bswapdi2_reg): Likewise.
16730 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
16732 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
16733 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
16735 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
16738 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
16740 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
16741 special handling for target option conflicts between dform
16742 options (-mpower9-dform, -mpower9-dform-vector,
16743 -mpower9-dform-scalar) and -mno-direct-move.
16745 2017-03-27 Richard Biener <rguenther@suse.de>
16747 PR tree-optimization/80181
16748 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
16750 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16752 * config/arc/predicates.md (move_double_src_operand): Replace the
16753 call to move_double_src_operand with a call to address_operand.
16755 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16757 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
16758 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
16759 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
16761 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16763 * config/arc/predicates.md (long_immediate_loadstore_operand):
16764 Consider scaled addresses cases.
16766 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
16768 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
16769 restored when in interrupt.
16770 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
16771 doesn't have delay slot.
16773 2017-03-27 Richard Biener <rguenther@suse.de>
16776 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
16777 inlined thunk clones.
16779 2017-03-27 Jakub Jelinek <jakub@redhat.com>
16782 * asan.c (instrument_derefs): Copy over last operand from
16783 original COMPONENT_REF to the new COMPONENT_REF with
16784 DECL_BIT_FIELD_REPRESENTATIVE.
16785 * ubsan.c (instrument_object_size): Likewise.
16787 2017-03-27 Richard Biener <rguenther@suse.de>
16789 PR tree-optimization/80170
16790 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
16791 sure DR/SCEV didnt fold in constants we do not see when looking
16792 at the reference base alignment.
16794 2017-03-27 Richard Biener <rguenther@suse.de>
16796 PR middle-end/80171
16797 * gimple-fold.c (fold_ctor_reference): Properly guard against
16798 NULL return value from canonicalize_constructor_val.
16800 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
16803 * config/i386/i386.c (ix86_expand_builtin)
16804 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
16805 flags reg setting and flags reg using instructions.
16806 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
16807 clobbering instructions to zero extend op2.
16809 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
16811 * doc/install.texi (Configuration) <--with-aix-soname>:
16812 Update link to AIX ld.
16814 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
16816 PR rtl-optimization/80160
16817 PR rtl-optimization/80159
16818 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
16819 reg_alternate_class into account.
16821 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
16824 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
16825 to consider in curr_insn_transform.
16827 2017-03-24 Jakub Jelinek <jakub@redhat.com>
16829 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
16830 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
16831 and emit_mode_inner.
16833 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16835 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
16836 argument to the overloaded builtin variants. Use the new flag to
16837 deprecate certain builtin variants.
16838 * config/s390/s390-builtin-types.def: Add new builtin types.
16839 * config/s390/s390-builtins.h: Support new flags field for
16840 overloaded builtins.
16841 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
16842 (s390_macro_to_expand): Enable vector float data type.
16843 (s390_cpu_cpp_builtins_internal): Indicate support of the new
16844 builtins by incrementing the __VEC__ version number.
16845 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
16847 (s390_resolve_overloaded_builtin): Emit error messages depending
16848 on the builtin flags.
16849 * config/s390/s390.c (s390_expand_builtin): Support additional
16850 flags argument. Change error message to match the messages
16851 emitted in s390-c.c.
16852 * config/s390/s390.md: New UNSPEC_* constants.
16853 (op_type): Add new instruction types.
16854 * config/s390/vecintrin.h: Add new builtins and test data class
16856 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
16857 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
16858 (VEC_INEXACT, VEC_NOINEXACT): New constants.
16859 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
16860 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
16861 ("vec_mergel<mode>"): V_HW -> VEC_HW.
16863 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
16864 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
16865 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
16866 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
16868 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
16869 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
16870 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
16871 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
16873 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
16874 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
16875 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
16876 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
16877 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
16878 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
16879 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
16881 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
16882 ("vec_scatter_element<V_HW_4:mode>_DI")
16883 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
16884 ("vec_fpint<mode>", "vflls")
16885 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
16886 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
16887 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
16888 ("*vec_cmphe<mode>_cc"): ... these.
16890 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
16891 mode constant instead of magic value.
16893 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16895 * config/s390/s390.c (s390_expand_vec_compare): Support other
16896 vector floating point modes than just V2DF.
16897 (s390_expand_vcond): Likewise.
16898 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
16899 (s390_cannot_change_mode_class): Prevent mode changes between TF
16900 and V1TF in vector registers.
16901 * config/s390/s390.md (DF, SF): New mode attributes.
16902 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
16903 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
16904 SFmode support for VRs.
16905 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
16907 (VFT, VF_HW): New mode iterators.
16908 (vw, sdx): New mode attributes.
16909 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
16910 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
16911 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
16912 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
16913 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
16914 also the new vector floating point modes. Renaming to ...
16916 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
16917 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
16918 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
16919 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
16920 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
16921 ("vec_unordered<mode>"): ... these.
16923 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
16924 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
16925 ("*vec_extendv2df"): New insn definitions.
16927 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16929 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
16930 ("mulditi3_2", "*muldi3_sign"): New patterns.
16931 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
16932 rename the pattern definition.
16934 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16936 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
16938 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
16940 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16942 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
16943 instruction if possible.
16944 * config/s390/vector.md (vec_halfnumelts): New mode
16946 ("*vec_vllezlf<mode>"): New pattern.
16948 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16950 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
16951 ("popcountv4si2", "popcountv2di2"): Rename to ...
16952 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
16953 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
16955 ("popcount<mode>2_vxe"): New pattern.
16957 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16959 * common/config/s390/s390-common.c (processor_flags_table): Add
16961 * config.gcc: Add arch12.
16962 * config/s390/driver-native.c (s390_host_detect_local_cpu):
16963 Default to arch12 for unknown CPU model numbers.
16964 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
16965 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
16966 PROCESSOR_max sanity check.
16967 * config/s390/s390-opts.h (enum processor_type): Add
16969 * config/s390/s390.c (processor_table): Add arch12.
16970 (s390_expand_builtin): Add check for B_VXE flag.
16971 (s390_issue_rate): Add PROCESSOR_ARCH12.
16972 (s390_get_sched_attrmask): Likewise.
16973 (s390_get_unit_mask): Likewise.
16974 (s390_sched_score): Enable z13 scheduling for arch12.
16975 (s390_sched_reorder): Likewise.
16976 (s390_sched_variable_issue): Likewise.
16977 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
16979 (s390_tune_attr): Use z13 scheduling also for arch12.
16980 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
16981 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
16982 (TARGET_VXE_P): New macros.
16983 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
16984 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
16985 * config/s390/s390.opt: Add arch12 as processor_type.
16987 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16989 * config/s390/s390.md
16990 ("fixuns_truncdddi2", "fixuns_trunctddi2")
16991 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
16992 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
16994 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
16995 Rename expanders to ...
16997 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
16998 ("fixuns_truncdddi2_emu"): ... these.
17000 ("fixuns_trunc<mode>si2_emu"): New expander.
17002 ("*fixuns_truncdfdi2_z13"): Rename to ...
17003 ("*fixuns_truncdfdi2_vx"): ... this.
17005 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17007 * config/s390/2964.md: Remove the single element vector compare
17008 instructions which are no longer used.
17009 * config/s390/s390.c (s390_select_ccmode): Remove handling of
17011 (s390_canonicalize_comparison): Remove handling of DFmode
17013 (s390_expand_vec_compare_scalar): Remove function.
17014 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
17015 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
17017 ("*cmp<mode>_ccs"): Add wfcdb instruction.
17019 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17021 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
17023 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
17024 will anyway by matched by mov<mode>_64dfp.
17026 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17028 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
17029 vlef/vstef. Add missing operand to vleif.
17031 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17033 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
17034 pair for all vector types with 64 bit elements.
17035 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
17036 * config/s390/vector.md (V_HW_64): ... here.
17037 (V_128_NOSINGLE): New mode iterator.
17038 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
17039 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
17040 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
17041 ("*vec_load_pairv2di"): Change to ...
17042 ("*vec_load_pair<mode>"): ... this one.
17044 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17046 * config/s390/constraints.md: Add comments.
17047 (jKK): Reject element sizes > 8 bytes.
17048 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
17050 * config/s390/s390.md: Add the s_operand checks formerly in
17051 s390_split_ok_p to various splitters where they are still
17053 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
17054 for 128 bit vectors. Plus two splitters.
17056 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17058 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
17061 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17064 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
17065 error if the boundary argument is not constant.
17067 2017-03-24 Jakub Jelinek <jakub@redhat.com>
17069 PR rtl-optimization/80112
17070 * loop-doloop.c (doloop_condition_get): Don't check condition
17071 if cmp isn't SET with IF_THEN_ELSE src.
17073 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17075 PR tree-optimization/80158
17076 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
17077 replacing a candidate statement, also replace it for the
17078 candidate's alternate interpretation.
17079 (replace_rhs_if_not_dup): Likewise.
17080 (replace_one_candidate): Likewise.
17082 2017-03-24 Richard Biener <rguenther@suse.de>
17084 PR tree-optimization/80167
17085 * graphite-isl-ast-to-gimple.c
17086 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
17088 (translate_isl_ast_to_gimple::get_rename): Likewise.
17090 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
17092 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
17093 handling of certain combinations of target options, including the
17094 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
17095 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
17097 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17100 * config/arm/arm.md (*load_multiple): Add reload_completed to
17101 matching condition.
17103 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17104 Richard Biener <rguenth@suse.de>
17106 PR tree-optimization/79908
17107 PR tree-optimization/80136
17108 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
17109 been cast away, gimplify_and_add suffices.
17111 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
17113 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
17115 2017-03-23 Richard Biener <rguenther@suse.de>
17117 PR tree-optimization/80032
17118 * gimplify.c (gimple_push_cleanup): Forced unconditional
17119 cleanups still have to go to the conditional_cleanups
17122 2017-03-22 Jakub Jelinek <jakub@redhat.com>
17124 PR tree-optimization/80072
17125 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
17127 (next_operand_entry_id): Change type to unsigned int.
17128 (sort_by_operand_rank): Make sure to return the right return value
17129 even if unsigned fields are bigger than INT_MAX.
17130 (struct oecount): Change cnt and id type to unsigned int.
17131 (oecount_hasher::equal): Formatting fix.
17132 (oecount_cmp): Make sure to return the right return value
17133 even if unsigned fields are bigger than INT_MAX.
17134 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
17137 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
17138 TREE_READONLY on result if writing it more than once.
17141 * doc/invoke.texi (-fsanitize=thread): Document that with
17142 -fnon-call-exceptions atomics are not able to throw
17146 * tsan.c: Include tree-eh.h.
17147 (instrument_builtin_call): Call maybe_clean_eh_stmt or
17148 maybe_clean_or_replace_eh_stmt where needed.
17149 (instrument_memory_accesses): Add cfg_changed argument.
17150 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
17151 if it returned true.
17152 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
17154 PR rtl-optimization/63191
17155 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
17156 wrapper function, moved the whole old content into ...
17157 (ix86_delegitimize_address_1): ... this. New inline function.
17158 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
17159 true as last argument instead of ix86_delegitimize_address.
17161 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
17163 * config/aarch64/aarch64.c (generic_branch_cost): Copy
17164 cortexa57_branch_cost.
17166 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
17168 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
17170 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17173 * doc/md.texi (Constraints): Document wA constraint.
17174 * config/rs6000/constraints.md (wA): New.
17175 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
17176 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
17177 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
17178 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
17180 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
17183 * gimplify.c (is_oacc_declared): New function.
17184 (oacc_default_clause): Use it to set default flags for acc declared
17185 variables inside parallel regions.
17186 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
17187 declared variables.
17188 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
17189 declare attribute to any decl as necessary.
17191 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
17194 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
17195 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
17196 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
17197 (arm_arch_lpae): This.
17198 * config/arm/arm.c (arm_arch7ve): Rename into ...
17199 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
17200 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
17203 2017-03-22 Martin Liska <mliska@suse.cz>
17206 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
17207 error message instead of an ICE.
17209 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17211 * doc/extend.texi (6.11 Additional Floating Types): Revise.
17213 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
17215 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
17217 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
17220 2017-03-21 Martin Sebor <msebor@redhat.com>
17222 * doc/extend.texi: Use "cannot" instead of "can't."
17223 * doc/hostconfig.texi: Same.
17224 * doc/install.texi: Same.
17225 * doc/invoke.texi: Same.
17226 * doc/loop.texi: Same.
17227 * doc/md.texi: Same.
17228 * doc/objc.texi: Same.
17229 * doc/rtl.texi: Same.
17230 * doc/tm.texi: Same.
17231 * doc/tm.texi.in: Same.
17232 * doc/trouble.texi: Same.
17234 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
17237 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
17238 (collect_checksum_attributes): Set it.
17239 (die_checksum_ordered): Use it.
17241 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17243 PR tree-optimization/79908
17244 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
17245 change: For a VA_ARG whose LHS has been cast away, use
17246 force_gimple_operand to construct the side effects.
17248 2017-03-21 David Malcolm <dmalcolm@redhat.com>
17250 PR translation/80001
17251 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
17252 more amenable to translation.
17253 (oacc_loop_auto_partitions): Likewise.
17255 2017-03-21 Marek Polacek <polacek@redhat.com>
17256 Martin Sebor <msebor@redhat.com>
17258 PR tree-optimization/80109
17259 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
17260 on INTEGRAL_TYPE_P.
17262 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17263 Segher Boessenkool <segher@kernel.crashing.org>
17266 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
17267 check reg_used_between_p between insn and one of succ or succ2
17268 depending on if succ is artificial insn not inserted into insn
17271 2017-03-21 Martin Liska <mliska@suse.cz>
17273 PR gcov-profile/80081
17274 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
17275 * doc/gcc.texi: Include gcov-dump stuff.
17276 * doc/gcov-dump.texi: New file.
17278 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
17280 PR rtl-optimization/79150
17281 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
17282 conditional jump, if the jump is the last insn of the loop.
17284 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17285 Richard Biener <rguenth@suse.de>
17287 PR tree-optimization/79908
17288 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
17289 been cast away, use force_gimple_operand to construct the side
17292 2017-03-21 Martin Liska <mliska@suse.cz>
17294 PR libfortran/79956
17295 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
17298 2017-03-21 Brad Spengler <spender@grsecurity.net>
17301 * plugin.c (htab_hash_plugin): New function.
17302 (add_new_plugin): Use it and adjust.
17303 (parse_plugin_arg_opt): Adjust.
17304 (init_one_plugin): Likewise.
17306 2017-03-21 Richard Biener <rguenther@suse.de>
17308 PR tree-optimization/80032
17309 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
17310 if set force the cleanup to happen unconditionally.
17311 (gimplify_target_expr): Push inserted clobbers with force_uncond
17312 to avoid them being removed by control-dependent DCE.
17314 2017-03-21 Richard Biener <rguenther@suse.de>
17316 PR tree-optimization/80122
17317 * tree-inline.c (copy_bb): Do not expans va-arg packs or
17318 va_arg_pack_len when the inlined call stmt requires pack
17320 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
17322 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17325 * tsan.c (instrument_builtin_call): If the memory model argument
17326 is not a constant, assume it is valid.
17329 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
17332 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17334 PR rtl-optimization/79910
17335 * combine.c (can_combine_p): Do not allow combining an I0 or I1
17336 if its dest is used by an insn before I2 (other than the combined
17337 insns themselves, which are properly handled already).
17339 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17342 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17344 * combine.c (record_used_regs): New static function.
17345 (try_combine): Handle situations where there is an additional
17346 instruction between I2 and I3 which needs to have a LOG_LINK
17350 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17352 * combine.c (try_combine): Delete redundant i1 test. Call
17353 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17355 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17358 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
17359 alternatives 13/14.
17361 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17363 PR tree-optimization/80054
17364 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
17365 the optimization if a PHI or any of its arguments is not dominated
17366 by the candidate's basis. Use gphi* rather than gimple* as
17368 (replace_profitable_candidates): Clean up a gimple* variable that
17369 should be a gphi* variable.
17371 2017-03-20 Martin Sebor <msebor@redhat.com>
17374 * doc/extend.texi (attribute constructor): Document present limitation.
17376 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
17379 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
17380 __POWER9_VECTOR__ #ifdef control, change template definition to
17381 use Power9-specific built-in function.
17382 (vec_any_eq): Likewise.
17383 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
17384 to control outcomes from this test.
17385 (vector_ae_<mode>p): For VEC_F modes, likewise.
17387 2017-03-20 Ian Lance Taylor <iant@google.com>
17389 * config/i386/i386.c (ix86_function_regparm): Save an extra
17390 register for -fsplit-stack with DECL_STATIC_CHAIN.
17392 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17395 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
17396 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
17398 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17400 * config/riscv/riscv.c (riscv_print_operand): Use "fence
17402 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
17405 2017-03-20 Marek Polacek <polacek@redhat.com>
17408 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
17410 2017-03-20 Richard Biener <rguenther@suse.de>
17412 PR tree-optimization/80113
17413 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
17414 allocate extra SSA name for PHI def.
17415 (add_close_phis_to_outer_loops): Likewise.
17416 (add_close_phis_to_merge_points): Likewise.
17417 (copy_loop_close_phi_args): Likewise.
17418 (copy_cond_phi_nodes): Likewise.
17420 2017-03-20 Martin Liska <mliska@suse.cz>
17422 PR middle-end/79753
17423 * tree-chkp.c (chkp_build_returned_bound): Do not build
17424 returned bounds for a LHS that's not a BOUNDED_P type.
17426 2017-03-20 Martin Liska <mliska@suse.cz>
17430 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
17431 COMPLEX_CST and VECTOR_CST.
17433 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17436 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
17437 target operand. A new splitter adds the clobber statement in case
17438 the target operand is dead anyway.
17440 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
17442 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
17443 to age-old versions of binutils and glibc.
17445 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
17447 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
17449 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17451 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
17453 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17455 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
17456 requirement for binutils 2.13.
17458 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17460 * combine.c (try_combine): Delete redundant i1 test. Call
17461 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17463 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
17465 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
17466 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
17468 <riscv64-*-elf>: Re-arrange section
17469 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
17470 <riscv32-*-linux>: Likewise.
17471 <riscv64-*-elf>: Likewise
17472 <riscv64-*-linux>: Likewise.
17474 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
17477 * aarch64.opt(verbose-cost-dump): Fix typo.
17479 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
17482 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
17483 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
17485 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17487 * reload.c (find_reloads): When reloading a nonoffsettable address,
17488 use RELOAD_OTHER for it and its address reloads.
17490 PR rtl-optimization/79910
17491 * combine.c (record_used_regs): New static function.
17492 (try_combine): Handle situations where there is an additional
17493 instruction between I2 and I3 which needs to have a LOG_LINK
17496 2017-03-17 Jeff Law <law@redhat.com>
17498 PR tree-optimization/71437
17499 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
17500 conditional in the hash table first.
17501 (vrp_dom_walker::before_dom_children): Extract condition from
17502 ASSERT_EXPR. Record condition, its inverion and any implied
17503 conditions as well.
17505 2017-03-17 Marek Polacek <polacek@redhat.com>
17506 Markus Trippelsdorf <markus@trippelsdorf.de>
17508 PR tree-optimization/80079
17509 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
17512 2017-03-17 Richard Biener <rguenther@suse.de>
17514 PR middle-end/80075
17515 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
17516 Properly verify the LHS before the RHS possibly claims to be
17518 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
17521 2017-03-17 Martin Jambor <mjambor@suse.cz>
17523 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
17524 (List of -O2 options): Likewise.
17525 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
17528 2017-03-17 Tom de Vries <tom@codesourcery.com>
17530 * gcov-dump.c (print_usage): Print bug_report_url.
17532 2017-03-17 Richard Biener <rguenther@suse.de>
17534 PR middle-end/80050
17535 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
17536 (parser::peek): Likewise.
17538 2017-03-17 Richard Biener <rguenther@suse.de>
17540 PR tree-optimization/80048
17541 * sese.c (free_sese_info): Properly release rename_map and
17542 copied_bb_map elements.
17544 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
17546 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
17547 Add linked-list forward and backlinks. Insert on
17548 construction, remove on destruction.
17549 (class pass_store_merging): Add m_stores_head field.
17550 (pass_store_merging::terminate_and_process_all_chains):
17551 Iterate over m_stores_head list.
17552 (pass_store_merging::terminate_all_aliasing_chains):
17554 (pass_store_merging::execute): Check for debug stmts first.
17555 Push new chains onto the m_stores_head stack.
17557 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
17560 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
17561 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
17562 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
17564 2017-03-16 Jeff Law <law@redhat.com>
17566 PR tree-optimization/71437
17567 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
17568 member function. Implementation moved into after_dom_children
17569 member function and into the threader's thread_outgoing_edges
17571 (dom_opt_dom_walker::after_dom_children): Simplify by moving
17572 some code into new thread_outgoing_edges.
17573 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
17574 definition. Simplify marker handling (do it here). Assume we always
17575 have the available expression and the const/copies tables.
17576 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
17578 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
17579 * tree-vrp.c (equiv_stack): No longer file scoped.
17580 (vrp_dom_walker): New class.
17581 (vrp_dom_walker::before_dom_children): New member function.
17582 (vrp_dom_walker::after_dom_children): Likewise.
17583 (identify_jump_threads): Setup domwalker. Use it rather than
17584 walking edges in a random order by hand. Simplify setup/finalization.
17585 (finalize_jump_threads): Remove.
17586 (vrp_finalize): Do not call identify_jump_threads here.
17587 (execute_vrp): Do it here instead and call thread_through_all_blocks
17590 PR tree-optimization/71437
17591 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
17593 (simplify_stmt_for_jump_threading): Add basic_block argument. All
17595 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
17596 (dom_opt_dom_walker::thread_across_edge): Remove
17597 handle_dominating_asserts argument. All callers changed.
17598 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
17599 changes. Remove calls to lhs_of_dominating_assert. Other
17600 uses of handle_dominating_asserts turn into unconditional code
17601 (simplify_control_stmt_condition_1): Likewise.
17602 (simplify_control_stmt_condition): Likewise.
17603 (thread_through_normal_block, thread_across_edge): Likewise.
17604 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
17605 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
17606 object if it is not an SSA_NAME.
17607 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
17608 before calling into the VRP specific simplifiers.
17609 (identify_jump_threads): Remove handle_dominating_asserts
17612 2017-03-16 Jakub Jelinek <jakub@redhat.com>
17615 * tree-diagnostic.c (default_tree_printer): No longer static.
17616 * tree-diagnostic.h (default_tree_printer): New prototype.
17618 2017-03-16 Tamar Christina <tamar.christina@arm.com>
17620 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
17621 Change ins into fmov.
17623 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17625 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
17626 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
17627 Use h_con constraint for operand 1.
17628 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
17629 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
17631 2017-03-15 Jeff Law <law@redhat.com>
17633 PR tree-optimization/71437
17634 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
17635 (record_temporary_equivalences): Use it.
17637 PR tree-optimization/71437
17638 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
17639 tree-ssa-scopedtables.
17640 (lookup_avail_expr, build_and_record_new_cond): Likewise.
17641 (record_conditions, record_cond, vuse_eq): Likewise.
17642 (record_edge_info): Adjust to API tweak of record_conditions.
17643 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
17644 (record_temporary_equivalences, optimize_stmt): Likewise.
17645 (eliminate_redundant_computations): Likewise.
17646 (record_equivalences_from_stmt): Likewise.
17647 * tree-ssa-scopedtables.c: Include options.h and params.h.
17648 (vuse_eq): New function, moved from tree-ssa-dom.c
17649 (build_and_record_new_cond): Likewise.
17650 (record_conditions): Likewise. Accept vector of conditions rather
17651 than edge_equivalence structure for first argument.
17652 for the first argument.
17653 (avail_exprs_stack::lookup_avail_expr): New member function, moved
17654 from tree-ssa-dom.c.
17655 (avail_exprs_stack::record_cond): Likewise.
17656 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
17657 from tree-ssa-dom.c.
17658 (avail_exprs_stack): Add new member functions lookup_avail_expr
17660 (record_conditions): Declare.
17662 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
17665 * lra-constraints.c (process_alt_operands): Increase reject for
17666 reloading an input/output operand.
17668 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
17671 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
17672 insns to convert from signed/unsigned char/short to IEEE 128-bit
17674 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
17676 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
17679 * config/i386/i386.c (ix86_vector_duplicate_value): Create
17680 subreg of inner mode for values already in registers.
17682 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
17684 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
17685 iteration reg is used after the loop.
17687 2017-03-14 Martin Sebor <msebor@redhat.com>
17689 PR tree-optimization/79800
17690 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
17691 precision in negative-positive range.
17692 (format_floating): Call non-const overload with adjusted precision.
17694 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
17697 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
17700 2017-03-14 Martin Sebor <msebor@redhat.com>
17702 PR middle-end/80020
17703 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
17704 * builtins.def (aligned_alloc): Use it.
17707 * Makefile.in (GTFILES): Add calls.c.
17708 * calls.c: Include "gt-calls.h".
17710 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
17712 PR rtl-optimization/79728
17713 * regs.h (struct target_regs): New field
17714 x_contains_allocatable_regs_of_mode.
17715 (contains_allocatable_regs_of_mode): New macro.
17716 * reginfo.c (init_reg_sets_1): Initialize it, and change
17717 contains_reg_of_mode so it includes global regs as well.
17718 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
17719 rather than contains_regs_of_mode.
17721 2017-03-14 Martin Liska <mliska@suse.cz>
17723 * doc/invoke.texi: Document options that can't be combined with
17724 -fcheck-pointer-bounds.
17726 2017-03-14 Martin Liska <mliska@suse.cz>
17728 PR middle-end/79831
17729 * doc/invoke.texi (-Wchkp): Document the option.
17731 2017-03-14 Martin Liska <mliska@suse.cz>
17733 * Makefile.in: Install gcov-dump.
17735 2017-03-14 Martin Liska <mliska@suse.cz>
17737 * multiple_target.c (expand_target_clones): Bail out for
17738 an invalid attribute.
17740 2017-03-14 Richard Biener <rguenther@suse.de>
17742 * alias.c (struct alias_set_entry): Pack properly.
17743 * cfgloop.h (struct loop): Likewise.
17744 * cse.c (struct set): Likewise.
17745 * ipa-utils.c (struct searchc_env): Likewise.
17746 * loop-invariant.c (struct invariant): Likewise.
17747 * lra-remat.c (struct cand): Likewise.
17748 * recog.c (struct change_t): Likewise.
17749 * rtl.h (struct address_info): Likewise.
17750 * symbol-summary.h (function_summary): Likewise.
17751 * tree-loop-distribution.c (struct partition): Likewise.
17752 * tree-object-size.c (struct object_size_info): Likewise.
17753 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
17754 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
17755 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
17756 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
17757 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
17758 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
17759 (struct _stmt_vec_info): Likewise.
17761 2017-03-14 Martin Liska <mliska@suse.cz>
17764 * multiple_target.c (create_dispatcher_calls): Check that
17765 a target can create a function dispatcher.
17767 2017-03-14 Martin Liska <mliska@suse.cz>
17770 * multiple_target.c (expand_target_clones): Drop local.local
17771 flag for default implementation.
17773 2017-03-14 Richard Biener <rguenther@suse.de>
17775 PR tree-optimization/80030
17776 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
17778 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
17780 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
17781 gcc_fallthrough() instead of __attribute__((fallthrough));
17783 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
17785 * doc/gcc.texi: Remove "up" link to (DIR).
17786 * doc/gccint.texi: Ditto.
17788 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
17790 * doc/install.texi (Specific) <avr>: Remove reference to
17793 2017-03-13 Jeff Law <law@redhat.com>
17795 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
17796 attribute rather than comments.
17798 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
17799 match_scratch operand is highest.
17801 2017-03-13 Martin Liska <mliska@suse.cz>
17803 PR middle-end/78339
17804 * ipa-pure-const.c (warn_function_noreturn): If the declarations
17805 is a CHKP clone, use original declaration.
17807 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17809 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
17810 (arc_conditional_register_usage): Use a different allocation order
17811 when optimizing for size.
17812 * common/config/arc/arc-common.c (arc_option_optimization_table):
17813 Section anchors default on when optimizing for size.
17815 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17817 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
17819 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17821 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
17822 * config/arc/arc.md (cpu_facility): Add cd variant.
17823 (*movqi_insn): Add code density variant.
17824 (*movhi_insn): Likewise.
17825 (*movqi_insn): Likewise.
17826 (*addsi3_mixed): Likewise.
17827 (subsi3_insn): Likewise.
17829 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17831 * config/arc/arc.md (movsi_cond_exec): Update constraint.
17833 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
17835 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
17836 expressions with MINUS and UNARY ops.
17838 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17841 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
17843 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
17844 between vec_select and vector argument.
17845 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
17846 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
17847 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
17848 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
17849 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
17850 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
17852 2017-03-13 Richard Biener <rguenther@suse.de>
17855 * params.def (vect-max-peeling-for-alignment): Fix typo.
17857 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17859 * doc/install.texi (Specific) <mips-*-*>: Remove description of
17860 issue that only occurred with binutils below 2.18.
17862 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17864 * doc/install.texi (Specific) <cris-axis-elf>: No longer
17865 refer to binutils 2.11/2.12 minimum.
17867 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
17869 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
17870 ftp.kernel.org and simplify binutils requirement.
17872 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
17874 * doc/invoke.texi (Warning Options): Fix spelling of link-time
17876 (Optimize Options): Ditto. Also remove redundancy.
17878 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17880 PR translation/79848
17881 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
17883 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
17884 to G_ to avoid double translation.
17886 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17888 PR translation/79923
17889 * auto-profile.c (get_combined_location): Convert leading
17890 character of diagnostics to lower case and remove trailing period.
17891 (read_profile): Likewise for various diagnostics.
17892 * config/arm/arm.c (arm_option_override): Remove trailing period
17893 from various diagnostics.
17894 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
17895 (msp430_expand_delay_cycles): Likewise.
17897 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17900 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
17901 full command-line argument, rather than just "str".
17902 (aarch64_validate_march): Likewise.
17903 (aarch64_validate_mtune): Likewise.
17905 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
17907 PR rtl-optimization/78911
17908 * lra-assigns.c (must_not_spill_p): New function.
17909 (spill_for): Use it.
17911 2017-03-10 Jakub Jelinek <jakub@redhat.com>
17913 PR tree-optimization/79981
17914 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
17915 ATOMIC_COMPARE_EXCHANGE ifn result.
17916 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
17917 IFN_ATOMIC_COMPARE_EXCHANGE.
17919 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17922 * opts.c (parse_sanitizer_options): Add missing question mark to
17923 "did you mean" message.
17925 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17927 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
17929 (VMULEUH_UNS): Likewise.
17930 (VMULOUB_UNS): Likewise.
17931 (VMULOUH_UNS): Likewise.
17932 * config/rs6000/rs6000.c (builtin_function_type): Remove
17933 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
17935 2017-03-10 David Malcolm <dmalcolm@redhat.com>
17938 * read-rtl-function.c (function_reader::read_rtx_operand): Update
17939 x with result of extra_parsing_for_operand_code_0.
17940 (function_reader::extra_parsing_for_operand_code_0): Convert
17941 return type from void to rtx, returning x. When reading
17942 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
17943 larger size containing struct block_symbol.
17945 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
17947 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
17948 -mfloat128-hardware without -m64.
17950 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
17953 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
17954 entries to the case statement that marks unsigned arguments to
17955 overloaded functions.
17957 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
17959 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
17960 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
17962 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
17965 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
17966 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
17968 2017-03-10 Martin Liska <mliska@suse.cz>
17972 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
17973 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
17976 2017-03-10 Olivier Hainque <hainque@adacore.com>
17978 * tree-switch-conversion (array_value_type): Start by resetting
17979 candidate type to it's main variant.
17981 2017-03-10 Jakub Jelinek <jakub@redhat.com>
17983 PR rtl-optimization/79909
17984 * combine.c (try_combine): Use simplify_replace_rtx on individual
17985 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
17986 of the whole CALL_INSN_FUNCTION_USAGE.
17988 PR tree-optimization/79972
17989 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
17990 get_range_info on SSA_NAMEs. Formatting fixes.
17992 2017-03-10 Richard Biener <rguenther@suse.de>
17993 Jakub Jelinek <jakub@redhat.com>
17995 PR tree-optimization/77975
17996 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
17997 edge to be constant.
17998 (get_val_for): For constant x return it. Formatting fix.
17999 (loop_niter_by_eval): Avoid pointless looping if the next iteration
18000 would use the same bases as the current one.
18002 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18004 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
18005 instead of vec_select for V1TImode.
18006 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
18008 (VSX_LE_128): Add V1TI to this mode iterator.
18009 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
18010 (*vsx_le_perm_store_<mode>): Likewise.
18011 (pre-reload splitter for VSX stores): Likewise.
18012 (post-reload splitter for VSX stores): Likewise.
18013 (*vsx_xxpermdi2_le_<mode>): Likewise.
18014 (*vsx_lxvd2x2_le_<mode>): Likewise.
18015 (*vsx_stxvd2x2_le_<mode>): Likewise.
18017 2017-03-09 Michael Eager <eager@eagercon.com>
18019 Correct failures with --enable-checking=yes,rtl.
18021 * config/microblaze/microblaze.c (microblaze_expand_shift):
18022 Replace GET_CODE test with CONST_INT_P and INTVAL test with
18023 test for const0_rtx.
18024 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
18025 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
18027 2017-03-09 Richard Biener <rguenther@suse.de>
18029 PR tree-optimization/79977
18030 * graphite-scop-detection.c (scop_detection::merge_sese):
18031 Handle the case of extra exits to blocks dominating the entry.
18033 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
18035 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
18038 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
18040 PR rtl-optimization/79949
18041 * lra-constraints.c (process_alt_operands): Check memory when
18042 trying to predict a cycle. Print about the overall increase.
18044 2017-03-09 Richard Biener <rguenther@suse.de>
18046 PR middle-end/79971
18047 * gimple-expr.c (useless_type_conversion_p): Preserve
18048 TYPE_SATURATING for fixed-point types.
18050 2017-03-09 Richard Biener <rguenther@suse.de>
18053 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
18054 alignment of BLKmode params.
18056 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18059 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
18060 (VALL_NO_V2Q): Likewise.
18062 * config/aarch64/aarch64-simd.md
18063 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
18065 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
18066 VALL_NO_V2Q mode iterator.
18067 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
18069 2017-03-09 Martin Liska <mliska@suse.cz>
18071 PR tree-optimization/79631
18072 * tree-chkp-opt.c (chkp_is_constant_addr): Call
18073 tree_int_cst_sign_bit just for INTEGER constants.
18075 2017-03-09 Martin Liska <mliska@suse.cz>
18079 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
18082 2017-03-09 Marek Polacek <polacek@redhat.com>
18085 * tree.c (inchash::add_expr): Handle TREE_VEC.
18087 2017-03-09 Martin Liska <mliska@suse.cz>
18090 (chkp_narrow_size_and_offset): New function.
18091 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
18092 (void chkp_parse_bit_field_ref): New function.
18093 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
18094 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
18096 2017-03-09 Martin Liska <mliska@suse.cz>
18099 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
18100 (chkp_find_bounds_1): Remove gcc_unreachable.
18102 2017-03-09 Jakub Jelinek <jakub@redhat.com>
18105 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
18106 BUILT_IN_SYNC*, determine the access type from the size suffix and
18107 always build a MEM_REF with that type. Handle forgotten
18108 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
18111 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
18112 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
18113 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
18114 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
18115 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
18116 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
18117 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
18118 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
18119 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
18120 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
18121 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
18122 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
18123 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
18124 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
18125 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
18126 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
18127 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
18128 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
18129 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
18130 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
18131 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
18132 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
18133 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
18134 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
18135 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
18136 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
18137 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
18138 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
18139 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
18140 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
18141 definitions outside of __OPTIMIZE__ guarded section.
18144 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
18145 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
18146 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
18147 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
18150 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18152 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
18153 ("vfenez<mode>"): Add missing constraints.
18155 2017-03-08 Martin Sebor <msebor@redhat.com>
18158 * config/nds32/nds32.c (nds32_option_override):
18159 Fix misspelled diagnostic.
18161 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18164 * gimplify.c (gimplify_omp_for): Replace index var in outer
18165 taskloop statement with an artificial variable and add
18166 OMP_CLAUSE_PRIVATE clause for it.
18168 2017-03-08 Richard Biener <rguenther@suse.de>
18170 PR tree-optimization/79955
18171 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
18172 for accesses that are completely outside of the variable.
18174 2017-03-08 Andrew Haley <aph@redhat.com>
18176 PR tree-optimization/79943
18177 * tree-ssa-loop-split.c (compute_new_first_bound): When
18178 calculating the new upper bound, (END-BEG) should be added, not
18181 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18183 * config/avr/avr.md (setmemhi): Make sure match_dup
18184 operand number comes before match_scratch.
18186 2017-03-08 Richard Biener <rguenther@suse.de>
18188 PR tree-optimization/79920
18189 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
18190 with ncopies == 1 to ...
18191 (vect_transform_slp_perm_load): ... here. Properly compute
18192 all element loads by iterating VF times over the group. Do
18193 not handle ncopies (computed in a broken way) in
18194 vect_create_mask_and_perm.
18196 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18199 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
18200 is a uniform vector, use uniform_vector_p return value instead of
18201 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
18203 2017-03-07 Marek Polacek <polacek@redhat.com>
18205 PR middle-end/79809
18206 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
18207 (alloca_call_type): Likewise.
18209 2017-03-07 Martin Liska <mliska@suse.cz>
18211 * gcov.c (process_args): Put comment to correct location.
18213 2017-03-07 Martin Liska <mliska@suse.cz>
18215 PR middle-end/68270
18216 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
18217 Use array_at_struct_end_p instead of DECL_CHAIN (field).
18218 (chkp_narrow_bounds_for_field): Likewise.
18219 (chkp_parse_array_and_component_ref): Pass one more argument to
18222 2017-03-07 Richard Biener <rguenther@suse.de>
18224 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
18227 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
18229 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
18230 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
18232 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18235 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
18236 to end of description.
18237 (PARAM_MAX_STORES_TO_MERGE): Likewise.
18239 2017-03-07 Jakub Jelinek <jakub@redhat.com>
18241 PR rtl-optimization/79901
18242 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
18244 (*avx512f_<code><mode>3<mask_name>): ... this.
18245 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
18246 iterator instead of VI8_AVX2_AVX512BW.
18248 PR rtl-optimization/79901
18249 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
18250 min/max expander, expand it using expand_vec_cond_expr.
18253 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
18256 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18259 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
18260 to void * for PCH reasons.
18261 * dwarf2out.c (output_loc_operands, output_die): Cast
18262 v.val_vec.array to unsigned char *.
18264 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
18267 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
18270 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
18272 PR rtl-optimization/79571
18273 * lra-constraints.c (process_alt_operands): Calculate static
18274 reject and subtract it from overall when only addresses will be
18277 2017-03-06 Julia Koval <julia.koval@intel.com>
18280 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
18281 incoming stack boundary to 128 for 64-bit targets.
18283 2017-03-06 Richard Biener <rguenther@suse.de>
18285 PR tree-optimization/79894
18286 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
18287 to NULL after folding it.
18289 2017-03-06 Richard Biener <rguenther@suse.de>
18291 PR tree-optimization/79824
18292 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
18293 check disabling peeling for gaps.
18295 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
18297 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
18298 attributes): Document gettimeofday.
18300 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18302 * config/s390/s390.c (s390_option_override_internal): Set
18303 PARAM_MIN_VECT_LOOP_BOUND
18305 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18307 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
18308 * config/s390/s390.md: Likewise.
18310 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18313 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
18314 (<avx2_avx512>_perm<mode>): Rename to ...
18315 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
18317 (<avx512>_perm<mode>_mask): Rename to ...
18318 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
18320 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
18321 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
18322 instead of VI8F_256_512.
18323 (avx512f_perm<mode>): New define_expand.
18324 (avx512f_perm<mode>_mask): Likewise.
18325 (avx512f_perm<mode>_1<mask_name>): New define_insn.
18326 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
18328 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18330 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
18331 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
18333 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
18335 2017-03-06 Martin Liska <mliska@suse.cz>
18338 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
18339 when having a SSA NAME w/o VAR_DECL assigned to it.
18341 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18343 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
18344 msa_dpsub_<su>_d): Fix MODE for vec_select.
18346 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18348 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
18350 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
18352 2017-03-06 Richard Biener <rguenther@suse.de>
18354 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
18355 * plugin.c (register_plugin_info): Likewise.
18356 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
18358 2017-03-05 Jakub Jelinek <jakub@redhat.com>
18360 * config/i386/sse.md (sse_storehps, sse_storelps,
18361 avx_<castmode><avxsizesuffix>_<castmode>,
18362 avx512f_<castmode><avxsizesuffix>_<castmode>,
18363 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
18364 in condition that at least one operand is not a MEM.
18366 2017-03-03 Jakub Jelinek <jakub@redhat.com>
18368 PR middle-end/79805
18369 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
18370 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
18372 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
18373 gimple_call_nothrow_p flag based on whether original builtin can throw.
18374 If it can, emit following stmts on the fallthrough edge.
18375 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
18376 don't create new bb if inserting just debug stmts on the edge, try to
18377 insert them on the fallthru bb or just reset debug stmts.
18379 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
18382 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
18383 restore recog_data (including the operand rtxes inside it) around
18384 the call to get_insn_template.
18386 2017-03-03 Martin Sebor <msebor@redhat.com>
18388 PR tree-optimization/79699
18389 * context.c (context::~context): Free MPFR caches to avoid
18390 a memory leak on program exit.
18392 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18394 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
18395 Use wide_int::ulow () instead of .elt (0).
18397 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18399 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
18400 (*pushxf): Limit oF constraint to 32bit targets and add oC
18401 constraint for 64bit targets.
18402 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
18403 (*pushdf): Change rmF constraint to rmC.
18405 2017-03-03 Martin Liska <mliska@suse.cz>
18407 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
18408 Remove unused variable.
18410 2017-03-03 Jakub Jelinek <jakub@redhat.com>
18413 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
18414 is a memory operand, increase num_memory.
18415 (ix86_expand_args_builtin): Likewise.
18417 2017-03-03 Jan Hubicka <jh@suse.cz>
18420 * ipa-devirt.c (maybe_record_node): Properly handle
18421 __cxa_pure_virtual visibility.
18423 2017-03-03 Martin Liska <mliska@suse.cz>
18425 PR tree-optimization/79803
18426 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
18428 (pass_loop_prefetch::execute): Disabled optimization if an
18429 assumption about L1 cache size is not met.
18431 2017-03-03 Martin Liska <mliska@suse.cz>
18433 PR rtl-optimization/79574
18434 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
18435 (hash_scan_set): Likewise.
18436 (dump_hash_table): Likewise.
18437 (hoist_code): Likewise.
18439 2017-03-03 Richard Biener <rguenther@suse.de>
18441 * fixed-value.c (fixed_from_string): Restore use of elt (1)
18442 in place of uhigh ().
18443 (fixed_convert_from_real): Likewise.
18445 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18448 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
18450 2017-03-03 Richard Biener <rguenther@suse.de>
18452 PR middle-end/79818
18453 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
18454 TYPE_OVERFLOW_UNDEFINED check.
18456 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18458 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
18460 (vector_ae_<mode>_p): Likewise.
18461 (vector_nez_<mode>_p): Likewise.
18462 (vector_ne_v2di_p): Likewise.
18463 (vector_ae_v2di_p): Likewise.
18464 (vector_ne_<mode>_p): Likewise.
18465 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
18467 (vsx_tsqrt<mode>2_fe): Likewise.
18469 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
18472 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
18474 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18476 PR rtl-optimization/79780
18477 * cprop.c (one_cprop_pass): When second and further conditional trap
18478 in a single basic block is turned into an unconditional trap, turn it
18479 into a deleted note to avoid RTL verification failures.
18481 2017-03-02 Richard Biener <rguenther@suse.de>
18483 * fold-const.c (const_binop): Use ulow () instead of elt (0).
18485 2017-03-02 Richard Biener <rguenther@suse.de>
18487 PR tree-optimization/79345
18489 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
18490 param and abort the walk, returning -1 if it is hit.
18491 (walk_aliased_vdefs): Take a limit param and pass it on.
18492 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
18493 defaulting to 0 and return a signed int.
18494 * tree-ssa-uninit.c (struct check_defs_data): New struct.
18495 (check_defs): New helper.
18496 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
18497 about uninitialized memory.
18498 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
18499 bogus uninitialized warning.
18500 (fixed_convert_from_real): Likewise.
18502 2017-03-02 Bin Cheng <bin.cheng@arm.com>
18504 PR tree-optimization/66768
18505 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
18506 iv_use if base object can't be determined.
18508 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18510 PR tree-optimization/79345
18511 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
18512 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
18513 (get_pattern_stats): Initialize it.
18514 * genemit.c (gen_expand): Verify match_scratch numbers come after
18515 match_operand/match_dup numbers.
18516 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
18517 match_scratch numbers.
18518 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
18520 * config/s390/s390.md (trunctdsd2): Likewise.
18522 2017-03-02 Richard Biener <rguenther@suse.de>
18524 * wide-int.h (wide_int_storage::operator=): Implement in terms
18527 2017-03-02 Richard Biener <rguenther@suse.de>
18529 PR tree-optimization/79777
18530 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
18531 the to insert expression to sth existing.
18533 2017-03-01 Martin Sebor <msebor@redhat.com>
18535 PR middle-end/79692
18536 * gimple-ssa-sprintf.c
18537 (directive::known_width_and_precision): New function.
18538 (format_integer): Use it.
18539 (get_mpfr_format_length): Consider the full range of precision
18540 when computing %g output with the # flag. Set the likely byte
18541 count to 3 rather than 1 when precision is indeterminate.
18542 (format_floating): Correct the lower bound of precision.
18544 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18546 * doc/invoke.texi: Document default code model for 64-bit Linux.
18548 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18551 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
18552 udiv rather than div since input pattern is unsigned.
18554 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
18556 * config/i386/i386.c (print_reg): Warn for values of
18557 unsupported size in integer register.
18559 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
18562 * config/rs6000/predicates.md (current_file_function_operand): Do
18563 not allow self calls to be local if the function is replaceable.
18565 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
18568 * config/rs6000/altivec.h (vec_ctz and others): Change the
18569 preprocessor macro that controls conditional compilation from
18570 _ARCH_PWR9 to __POWER9_VECTOR__.
18571 (vec_all_ne): Change parameterization of __altivec_scalar_pred
18572 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
18573 control (instead of _ARCH_PWR9 control) so that template
18574 definition uses power9-specific function.
18575 (vec_any_eq): Likewise.
18576 (vec_all_ne): Change macro definition to use a power9-specific
18577 expansion under #ifdef __POWER9_VECTOR__ control (instead of
18578 _ARCH_PWR9 control).
18579 (vec_any_eq) Likewise.
18580 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
18581 expansion for CMPNEF to remove support for xvcmpnesp instruction.
18582 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
18583 support for xvcmpnedp instruction.
18584 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
18585 macro expansion so that Power9 implementation of vec_all_ne does
18586 not use the AltiVec predicate framework.
18587 (VCMPNEH_P): Likewise.
18588 (VCMPNEW_P): Likewise.
18589 (VCMPNED_P): Likewise.
18590 (VCMPNEFP_P): Likewise.
18591 (VCMPNEDP_P): Likewise.
18592 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
18593 implementation of vec_any_eq to not use AltiVec predicate
18595 (VCMPAEH_P): Likewise.
18596 (VCMPAEW_P): Likewise.
18597 (VCMPAED_P): Likewise.
18598 (VCMPAEFP_P): Likewise.
18599 (VCMPAEDP_P): Likewise.
18600 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
18601 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
18602 not use the AltiVec predicate framework.
18603 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
18604 of vec_any_eq to not use AltiVec predicate framework.
18605 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
18606 support for predefined __POWER9_VECTOR__ macro to indicate that
18607 Power9 instruction selection is enabled.
18608 (altivec_overloaded_builtins): Remove extraneous
18609 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
18610 function argument types RS6000_BTI_bool_V16QI and
18611 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
18612 entry for overloaded function argument types RS6000_BTI_bool_V4SI
18613 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
18614 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
18615 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
18616 Power9 for implementations of vec_cmpne. Change the signature for
18617 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
18618 (representing vec_all_ne) to remove the previously described first
18619 argument of type RS6000_BTI_INTSI, as this was an artifact of
18620 reliance on the AltiVec predicate framework, which is no longer
18621 used in the implementation of these functions. Add
18622 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
18623 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
18624 since, unlike the AltiVec predicate framework implementation, we
18625 do not share function descriptors between vec_alle and vec_anyeq.
18626 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
18627 set of modes that receive special treatment even when
18628 TARGET_P9_VECTOR is true. The special treatment emits code that
18629 does not depend on Power9 instructions.
18630 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
18631 define_expand to not rely on AltiVec predicate framework.
18632 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
18634 (vector_ne_v2di_p): Change this define_expand to not rely on
18635 AltiVec predicate framework.
18636 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
18638 (vector_ne_<mode>_p): Change this define_expand to not rely on
18639 AltiVec predicate framework.
18640 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
18642 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
18643 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
18644 define_insn pattern.
18645 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
18646 define_insn pattern because the xvcmpne<VSs>. instruction is not
18648 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
18649 instruction is not supported.
18651 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18653 * config/nvptx/nvptx.c: Include intl.h.
18655 2017-03-01 Martin Jambor <mjambor@suse.cz>
18658 * ipa-prop.h (ipa_bits): Removed field known.
18659 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
18660 to pointers. Adjusted their comments to warn about their sharing.
18661 (ipcp_transformation_summary): Change bits to a vector of pointers.
18662 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
18663 (ipa_get_ipa_bits_for_value): Declare.
18664 * tree-vrp.h (value_range): Mark as GTY((for_user)).
18665 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
18666 (ipa_bits_hash_table): Likewise.
18667 (ipa_vr_ggc_hash_traits): Likewise.
18668 (ipa_vr_hash_table): Likewise.
18669 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
18670 being pointers and vr_known being removed.
18671 (ipa_set_jf_unknown): Likewise.
18672 (ipa_get_ipa_bits_for_value): New function.
18673 (ipa_set_jfunc_bits): Likewise.
18674 (ipa_get_value_range): New overloaded functions.
18675 (ipa_set_jfunc_vr): Likewise.
18676 (ipa_compute_jump_functions_for_edge): Use the above functions to
18677 construct bits and vr parts of jump functions.
18678 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
18679 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
18681 (ipcp_grow_transformations_if_necessary): Also allocate
18682 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
18684 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
18685 them. Fix too long lines.
18686 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
18687 vr_known being removed.
18688 (ipa_read_jump_function): Use new setter functions to construct bits
18689 and vr parts of jump functions or set them to NULL.
18690 (write_ipcp_transformation_info): Adjust for bits being pointers.
18691 (read_ipcp_transformation_info): Likewise.
18692 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
18694 Include gt-ipa-prop.h.
18695 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
18697 (ipcp_store_bits_results): Likewise.
18698 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
18699 Do not write to existing jump functions but use a temporary instead.
18701 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18704 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
18705 attempt to use its first operand as BIT_FIELD_REF base.
18707 2017-03-01 Richard Biener <rguenther@suse.de>
18709 PR middle-end/79721
18710 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
18711 interpolating formula in wrapping arithmetic.
18712 (chrec_apply): Convert chrec_evaluate return value to wanted type.
18714 2017-03-01 Jakub Jelinek <jakub@redhat.com>
18716 PR tree-optimization/79734
18717 * tree-vect-generic.c (expand_vector_condition): Optimize
18718 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
18719 Handle VEC_COND_EXPR where comparison has different inner width from
18720 type's inner width.
18722 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
18724 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
18725 markup, and similar issues. Remove @opindex entries for things
18726 that aren't options. Add missing -mmpy-option entries.
18728 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18730 PR tree-optimization/79737
18731 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
18732 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
18733 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
18734 instead of byte_size. Formatting fix.
18735 (shift_bytes_in_array_right): Formatting fix.
18737 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
18740 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
18741 condition on optimize for the leaf function test.
18743 2017-02-28 Martin Liska <mliska@suse.cz>
18746 * read-rtl-function.c (function_reader::handle_unknown_directive):
18747 Bail out when one uses -flto.
18749 2017-02-28 Martin Liska <mliska@suse.cz>
18751 * common.opt: Replace space with tabular for options of <number>
18753 * config/i386/i386.opt: Show <number> value for
18754 -mlarge-data-threshold.
18755 * opts.c (print_filtered_help): Do not display number in hexadecimal
18758 2017-02-28 Martin Liska <mliska@suse.cz>
18760 * common.opt: Fix --help=option -Q for options which are of
18763 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
18765 * config/i386/i386.c (print_reg): Error out for values
18766 of 8-bit size in invalid integer register.
18768 2017-02-28 Martin Sebor <msebor@redhat.com>
18770 PR tree-optimization/79691
18771 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
18773 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18776 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
18777 gcc_unreachable with output_operand_lossage.
18779 2017-02-28 Richard Biener <rguenther@suse.de>
18781 PR tree-optimization/79740
18782 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
18784 (visit_nary_op): Insert the nary into the hashtable if we
18785 pattern-matched sth.
18786 * tree-ssa-pre.c (eliminate_insert): Robustify.
18788 2017-02-28 Richard Biener <rguenther@suse.de>
18790 PR middle-end/79731
18791 * fold-const.c (decode_field_reference): Reject out-of-bound
18794 2017-02-28 Jakub Jelinek <jakub@redhat.com>
18796 * config/i386/i386.c: Include intl.h.
18797 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
18798 instead of just cond ? "..." : "...".
18799 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
18800 * coverage.c (read_counts_file): Likewise.
18801 * omp-offload.c: Include intl.h.
18802 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
18803 of just cond ? "..." : "...".
18804 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
18805 of just cond ? "..." : "...".
18807 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
18810 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
18812 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
18813 'tune for' CPU name.
18814 * config/arm/arm-cpu-data.h: Regenerated.
18816 2017-02-28 Richard Biener <rguenther@suse.de>
18818 PR tree-optimization/79732
18819 * tree-inline.c (expand_call_inline): Do not shadow var.
18821 2017-02-28 Richard Biener <rguenther@suse.de>
18823 PR tree-optimization/79723
18824 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
18825 address-space properly.
18827 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
18829 * doc/optinfo.texi (Optimization groups): Fix option used for
18831 * doc/invoke.texi (-fopt-info): Document "omp".
18832 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
18833 (OPTGROUP_ALL): Add OPTGROUP_OMP.
18834 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
18835 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
18836 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
18838 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
18840 * dumpfile.c (optgroup_options): Instead of "openmp", associate
18841 OPTGROUP_OMP with "omp".
18843 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
18846 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
18847 for arithmetic shift of unsigned V2DI.
18849 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
18851 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
18852 arc/linux.h headers.
18853 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
18854 (LINK_SPEC): Likewise.
18855 (ARC_TLS_EXTRA_START_SPEC): Likewise.
18856 (EXTRA_SPECS): Likewise.
18857 (STARTFILE_SPEC): Likewise.
18858 (ENDFILE_SPEC): Likewise.
18859 (LIB_SPEC): Likewise.
18860 (TARGET_SDATA_DEFAULT): Likewise.
18861 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
18862 (MULTILIB_DEFAULTS): Likewise.
18863 (DWARF2_UNWIND_INFO): Likewise.
18864 * config/arc/big.h: New file.
18865 * config/arc/elf.h: Likewise.
18866 * config/arc/linux.h: Likewise.
18867 * config/arc/t-uClibc: Remove.
18869 2017-02-27 Bin Cheng <bin.cheng@arm.com>
18871 PR tree-optimization/77536
18872 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
18873 (tree_transform_and_unroll_loop): Use above function to compute the
18874 estimated niter of unrolled loop and use it when scaling profile.
18875 Also use count info rather than frequency if it's non-zero.
18876 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
18877 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
18878 (vect_transform_loop): Call above function.
18880 2017-02-27 Richard Biener <rguenther@suse.de>
18882 PR tree-optimization/45397
18883 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
18884 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
18885 (visit_nary_op): Add pattern matching for CSEing sign-changed
18886 or truncated operations with wider ones.
18888 2017-02-27 Richard Biener <rguenther@suse.de>
18890 PR tree-optimization/79690
18891 * tree-vect-stmts.c (vectorizable_store): Use vector type
18892 built from the DR with address-space.
18894 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
18896 * doc/invoke.texi (Optimize Options): Refine the description
18897 of asan-use-after-return.
18899 2017-02-25 Alan Modra <amodra@gmail.com>
18901 PR rtl-optimization/79584
18902 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
18903 base, not ad->base_term, the reg within base. Remove assertion
18904 that ad->base == ad->base_term. Replace gen_int_mode using
18905 bogus mode with const0_rtx.
18907 2017-02-25 Jakub Jelinek <jakub@redhat.com>
18909 PR middle-end/79396
18910 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
18911 FMA_EXPR like tcc_binary or tcc_unary.
18913 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
18916 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
18918 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
18919 (output_loc_operands): Handle DW_OP_call_ref and
18920 DW_OP_GNU_variable_value.
18921 (struct variable_value_struct): New type.
18922 (struct variable_value_hasher): Likewise.
18923 (variable_value_hash): New variable.
18924 (string_types): Remove.
18925 (copy_loc_descr): New function.
18926 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
18927 (prepend_loc_descr_to_each): New function.
18928 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
18929 instead of add_loc_descr_to_each if the first argument is single
18930 location list and the second has multiple.
18931 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
18932 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
18933 when looking for variable value which doesn't have other location info.
18934 (loc_list_from_tree): Formatting fix.
18935 (gen_array_type_die): Simplify DW_AT_string_length handling.
18936 (adjust_string_types): Remove.
18937 (gen_subprogram_die): Don't call adjust_string_types nor test/set
18938 string_types. Call resolve_variable_values.
18939 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
18940 (resolve_addr_in_expr): Likewise. Add A argument.
18941 (copy_deref_exprloc): Remove deref argument. Adjust for the
18942 original expression being DW_OP_GNU_variable_value with optionally
18943 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
18944 optionally after it.
18945 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
18946 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
18947 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
18948 (variable_value_hasher::hash, variable_value_hasher::equal): New
18950 (resolve_variable_value_in_expr, resolve_variable_value,
18951 resolve_variable_values, note_variable_value_in_expr,
18952 note_variable_value): New functions.
18953 (dwarf2out_early_finish): Call note_variable_value on all toplevel
18956 2017-02-24 Jakub Jelinek <jakub@redhat.com>
18959 * opts.h (handle_generated_option): Add GENERATED_P argument.
18960 * opts-common.c (handle_option): Adjust function comment.
18961 (handle_generated_option): Add GENERATED_P argument, pass it to
18963 (control_warning_option): Pass false to handle_generated_option
18965 * opts.c (maybe_default_option): Pass true to handle_generated_option
18967 * optc-gen.awk: Likewise.
18969 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18971 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
18972 a REG, look at the REG it is a SUBREG of.
18973 (splitter for cmpeqsi_t): Ditto.
18975 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18977 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
18978 the special USEs with the pattern of the insn, not the insn itself.
18980 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
18983 * doc/invoke.texi: Document -mload-store-pairs.
18985 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18986 Sandra Loosemore <sandra@codesourcery.com>
18988 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
18989 argument isn't a CONST_INT.
18990 (nios2_alternate_compare_const): Assert op is a CONST_INT.
18991 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
18992 (nios2_validate_compare): Bypass alternate compare logic if *op2
18993 is not a CONST_INT.
18994 (ldstwm_operation_p): Return false if first_base is not a REG or
18995 if first_offset is not a CONST_INT.
18997 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
18999 * config/cris/cris.md: Use correct operand in a define_peephole2.
19001 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19003 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
19005 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19007 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
19008 this_insn if it is an INSN or JUMP_INSN.
19009 (force_offsettable): Look at base, not at addr.
19010 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
19011 on things that aren't necessarily CONST_INTs.
19013 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
19015 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
19016 -mfpmath=sse is the default also for x86-32 targets with SSE2
19017 instruction set when @option{-ffast-math} is enabled
19019 2017-02-24 Jeff Law <law@redhat.com>
19021 PR rtl-optimizatoin/79286
19022 * ira.c (update_equiv_regs): Drop may_trap_p exception to
19025 2017-02-24 Richard Biener <rguenther@suse.de>
19027 PR tree-optimization/79389
19028 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
19031 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
19033 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
19034 function comment to reflect reality.
19035 (loop_exits_before_overflow): Fix typo in function description.
19037 2017-02-24 Richard Biener <rguenther@suse.de>
19039 PR tree-optimization/79389
19040 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
19041 properly that a threading opportunity exists. Detect conditional
19042 copy/constant propagation opportunities.
19044 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
19046 * config/visium/visium.md (type): Add trap.
19047 (b): New mode attribute.
19048 (*btst): Rename into...
19049 (*btst<mode>): ...this and adjust.
19050 (*cbranchsi4_btst_insn): Rename into...
19051 (*cbranch<mode>4_btst_insn): ...this and adjust.
19052 (trap): New define_insn.
19054 2017-02-23 Jakub Jelinek <jakub@redhat.com>
19056 PR tree-optimization/79389
19057 * ifcvt.c (struct noce_if_info): Add rev_cond field.
19058 (noce_reversed_cond_code): New function.
19059 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
19060 reversed_comparison_code. Formatting fix.
19061 (noce_try_store_flag): Test rev_cond != NULL in addition to
19062 reversed_comparison_code.
19063 (noce_try_store_flag_constants): Likewise.
19064 (noce_try_store_flag_mask): Likewise.
19065 (noce_try_addcc): Use rev_cond if non-NULL instead of
19066 reversed_comparison_code.
19067 (noce_try_cmove_arith): Likewise. Formatting fixes.
19068 (noce_try_minmax, noce_try_abs): Clear rev_cond.
19069 (noce_find_if_block): Initialize rev_cond.
19070 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
19071 instead of false as last argument never attempt to reverse it
19074 2017-02-23 Bin Cheng <bin.cheng@arm.com>
19076 PR tree-optimization/79663
19077 * tree-predcom.c (combine_chains): Process refs in reverse order
19078 only for ZERO length chains, and add explaining comment.
19080 2017-02-23 Jeff Law <law@redhat.com>
19082 PR tree-optimization/79578
19083 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
19084 in call to operand_equal_p.
19086 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
19089 * config/i386/cpuid.h: Fix another undefined behavior.
19091 2017-02-23 Richard Biener <rguenther@suse.de>
19093 PR tree-optimization/79683
19094 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
19095 vector types for data-refs.
19097 2017-02-23 Martin Liska <mliska@suse.cz>
19099 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
19101 2017-02-23 Jakub Jelinek <jakub@redhat.com>
19103 PR middle-end/79665
19104 * internal-fn.c (get_range_pos_neg): Moved to ...
19105 * tree.c (get_range_pos_neg): ... here. No longer static.
19106 * tree.h (get_range_pos_neg): New prototype.
19107 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
19108 are known to be in between 0 and signed maximum inclusive, try to
19109 expand both unsigned and signed divmod and use the cheaper one from
19112 2017-02-22 Jeff Law <law@redhat.com>
19114 PR tree-optimization/79578
19115 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
19116 to compare base operands.
19118 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
19121 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
19122 gpc_reg_operand instead of fpr_reg_operand.
19124 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
19126 * config/mips/mips.c (mips_return_in_memory): Force FP
19127 vector types to be returned in memory for o32 ABI.
19129 2017-02-22 Jakub Jelinek <jakub@redhat.com>
19131 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
19132 instead of DW_TAG_member for static data member declarations and don't
19133 set no_linkage_name for static inline data members.
19134 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
19137 2017-02-22 Martin Liska <mliska@suse.cz>
19139 * doc/invoke.texi: Replace inequality signs with square brackets
19142 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19144 PR tree-optimization/68644
19145 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
19147 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
19150 * lra-constraints.c (simplify_operand_subreg): Handle
19151 WORD_REGISTER_OPERATIONS targets.
19153 2017-02-22 Jakub Jelinek <jakub@redhat.com>
19156 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
19157 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
19158 elimination by swapping fld*.
19160 2017-02-22 Richard Biener <rguenther@suse.de>
19162 PR tree-optimization/79673
19163 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
19164 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
19165 irrelevant address-space qualifiers and avoiding a
19166 ADDR_SPACE_CONVERT_EXPR from fold_convert.
19168 2017-02-22 Richard Biener <rguenther@suse.de>
19170 PR tree-optimization/79666
19171 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
19172 to not symbolically negate if that may introduce undefined
19175 2017-02-22 Martin Liska <mliska@suse.cz>
19178 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
19179 * data-streamer-out.c (streamer_write_gcov_count_stream):
19181 * value-prof.c (stream_out_histogram_value): Make assert more
19182 precise based on type of counter.
19184 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
19187 * config/i386/i386.md (standard_x87sse_constant_load splitter):
19188 Use nonimmediate_operand instead of memory_operand for operand 1.
19189 (float-extend standard_x87sse_constant_load splitter): Ditto.
19191 2017-02-21 Jeff Law <law@redhat.com>
19193 PR tree-optimization/79621
19194 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
19195 blocks with edges to themselves.
19197 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19200 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
19201 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
19202 Use gimple_call_builtin_p.
19205 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
19206 on temporarily removed DEBUG_INSNs.
19208 PR tree-optimization/79649
19209 * tree-loop-distribution.c (classify_partition): Give up on
19210 non-generic address space loads/stores.
19212 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
19214 * doc/loop.texi (Loop manipulation): Remove nonexistent
19215 tree_ssa_loop_version from the documentation.
19216 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
19218 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19221 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
19222 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
19223 * config/rs6000/rs6000.c: Include except.h.
19224 (rs6000_expand_split_stack_prologue): Call
19225 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
19227 2017-02-21 Martin Jambor <mjambor@suse.cz>
19230 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
19231 have been analyzed.
19233 2017-02-21 Martin Jambor <mjambor@suse.cz>
19235 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
19236 for backward compatibility only.
19237 * doc/invoke.texi (Option Summary): Remove all references to
19238 -fipa-cp-alignment.
19240 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
19244 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19246 * lra-constraints.c (curr_insn_transform): Handle
19247 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19249 2017-02-21 Martin Liska <mliska@suse.cz>
19251 * config/i386/i386.opt: Replace -masm-dialect with -masm.
19253 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
19255 PR translation/79638
19256 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
19258 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
19261 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
19262 (arm_function_ok_for_sibcall): Return false for an indirect call by
19263 descriptor if all the argument registers are used.
19264 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
19265 alignment of the function.
19267 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19269 PR tree-optimization/61441
19270 * simplify-rtx.c (simplify_const_unary_operation): For
19271 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
19272 the sNaN unmodified.
19274 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19276 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19277 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19278 instead of SYSTEM_HEADER_DIR.
19280 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
19281 Martin Liška <mliska@suse.cz>
19283 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
19284 Fix typos and grammar, use active voice, and clarify.
19286 2017-02-20 Marek Polacek <polacek@redhat.com>
19288 PR middle-end/79537
19289 * gimplify.c (gimplify_expr): Handle unused *&&L;.
19292 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
19294 2017-02-20 Jakub Jelinek <jakub@redhat.com>
19297 * config/i386/i386.c (ix86_expand_builtin): Handle
19298 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
19299 ix86_builtins_isa[fcode].isa as a requirement of those
19300 flags and any other flag in the bitmask.
19301 (ix86_init_mmx_sse_builtins): Use 0 instead of
19302 ~OPTION_MASK_ISA_64BIT as mask.
19303 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
19304 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
19305 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
19306 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
19308 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19311 * lra-constraints.c (split_reg): Check requested split mode
19312 is supported by the register.
19314 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19316 * lra-constraints.c (simplify_operand_subreg): Remove early
19319 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19322 * lra-constraints.c (curr_insn_transform): Tighten condition
19323 for converting SUBREG reloads from OP_OUT to OP_INOUT.
19325 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19328 * lra-constraints.c (curr_insn_transform): Handle
19329 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19331 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
19334 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
19336 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
19338 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
19341 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
19344 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19346 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
19349 2017-02-18 Jakub Jelinek <jakub@redhat.com>
19351 * final.c (last_columnnum, override_columnnum): New variables.
19352 (final_start_function): Set last_columnnum, pass it to begin_prologue
19353 hook and pass 0 to dwarf2out_begin_prologue.
19354 (final_scan_insn): Update override_columnnum. Pass last_columnnum
19355 to source_line debug hook.
19356 (notice_source_line): Compute last_columnnum and for debug_column_info
19357 return true on column changes.
19358 * debug.h (struct gcc_debug_hooks): Add column argument to
19359 source_line and begin_prologue hooks.
19360 (debug_nothing_int_charstar_int_bool): Remove prototype.
19361 (debug_nothing_int_int_charstar,
19362 debug_nothing_int_int_charstar_int_bool): New prototypes.
19363 (dwarf2out_begin_prologue): Add column argument.
19364 * debug.c (do_nothing_debug_hooks): Adjust source_line and
19365 begin_prologue hooks.
19366 (debug_nothing_int_charstar_int_bool): Remove.
19367 (debug_nothing_int_int_charstar,
19368 debug_nothing_int_int_charstar_int_bool): New functions.
19369 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
19370 through to dwarf2out_source_line.
19371 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
19372 (dwarf2out_source_line): Add column argument, emit it if requested.
19373 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
19375 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19376 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19377 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
19378 through to dwarf2out_begin_prologue.
19379 (vmsdbgout_source_line): Add column argument, pass it through to
19380 dwarf2out_source_line.
19381 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
19382 dbxout_source_line caller.
19383 (dbxout_source_line): Add column argument.
19385 * common.opt (gno-column-info, gcolumn-info): New options.
19386 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
19387 (check_die): Also test for multiple DW_AT_decl_column attributes.
19388 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
19389 DW_AT_decl_column if requested.
19390 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
19392 (gen_variable_die): Likewise.
19393 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
19394 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
19397 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
19398 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
19399 (ix86_handle_option): Handle OPT_m3dnowa.
19400 * doc/invoke.texi (-m3dnowa): Document.
19401 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
19402 -m3dnowa instead of -m3dnow -march=athlon.
19405 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
19406 instead of gcc_assert for K, r and R code checks. Formatting fixes.
19408 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19411 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
19412 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
19413 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
19414 generator for vsx_xxpermdi_<mode>_be.
19415 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
19416 force big-endian semantics.
19417 (vsx_xxpermdi_<mode>_be): New define_expand with same
19418 implementation as previous version of vsx_xxpermdi_<mode>.
19420 2017-02-17 Jakub Jelinek <jakub@redhat.com>
19422 PR tree-optimization/79327
19423 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
19424 variable, its initialization and use.
19426 2017-02-17 Julia Koval <julia.koval@intel.com>
19428 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
19429 (OPTION_MASK_ISA_PKU_UNSET): New.
19430 (ix86_handle_option): Handle -mrdpid.
19431 * config/i386/cpuid.h (bit_RDPID): New.
19432 * config/i386/driver-i386.c (host_detect_local_cpu):
19433 Detect RDPID feature.
19434 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
19435 * config/i386/i386-c.c (ix86_target_macros_internal):
19437 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
19438 (ix86_valid_target_attribute_inner_p): Add "rdpid".
19439 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
19440 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
19441 * config/i386/i386.md (define_insn "rdpid"): New.
19442 * config/i386/i386.opt Add -mrdpid.
19443 * config/i386/immintrin.h (_rdpid_u32): New.
19445 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
19447 PR rtl-optimization/79541
19448 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
19449 instead of transforming it into USE.
19451 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
19453 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
19454 If HONOR_SNANS (SFmode) force the input to a register.
19455 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
19456 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
19457 an frsp or similar insn.
19459 2017-02-17 Martin Liska <mliska@suse.cz>
19461 PR rtl-optimization/79577
19462 * params.def (selsched-max-sched-times): Increase minimum to 1.
19464 2017-02-17 Martin Liska <mliska@suse.cz>
19466 PR rtl-optimization/79574
19467 * gcse.c (want_to_gcse_p): Prevent integer overflow.
19469 2017-02-17 Martin Liska <mliska@suse.cz>
19471 PR tree-optimization/79529
19472 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
19473 ssa_defined_default_def_p to handle cases which are implicitly
19475 * tree-ssa.c (ssa_defined_default_def_p): New function.
19476 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
19477 which are implicitly defined.
19478 * tree-ssa.h (ssa_defined_default_def_p): Declare.
19480 2017-02-17 Richard Biener <rguenther@suse.de>
19482 PR middle-end/79576
19483 * params.def (max-ssa-name-query-depth): Limit to 10.
19485 2017-02-17 Richard Biener <rguenther@suse.de>
19487 PR tree-optimization/79552
19488 * tree-ssa-structalias.c (visit_loadstore): Properly verify
19491 2017-02-17 Richard Biener <rguenther@suse.de>
19494 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
19496 2017-02-17 Marek Polacek <polacek@redhat.com>
19498 PR middle-end/79536
19499 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
19500 (fold_negate_expr): New wrapper.
19502 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
19504 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
19505 Correct terminology and de-emphasize pre-standard behavior.
19507 2017-02-16 Alan Modra <amodra@gmail.com>
19509 PR rtl-optimization/79286
19510 * ira.c (def_dominates_uses): New function.
19511 (update_equiv_regs): Don't create an equivalence for insns that
19512 may trap where the register def does not dominate the use.
19514 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
19516 PR rtl-optimization/78127
19517 * lra.c (lra): Call lra_eliminate before finish the loop after
19520 2017-02-16 Richard Biener <rguenther@suse.de>
19522 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
19524 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
19525 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
19526 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
19527 (isl_val_int_from_wi): New function.
19528 (extract_affine_gmp): Rename to ...
19529 (extract_affine_wi): ... this, take a widest_int.
19530 (extract_affine_int): Just wrap extract_affine_wi.
19531 (add_param_constraints): Use isl_val_int_from_wi.
19532 (add_loop_constraints): Likewise, and extract_affine_wi.
19534 2017-02-15 Jeff Law <law@redhat.com>
19536 PR middle-end/79521
19537 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
19538 ira_init_register_move_cost_if_necessary.
19540 2017-02-15 Martin Sebor <msebor@redhat.com>
19542 PR middle-end/32003
19543 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
19544 removed in a prior commit.
19546 2017-02-15 Bin Cheng <bin.cheng@arm.com>
19548 PR tree-optimization/79347
19549 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
19550 counters during peeling.
19552 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
19554 * Makefile.in (site.exp): Remove "set ISLVER".
19556 2017-02-15 Jakub Jelinek <jakub@redhat.com>
19559 * real.c (real_from_integer): Call real_convert even for decimal.
19561 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
19564 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
19566 2017-02-14 Andrew Pinski <apinski@cavium.com>
19568 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
19569 cores and change the partno/implementer to be correct.
19570 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
19571 the 'B" as the implementer.
19572 * config/aarch64/aarch64-tune.md: Regenerate.
19574 2017-02-14 Carl Love <cel@us.ibm.com>
19576 * config/rs6000/rs6000.c: Add case statement entry to make the
19577 xvcvuxdsp built-in argument unsigned.
19578 * config/rs6000/vsx.md: Fix the source and return operand types so they
19579 match the instruction definitions from the ISA document. Fix typo
19580 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
19583 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
19586 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
19587 member early_clobber_alts.
19588 * lra-lives.c (reg_early_clobber_p): New.
19589 (process_bb_lives): Use it.
19590 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
19591 (debug_operand_data): Initialize early_clobber_alts.
19592 (setup_operand_alternative): Set up early_clobber_alts.
19593 (collect_non_operand_hard_regs): Ditto. Pass early clobber
19594 alternatives to new_insn_reg.
19595 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
19597 (lra_update_insn_regno_info): Pass the new arg.
19599 2017-02-14 Jakub Jelinek <jakub@redhat.com>
19601 PR middle-end/79505
19602 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
19603 (new_oacc_loop_raw): Don't clear already cleared fields.
19606 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
19607 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
19608 _mm512_prefetch_i64gather_ps): New inline functions and macros.
19610 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
19613 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
19615 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
19618 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
19619 the extra instruction to the right place to store 128-bit constant
19622 2017-02-14 Martin Sebor <msebor@redhat.com>
19624 PR middle-end/79448
19625 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
19626 warning for strings of unknown length.
19628 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
19630 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
19632 2017-02-14 Jeff Law <law@redhat.com>
19635 * ira-costs.c (scan_one_insn): Initialize register move costs
19636 for pseudos seen in USE/CLOBBER insns.
19638 PR tree-optimization/79095
19639 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
19640 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
19641 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
19642 if the operands are known to be not equal, then the resulting range
19644 (intersect_ranges): If the new range is ~[0,0] and the old range is
19645 wide, then prefer ~[0,0].
19646 * tree-vrp.c (overflow_comparison_p_1): New function.
19647 (overflow_comparison_p): New function.
19648 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
19649 if NAME is used in an overflow test.
19650 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
19651 overflow check that can be expressed as an equality test, then adjust
19652 ops to be that equality test.
19654 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19656 * config/s390/s390-builtin-types.def: Remove flags argument.
19657 * config/s390/s390.c (s390_init_builtins): Likewise.
19659 2017-02-14 Martin Liska <mliska@suse.cz>
19661 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
19662 vector. Fix trailing white spaces.
19664 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
19666 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
19669 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19671 PR rtl-optimization/68664
19672 * config/arm/arm.c (arm_sched_can_speculate_insn):
19673 New function. Declare prototype.
19674 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
19676 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19678 PR rtl-optimization/68664
19679 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
19681 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
19683 2017-02-14 Amit Pawar <amit.pawar@amd.com>
19685 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
19686 max skip bytes for function, loop and jump.
19688 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19690 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
19691 ABS_EXPR for gimple dump.
19693 2017-02-14 Jakub Jelinek <jakub@redhat.com>
19696 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
19698 PR tree-optimization/79408
19699 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
19700 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
19701 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
19702 also if rhs1 is INTEGER_CST.
19704 2017-02-14 Richard Biener <rguenther@suse.de>
19706 PR middle-end/79432
19707 * tree-into-ssa.c (insert_phi_nodes): When the function can
19708 have abnormal edges rewrite SSA names with broken use-def
19709 dominance out of SSA and register them for PHI insertion.
19711 2017-02-13 Martin Sebor <msebor@redhat.com>
19713 PR middle-end/79496
19714 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
19715 clearing info.nowrite flag when snprintf size argument is a range.
19717 2017-02-13 Jakub Jelinek <jakub@redhat.com>
19719 * cprop.c (cprop_jump): Add missing space in string literal.
19720 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
19721 (get_constraint_for_component_ref): Likewise.
19722 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
19723 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
19724 * lra-constraints.c (process_alt_operands): Likewise.
19725 * ipa-inline.c (inline_small_functions): Likewise.
19726 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
19727 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
19728 * trans-mem.c (diagnose_tm_1_op): Likewise.
19729 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
19730 (grid_parallel_clauses_gridifiable): Likewise.
19732 * config/nvptx/mkoffload.c (process): Add space in between
19735 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
19736 "MOD4_SSE_REGS" and "ALL_REGS".
19738 * spellcheck.c (test_data): Add , in between "foo" and "food".
19740 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19743 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
19744 boundary crossing check and subsequent code generation agree.
19746 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19748 * config/aarch64/aarch64.c (has_memory_op): Delete.
19749 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
19752 2017-02-13 Jakub Jelinek <jakub@redhat.com>
19754 PR rtl-optimization/79388
19755 PR rtl-optimization/79450
19756 * combine.c (distribute_notes): When removing TEM_INSN for which
19757 corresponding dest has last value recorded, invalidate that last
19760 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19762 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
19763 of explicit '@'. Add missing assembly comment marker on branch costs
19766 2017-02-13 Nathan Sidwell <nathan@acm.org>
19768 * gengtype-lex.l (<in_struct>): Add '/'.
19770 2017-02-13 Martin Liska <mliska@suse.cz>
19773 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
19775 2017-02-13 Richard Biener <rguenther@suse.de>
19777 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
19779 * configure: Re-generate.
19780 * config.in: Likewise.
19781 * graphite-dependences.c: Simplify as if
19782 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
19783 * graphite-isl-ast-to-gimple.c: Likewise.
19784 * graphite-optimize-isl.c: Likewise.
19785 * graphite-poly.c: Likewise.
19786 * graphite-sese-to-poly.c: Likewise.
19787 * graphite.h: Likewise.
19788 * toplev.c: Include isl/version.h and use isl_version () for
19789 printing the ISL version.
19790 * doc/install.texi: Update ISL requirement.
19792 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
19794 * doc/standards.texi (Standards): Update reference to
19797 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
19799 * doc/extend.texi (Named Address Spaces): sourceware.org now
19801 * doc/install.texi (Binaries): Ditto.
19804 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
19806 * doc/cpp.texi: Replace "stringify"/"stringification" with C
19807 standard terminology "stringize"/"stringizing" throughout.
19808 * doc/cppinternals.texi: Likewise.
19810 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
19812 * doc/extend.texi: Fix some spelling mistakes and typos.
19813 * doc/invoke.texi: Likewise.
19815 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
19818 * params.def (inline-min-speedup) Change from 10 to 8.
19820 2017-02-11 Jakub Jelinek <jakub@redhat.com>
19822 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
19825 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
19828 * ipa-inline-analysis.c (get_minimal_bb): New function.
19829 (record_modified): Use it.
19830 (remap_edge_change_prob): Handle also ancestor functions.
19832 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
19834 * doc/contrib.texi (Contributors): Remove broken link into
19835 the Mauve CVS repository.
19837 2017-02-11 Jakub Jelinek <jakub@redhat.com>
19839 PR middle-end/79454
19840 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
19841 result computation whenever lhs doesn't have vector mode, not
19842 just when it has BLKmode.
19844 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
19846 * doc/makefile.texi (profiledbootstrap): Refer to the
19847 installation instructions only in textual form.
19849 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19852 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
19854 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
19856 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
19857 (Specific): Update mingw-w64 reference.
19859 (Specific): Remove broken link to Renesas RX processor.
19861 2017-02-10 Richard Biener <rguenther@suse.de>
19863 * toplev.c (process_options): Do not mention obsolete graphite
19864 options when printing sorry message about missing graphite support.
19865 Mention -floop-nest-optimize.
19867 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
19869 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
19870 (vtst_p16): Likewise.
19871 (vtstq_p8): Likewise.
19872 (vtstq_p16): Likewise.
19874 (vtstq_p64): Likewise.
19875 * config/arm/arm_neon.h (vgetq_lane_p64): New.
19876 (vset_lane_p64): New.
19877 (vsetq_lane_p64): New.
19879 2017-02-10 Jakub Jelinek <jakub@redhat.com>
19881 PR tree-optimization/79411
19882 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
19883 stmt operands are SSA_NAMEs used in abnormal phis.
19884 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
19887 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
19890 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
19893 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
19895 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
19897 2017-02-09 Jakub Jelinek <jakub@redhat.com>
19899 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
19903 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
19904 not arbitrary TREE_CONSTANT.
19907 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
19908 "omp declare target link" attribute unless is_global_var.
19909 * omp-offload.c (find_link_var_op): Likewise.
19911 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
19912 Chung-Lin Tang <cltang@codesourcery.com>
19914 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
19916 (gimplify_adjust_omp_clauses): Don't delete TILE.
19917 (gimplify_omp_for): Deal with TILE.
19918 * internal-fn.c (expand_GOACC_TILE): New function.
19919 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
19921 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
19922 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
19924 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
19925 avoid DIV for outermost collapse var.
19926 (expand_oacc_for): Insert tile element loop as needed. Adjust.
19927 Remove out of date comments, fix whitespace.
19928 * omp-general.c (omp_extract_for_data): Deal with tiling.
19929 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
19930 adjust OLF_DIM_BASE value.
19931 (struct omp_for_data): Add tiling field.
19932 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
19933 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
19934 for auto loops. Remove default auto determining, moved to
19935 oacc_loop_fixed_partitions.
19936 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
19937 stmts, add e_mask field.
19938 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
19939 (oacc_thread_numbers): Use oacc_dim_call.
19940 (oacc_xform_tile): New.
19941 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
19942 (finish_oacc_loop): Adjust for ifns vector.
19943 (oacc_loop_discover_walk): Append loop abstraction sites to list,
19944 add case for GOACC_TILE fns.
19945 (oacc_loop_xform_loop): Delete.
19946 (oacc_loop_process): Iterate over call list directly, and add
19947 handling for GOACC_TILE fns.
19948 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
19950 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
19951 vector partitioning to outer loops. Assign 2 partitions to loops
19952 when available. Add TILE handling.
19953 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
19954 (execite_oacc_device_lower): Process GOACC_TILE fns,
19955 ignore unknown specs.
19956 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
19957 * tree.c (omp_clause_num_ops): Adjust TILE ops.
19958 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
19960 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
19962 * configure.ac (ACX_BUGURL): Update.
19963 * configure: Regenerate.
19965 2017-02-09 Richard Biener <rguenther@suse.de>
19967 PR tree-optimization/69823
19968 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
19969 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
19971 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
19973 * config/arc/arc-c.def: Add __NPS400__ definition.
19974 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
19975 (TARGET_NPS400): Define.
19977 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
19979 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
19981 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
19982 pointer, arch_info.
19983 (arc_cpu_types): Fill the arch_info field with a pointer into the
19984 arc_arch_types table.
19985 (arc_selected_cpu): Declare.
19986 * config/arc/arc.c (arc_selected_cpu): Make global.
19987 (arc_selected_arch): Delete.
19988 (arc_base_cpu): Delete.
19989 (arc_override_options): Remove references to deleted variables,
19990 update access to arch information.
19991 (ARC_OPT): Update access to arch information.
19992 (ARC_OPTX): Likewise.
19993 * config/arc/arc.h (arc_base_cpu): Remove declaration.
19994 (TARGET_ARC600): Update access to arch information.
19995 (TARGET_ARC601): Likewise.
19996 (TARGET_ARC700): Likewise.
19997 (TARGET_EM): Likewise.
19998 (TARGET_HS): Likewise.
19999 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
20002 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
20005 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
20006 condition/operands for integer GE/LE/GEU/LEU operations.
20008 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
20010 PR translation/79397
20011 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
20014 2017-02-08 Martin Jambor <mjambor@suse.cz>
20017 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
20018 whether allocation happened.
20019 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
20020 nothing was allocated.
20022 2017-02-08 Jakub Jelinek <jakub@redhat.com>
20024 PR tree-optimization/79408
20025 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
20026 constant, but SSA_NAME with a known integer range, use the minimum
20027 of that range instead of op1 to determine if modulo can be replaced
20028 with its first operand.
20030 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20032 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
20034 2017-02-08 Richard Biener <rguenther@suse.de>
20036 PR tree-optimization/71824
20037 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
20038 Check all loops contained in the merged region.
20040 2017-02-07 Andrew Pinski <apinski@cavium.com>
20042 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
20044 2017-02-07 Andrew Pinski <apinski@cavium.com>
20046 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
20047 (thunderxt88): Likewise.
20048 (thunderxt81): Disable LSE and change v8.1 to v8.
20049 (thunderxt83): Likewise.
20051 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20052 Richard Biener <rguenther@suse.de>
20054 PR middle-end/79399
20055 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
20056 type from int to size_t.
20057 * ira-costs.c (struct_costs_size): Change type from int to size_t.
20059 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20061 PR rtl-optimization/79386
20062 * cprop.c (bypass_conditional_jumps): Initialize
20063 bypass_last_basic_block already before splitting bbs after
20064 unconditional traps...
20065 (bypass_conditional_jumps): ... rather than here.
20068 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
20069 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
20070 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
20071 fix -masm=intel patterns.
20073 2017-02-07 Richard Biener <rguenther@suse.de>
20075 PR tree-optimization/79256
20076 PR middle-end/79278
20077 * builtins.c (get_object_alignment_2): Use min_align_of_type
20078 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
20079 and ADJUST_FIELD_ALIGN.
20081 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
20083 * doc/tm.texi: Regenerate.
20084 * stor-layout.c (layout_decl): Adjust.
20085 (update_alignment_for_field): Likewise.
20086 (place_field): Likewise.
20087 (min_align_of_type): Likewise.
20088 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
20089 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
20090 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
20091 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
20092 * config/frv/frv.c (frv_adjust_field_align): Likewise.
20093 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
20094 * config/i386/i386.c (x86_field_alignment): Likewise.
20095 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
20096 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
20097 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
20098 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
20099 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
20100 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
20104 2017-01-30 Richard Biener <rguenther@suse.de>
20106 PR tree-optimization/79256
20107 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
20108 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
20111 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
20113 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
20114 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
20115 builtins to SImode and emit a zero-extend, if necessary.
20117 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20119 * docs/invoke.texi (RISC-V Options): Alphabetize.
20121 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20123 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
20126 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20128 * config/riscv/riscv.c: New file.
20129 * common/config/riscv/riscv-common.c: Likewise.
20130 * config.gcc: Likewise.
20131 * config/riscv/constraints.md: Likewise.
20132 * config/riscv/elf.h: Likewise.
20133 * config/riscv/generic.md: Likewise.
20134 * config/riscv/linux.h: Likewise.
20135 * config/riscv/multilib-generator: Likewise.
20136 * config/riscv/peephole.md: Likewise.
20137 * config/riscv/pic.md: Likewise.
20138 * config/riscv/predicates.md: Likewise.
20139 * config/riscv/riscv-builtins.c: Likewise.
20140 * config/riscv/riscv-c.c: Likewise.
20141 * config/riscv/riscv-ftypes.def: Likewise.
20142 * config/riscv/riscv-modes.def: Likewise.
20143 * config/riscv/riscv-opts.h: Likewise.
20144 * config/riscv/riscv-protos.h: Likewise.
20145 * config/riscv/riscv.h: Likewise.
20146 * config/riscv/riscv.md: Likewise.
20147 * config/riscv/riscv.opt: Likewise.
20148 * config/riscv/sync.md: Likewise.
20149 * config/riscv/t-elf-multilib: Likewise.
20150 * config/riscv/t-linux: Likewise.
20151 * config/riscv/t-linux-multilib: Likewise.
20152 * config/riscv/t-riscv: Likewise.
20153 * configure.ac: Likewise.
20154 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
20155 Waterman as RISC-V maintainers.
20156 * doc/install.texi: Add RISC-V entries.
20157 * doc/invoke.texi: Add RISC-V options section.
20158 * doc/md.texi: Add RISC-V constraints section.
20159 * configure: Regenerated.
20161 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
20164 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
20165 false values to be constant vectors with all 0 or all 1 bits set.
20166 (vcondu<mode><mode>): Likewise.
20167 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
20169 (fpmask_comparison_operator): Update comment.
20170 (vecint_comparison_operator): New predicate.
20171 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
20172 vector conditionals when the true and false values are constant
20173 vectors with all 0 bits or all 1 bits set.
20175 2017-02-06 Martin Sebor <msebor@redhat.com>
20177 PR tree-optimization/79376
20178 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
20180 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
20182 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
20183 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
20184 to simplify split condition.
20186 2017-02-06 Jakub Jelinek <jakub@redhat.com>
20188 * omp-expand.c (oxpand_omp_atomic_fetch_op,
20189 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
20192 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
20194 PR rtl-optimization/68664
20195 * target.def (can_speculate_insn): New hook.
20196 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
20197 * doc/tm.texi: Regenerate.
20198 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
20199 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
20200 (rs6000_sched_can_speculate_insn): New function.
20202 2017-02-06 Jakub Jelinek <jakub@redhat.com>
20204 PR tree-optimization/79284
20205 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
20206 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
20207 vectorizable_mask_load_store, vectorizable_operation,
20208 vect_is_simple_cond, get_same_sized_vectype): Use it instead
20209 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
20210 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
20211 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
20212 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
20213 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
20214 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
20215 is_gimple_assign (stmt). Replace another such test with
20216 is_gimple_assign (stmt).
20218 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
20221 * config/avr/avr.c (rtl-iter.h): Include it.
20222 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
20223 (avr_legitimate_combined_insn): ...and implementation.
20225 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
20227 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
20228 * config/s390/s390.c (s390_const_operand_ok)
20229 (s390_canonicalize_comparison, s390_extract_part)
20230 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
20231 (s390_contiguous_bitmask_p, s390_rtx_costs)
20232 (legitimize_pic_address): Likewise.
20233 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
20234 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
20235 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
20236 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
20237 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
20239 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
20241 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
20242 REGNO($0) == REGNO($1).
20244 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20246 * config/s390/linux.h(SIZE_TYPE): Add comment.
20248 2017-02-06 Julian Brown <julian@codesourcery.com>
20249 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20250 Virendra Pathak <virendra.pathak@broadcom.com>
20252 * config/aarch64/aarch64-cores.def: Change the scheduler
20254 * config/aarch64/aarch64.md: Include thunderx2t99.md.
20255 * config/aarch64/thunderx2t99.md: New file.
20257 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20259 * doc/standards.texi (Go Language): Update link to language
20262 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20264 * tree-eh.c (lower_resx): Sanitize profile.
20265 (cleanup_empty_eh_move_lp): Likewise.
20267 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20270 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
20272 * cfgloopmanip.h (loop_version): Update prototype.
20273 * modulo-sched.c (sms_schedule): Update call of loop_version.
20274 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
20275 * tree-parloops.c (gen_parallel_loop): Likewise.
20276 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
20277 * tree-ssa-loop-split.c (split_loop): Likewise.
20278 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
20279 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
20281 2017-02-05 Martin Liska <mliska@suse.cz>
20284 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
20286 (print_operand_address): Initialize a struct to zero.
20288 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20290 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
20291 garbage collector only in textual form.
20293 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20295 * doc/extend.texi (x86 specific memory model extensions for
20296 transactional memory): Simplify a phrase.
20298 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
20301 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
20302 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
20303 (atomic_storedi_1): Likewise.
20305 2017-02-04 Jakub Jelinek <jakub@redhat.com>
20307 PR tree-optimization/79338
20308 * tree-parloops.c (gather_scalar_reductions): Don't call
20309 vect_analyze_loop_form for loop->inner before destroying loop's
20312 2017-02-03 Martin Sebor <msebor@redhat.com>
20314 PR tree-optimization/79327
20315 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
20316 when precision has resulted in leading zeros.
20317 (format_integer): Adjust the likely counter to assume an unknown
20318 argument that may be zero is non-zero.
20320 2017-02-03 Jason Merrill <jason@redhat.com>
20323 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
20324 avoid copying non-taken branch.
20326 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20328 PR tree-optimization/79340
20329 * tree-vect-loop.c (vectorizable_reduction): Release
20330 vec_defs elements after safe_splicing them into other vectors.
20333 PR tree-optimization/79327
20334 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
20335 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
20337 (format_integer): Use wide_int_to_tree instead of build_int_cst
20338 + to_?hwi. If argmin is NULL, just set argmin and argmax to
20339 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
20340 of shortest and longest sequence.
20342 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
20344 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
20345 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
20347 2017-02-03 Walter Lee <walt@tilera.com>
20350 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
20351 after initial stackframe link reg save.
20352 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
20354 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20357 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
20358 wu for stxssp alternative.
20360 2017-02-03 Martin Sebor <msebor@redhat.com>
20362 PR tree-optimization/79352
20363 * gimple-fold.c (get_range_strlen): Add argument.
20364 (get_range_strlen): Change return type to bool.
20365 (get_maxval_strlen): Pass in a dummy argument.
20366 * gimple-fold.h (get_range_strlen): Change return type to bool.
20367 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
20368 * tree.h (array_at_struct_end_p): Add argument.
20369 * tree.c (array_at_struct_end_p): Handle it.
20371 2017-02-03 Martin Liska <mliska@suse.cz>
20374 * multiple_target.c (create_dispatcher_calls): Redirect edge
20375 from a caller of a dispatcher.
20376 (expand_target_clones): Make the clones local.
20377 (ipa_target_clone): Do both target clones and resolvers.
20378 (ipa_dispatcher_calls): Remove the pass.
20379 (pass_dispatcher_calls::gate): Likewise.
20380 (make_pass_dispatcher_calls): Likewise.
20381 * passes.def (pass_target_clone): Put as very first IPA early
20384 2017-02-03 Martin Liska <mliska@suse.cz>
20386 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
20387 in case of a function with ifunc attribute.
20389 2017-02-03 Martin Liska <mliska@suse.cz>
20391 * cgraph.c (cgraph_node::dump): Dump function version info.
20392 * symtab.c (symtab_node::dump_base): Add missing new line.
20394 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20396 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
20397 (ifcombine_ifandif): Use it.
20399 2017-02-03 Martin Liska <mliska@suse.cz>
20401 * doc/invoke.texi: Document default value for
20402 use-after-scope-direct-emission-threshold.
20404 2017-02-03 Martin Liska <mliska@suse.cz>
20406 PR tree-optimization/79339
20407 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
20408 (format_floating): Likewise.
20410 2017-02-03 Martin Liska <mliska@suse.cz>
20413 * ipa-prop.c (ipa_node_params_t::insert): Remove current
20415 (ipa_node_params_t::remove): Likewise.
20416 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
20417 initialization from removed ipa_node_params_t::insert.
20418 (ipa_node_params::~ipa_node_params): Move from removed
20419 ipa_node_params_t::release.
20420 * symbol-summary.h (symbol_summary::m_released): New member.
20421 Do not release a summary twice. Do not allow to call finalizer
20422 for types of a summary that live in GGC memory.
20424 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20426 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
20429 2017-02-02 Martin Sebor <msebor@redhat.com>
20431 PR middle-end/79275
20432 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
20433 (format_string): Tighten up the range of output for non-constant
20434 strings and correct the expected range for wide non-constant strings.
20436 2017-02-02 Martin Sebor <msebor@redhat.com>
20438 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
20440 PR middle-end/32003
20441 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
20443 (-fdump-tree-@var): Add to index and document how to come up
20444 with pass-specific option and dump file names.
20445 (-fdump-passes): Clarify where to look for output.
20447 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20449 PR middle-end/77445
20450 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
20451 statistics of the analyzed path; allow threading for speed when
20452 any of BBs along the path are optimized for speed.
20454 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
20456 PR middle-end/78468
20457 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
20458 settings of the virtual registers.
20461 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
20463 * explow.c (get_dynamic_stack_size): Take known alignment of stack
20464 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
20467 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20469 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
20470 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
20472 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20474 * config/s390/s390.md: Add missing comments with the expanded
20476 * config/s390/vector.md: Likewise.
20477 * config/s390/vx-builtins.md: Likewise.
20479 2017-02-02 Jakub Jelinek <jakub@redhat.com>
20482 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
20483 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
20484 conditions on a single line.
20486 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20488 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
20489 __S390_VX__ to __VX__.
20491 2017-02-01 Andrew Pinski <apinski@cavium.com>
20493 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
20494 stmt_info to record_stmt_cost.
20495 (vect_get_known_peeling_cost): Pass stmt_info if known to
20497 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
20498 cpu_vector_cost field into
20499 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
20500 field into vec_int_stmt_cost and vec_fp_stmt_cost.
20501 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
20502 splitting of scalar_stmt_cost and vec_stmt_cost.
20503 (thunderx_vector_cost): Likewise.
20504 (cortexa57_vector_cost): LIkewise.
20505 (exynosm1_vector_cost): Likewise.
20506 (xgene1_vector_cost): Likewise.
20507 (thunderx2t99_vector_cost): Improve after the splitting of the two
20509 (aarch64_builtin_vectorization_cost): Update for the splitting of
20510 scalar_stmt_cost and vec_stmt_cost.
20512 2017-02-01 Torvald Riegel <triegel@redhat.com>
20513 Richard Henderson <rth@redhat.com>
20515 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
20516 conditional on existance of a fast atomic load.
20517 * optabs-query.c (can_atomic_load_p): New function.
20518 * optabs-query.h (can_atomic_load_p): Declare it.
20519 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
20520 no fast atomic load is available for the particular size of access.
20521 (expand_atomic_compare_and_swap): Likewise.
20522 (expand_atomic_load): Likewise.
20523 (expand_atomic_store): Likewise.
20524 (expand_atomic_fetch_op): Likewise.
20525 * testsuite/lib/target-supports.exp
20526 (check_effective_target_sync_int_128): Remove x86 because it provides
20527 no fast atomic load.
20528 (check_effective_target_sync_int_128_runtime): Likewise.
20530 2017-02-01 Richard Biener <rguenther@suse.de>
20532 * graphite.c: Include tree-vectorizer.h for find_loop_location.
20533 (graphite_transform_loops): Provide opt-info for optimized nests.
20534 * tree-parloop.c (parallelize_loops): Provide opt-info for
20535 parallelized loops.
20537 2017-02-01 Richard Biener <rguenther@suse.de>
20539 PR middle-end/79315
20540 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
20541 was not set before.
20543 2017-02-01 Richard Biener <rguenther@suse.de>
20545 PR tree-optimization/71824
20546 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
20547 Verify the loops are valid in the merged SESE region.
20548 (scop_detection::can_represent_loop_1): Check analyzing the
20549 evolution of the number of iterations in the region succeeds.
20551 2017-01-31 Ian Lance Taylor <iant@golang.org>
20553 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
20554 REG_ARGS_SIZE note to 32-bit push insns and call insn.
20556 2017-01-31 David Malcolm <dmalcolm@redhat.com>
20558 PR preprocessor/79210
20559 * input.c (get_substring_ranges_for_loc): Replace line_width
20560 assertion with error-handling.
20562 2017-01-31 Richard Biener <rguenther@suse.de>
20564 PR tree-optimization/77318
20565 * graphite-sese-to-poly.c (extract_affine): Fix assert.
20566 (create_pw_aff_from_tree): Take loop parameter.
20567 (add_condition_to_pbb): Pass loop of the condition to
20568 create_pw_aff_from_tree.
20570 2017-01-31 Jakub Jelinek <jakub@redhat.com>
20572 * config/s390/s390.c (s390_asan_shadow_offset): New function.
20573 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
20575 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
20579 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
20581 (convert_int_to_float128): Likewise.
20582 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
20583 (convert_int_to_float128): Likewise.
20584 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
20585 (UNSPEC_IEEE128_CONVERT): Likewise.
20586 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
20587 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
20588 Use local variables for IBM extended format.
20589 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
20590 (fix_trunc<mode>si2_fprs): Likewise.
20591 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
20592 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
20593 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
20594 to know that we can now have integers of all sizes in vector
20596 (fix<uns>_<mode>di2_hw): Likewise.
20597 (float<uns>_<mode>si2_hw): Likewise.
20598 (fix_<mode>si2_hw): Likewise.
20599 (fixuns_<mode>si2_hw): Likewise.
20600 (float<uns>_<mode>di2_hw): Likewise.
20601 (float_<mode>di2_hw): Likewise.
20602 (float_<mode>si2_hw): Likewise.
20603 (floatuns_<mode>di2_hw): Likewise.
20604 (floatuns_<mode>si2_hw): Likewise.
20605 (xscvqp<su>wz_<mode>): Delete, no longer used.
20606 (xscvqp<su>dz_<mode>): Likewise.
20607 (xscv<su>dqp_<mode>): Likewise.
20608 (ieee128_mfvsrd_64bit): Likewise.
20609 (ieee128_mfvsrd_32bit): Likewise.
20610 (ieee128_mfvsrwz): Likewise.
20611 (ieee128_mtvsrw): Likewise.
20612 (ieee128_mtvsrd_64bit): Likewise.
20613 (ieee128_mtvsrd_32bit): Likewise.
20615 2017-01-31 Martin Liska <mliska@suse.cz>
20618 * ipa-prop.c (ipa_free_all_node_params): Call release method
20619 instead of ~sumbol_summary to not to trigger double times
20622 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
20624 PR tree-optimization/71691
20625 * bitmap.h (class auto_bitmap): New.
20626 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
20627 is_maybe_undefined instead of ssa_undefined_value_p.
20629 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20631 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
20632 __S390_ARCH_LEVEL__ to __ARCH__.
20634 2017-01-31 Jakub Jelinek <jakub@redhat.com>
20636 PR tree-optimization/79267
20637 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
20638 if should_remove_lhs_p is true.
20640 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
20643 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
20644 (add_alignment_attribute): New.
20645 (base_type_die): Add alignment attribute.
20646 (subrange_type_die): Likewise.
20647 (modified_type_die): Likewise.
20648 (gen_array_type_die): Likewise.
20649 (gen_descr_array_type_die: Likewise.
20650 (gen_enumeration_type_die): Likewise.
20651 (gen_subprogram_die): Likewise.
20652 (gen_variable_die): Likewise.
20653 (gen_field_die): Likewise.
20654 (gen_ptr_to_mbr_type_die): Likewise.
20655 (gen_struct_or_union_type_die): Likewise.
20656 (gen_subroutine_type_die): Likewise.
20657 (gen_typedef_die): Likewise.
20658 (base_type_cmp): Compare alignment attribute.
20660 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20663 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
20664 (setb_unsigned) New pattern for setb with CCUNS.
20665 * config/rs6000/rs6000.c (expand_block_compare): Use a different
20666 subfc./subfe sequence to avoid overflow problems. Generate a
20667 shorter sequence with cmpld/setb for power9.
20668 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
20669 for generating subfc. instruction.
20670 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
20671 now uses this instruction.
20673 2017-01-30 Ian Lance Taylor <iant@google.com>
20676 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
20677 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
20679 2017-01-30 Martin Sebor <msebor@redhat.com>
20681 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
20682 Move constant to the right of a relational operator.
20683 (get_mpfr_format_length, format_character, format_string): Ditto.
20684 (should_warn_p, maybe_warn): Same.
20686 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
20688 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
20691 * asan.c (get_translation_unit_decl): Remove function.
20692 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
20694 2017-01-30 Martin Liska <mliska@suse.cz>
20696 PR gcov-profile/79259
20697 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
20698 -fprofile-generate.
20700 2017-01-30 Martin Liska <mliska@suse.cz>
20703 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
20704 Initialize variables with NULL value.
20706 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
20709 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
20711 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
20713 2017-01-30 Richard Biener <rguenther@suse.de>
20715 PR tree-optimization/79276
20716 * tree-vrp.c (process_assert_insertions): Properly adjust common
20717 when removing a duplicate.
20719 * gcc.dg/torture/pr79276.c: New testcase.
20721 2017-01-30 Richard Biener <rguenther@suse.de>
20723 PR tree-optimization/79256
20724 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
20725 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
20727 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
20729 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
20732 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
20733 ("*r<noxa>sbg_<mode>_sll_bitmask")
20734 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
20735 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
20736 Use contiguous_bitmask_nowrap_operand.
20738 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20741 * config/rs6000/altivec.h (vec_xl): Revise #define.
20742 (vec_xst): Likewise.
20744 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
20746 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
20748 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
20750 PR rtl-optimization/79194
20751 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
20752 traps before call to bypass_conditional_jumps.
20754 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
20756 PR tree-optimization/71374
20757 * lra-constraints.c (check_conflict_input_operands): New.
20758 (match_reload): Use it.
20760 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
20763 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
20764 account to calculate conflict_set.
20766 2017-01-27 Bin Cheng <bin.cheng@arm.com>
20768 PR rtl-optimization/78559
20769 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
20770 other_insn in combine.
20772 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20774 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
20775 uint16_type_node for BT_UINT16.
20777 2017-01-27 David Malcolm <dmalcolm@redhat.com>
20779 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
20780 "RTL Tests" to menu.
20781 (GIMPLE Tests): New node.
20782 (RTL Tests): New node.
20784 2017-01-27 Richard Biener <rguenther@suse.de>
20786 PR tree-optimization/79245
20787 * tree-loop-distribution.c (distribute_loop): Apply cost
20788 modeling also to detected patterns.
20790 2017-01-27 Richard Biener <rguenther@suse.de>
20792 PR tree-optimization/71433
20793 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
20794 (compare_assert_loc): New function.
20795 (process_assert_insertions): Sort and optimize assert locations
20796 to remove duplicates and push down identical assertions on
20797 edges to their destination block.
20799 2017-01-27 Richard Biener <rguenther@suse.de>
20801 PR tree-optimization/79244
20802 * tree-vrp.c (remove_range_assertions): Forcefully propagate
20803 out SSA names even if abnormal.
20805 2017-01-27 Jakub Jelinek <jakub@redhat.com>
20807 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
20808 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
20809 instead of MPFR_RNDN.
20811 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
20814 * arm.c (arm_option_override): Don't call build_target_option_node
20815 until after doing all option overrides.
20816 (arm_valid_target_attribute_tree): Likewise.
20818 2017-01-27 Martin Liska <mliska@suse.cz>
20820 * doc/invoke.texi (-fprofile-arcs): Document profiling support
20821 for {cd}tors and C++ {cd}tors.
20823 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
20825 * config/s390/s390.md ("*setmem_long_and")
20826 ("*setmem_long_and_31z"): Use zero_extend instead of and.
20828 2017-01-26 Martin Sebor <msebor@redhat.com>
20830 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
20833 2017-01-26 Martin Sebor <msebor@redhat.com>
20835 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
20836 HAVE_DFmode before using XFmode or DFmode.
20837 (parse_directive): Avoid using the z length modifier to avoid
20838 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
20840 PR middle-end/78703
20841 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
20842 to accept adjustment as an array.
20843 (get_int_range): New function.
20844 (struct directive): Make width and prec arrays.
20845 (directive::set_width, directive::set_precision): Call get_int_range.
20846 (format_integer, format_floating): Handle width and precision ranges.
20847 (format_string, parse_directive): Same.
20849 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20852 * dwarf2out.c (generate_skeleton_bottom_up): For children with
20853 comdat_type_p set, just clone them, but keep the children in the
20857 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
20858 which have direct callers with -fvar-tracking-assignments enabled
20860 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
20861 inside of type units.
20863 2017-01-26 Martin Sebor <msebor@redhat.com>
20865 PR middle-end/78703
20866 * gimple-ssa-sprintf.c (struct result_range): Add likely and
20868 (struct format_result): Replace number_chars, number_chars_min,
20869 and number_chars_max with a single member of struct result_range.
20871 (format_result::operator+=): Adjust.
20872 (struct fmtresult): Remove bounded. Handle likely and unlikely
20874 (fmtresult::adjust_for_width_or_precision): New function.
20875 (fmtresult:type_max_digits): New function.
20876 (bytes_remaining): Handle likely and unlikely counters.
20877 (min_bytes_remaining): Remove.
20878 (format_percent): Simplify.
20879 (format_integer, format_floating): Set likely and unlikely counters.
20880 (get_string_length, format_character, format_string): Same.
20881 (format_plain, should_warn_p): New function.
20882 (maybe_warn): Call should_warn_p. Update diagnostic messages
20883 and handle those for all directives, including plain strings.
20884 (format_directive): Handle likely and unlikely counters.
20885 Remove unnecessary quoting from diagnostics. Add an informational
20887 (add_bytes): Remove.
20888 (pass_sprintf_length::compute_format_length): Simplify.
20889 (try_substitute_return_value): Handle likely and unlikely counters.
20891 2017-01-26 Carl Love <cel@us.ibm.com>
20893 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
20894 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
20896 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
20899 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
20900 endianess for subregs into account.
20901 * lra-constraints.c (lra_constraints): Do risky transformations
20902 always on the first iteration.
20903 * lra-lives.c (check_pseudos_live_through_calls): Add arg
20904 last_call_used_reg_set.
20905 (process_bb_lives): Define and use last_call_used_reg_set.
20906 * lra.c (lra): Always continue after lra_constraints on the first
20909 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
20911 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
20913 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
20915 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20917 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
20918 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
20919 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
20920 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
20921 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
20922 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
20923 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
20924 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
20925 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
20927 2017-01-26 Marek Polacek <polacek@redhat.com>
20930 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
20931 for the third operand.
20933 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20935 PR middle-end/79236
20936 * omp-low.c (struct omp_context): Add simt_stmt field.
20937 (scan_omp_for): Return omp_context *.
20938 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
20939 context to the _simt_ SIMD stmt.
20940 (lower_omp_for): For combined SIMD with sibling _simt_
20941 SIMD, make sure to use the same decls in _looptemp_
20942 clauses as in the sibling.
20944 2017-01-26 David Sherwood <david.sherwood@arm.com>
20946 PR middle-end/79212
20947 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
20950 2017-01-26 Jakub Jelinek <jakub@redhat.com>
20953 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
20954 emit fld b; fld a; if possible.
20956 * brig-builtins.def: Update copyright years.
20957 * config/arm/arm_acle_builtins.def: Update copyright years.
20959 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
20962 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
20963 constraint instead of o for the stxsd instruction.
20965 2017-01-25 Carl Love <cel@us.ibm.com>
20967 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
20968 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
20970 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
20972 * doc/invoke.texi (C++ Dialect Options): Fix typo.
20974 2017-01-25 Richard Biener <rguenther@suse.de>
20976 PR tree-optimization/69264
20977 * target.def (vector_alignment_reachable): Improve documentation.
20978 * doc/tm.texi: Regenerate.
20979 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
20981 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
20982 earlier changes with respect to TYPE_USER_ALIGN.
20983 (vector_alignment_reachable_p): Likewise. Improve dumping.
20985 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20988 * config/arm/arm.md (xordi3): Force constant operand into a register
20991 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20993 * doc/invoke.texi (-fstore-merging): Correct default optimization
20994 levels at which it is enabled.
20995 (-O): Move -fstore-merging from list to...
20998 2017-01-25 Richard Biener <rguenther@suse.de>
21001 * omp-expand.c: Include debug.h.
21002 (expand_omp_taskreg): Make sure to generate early debug before
21003 outlining anything from a function.
21004 (expand_omp_target): Likewise.
21005 (grid_expand_target_grid_body): Likewise.
21007 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
21010 * asan.c (get_translation_unit_decl): New function.
21011 (asan_add_global): Extract modules file name from globals
21012 TRANSLATION_UNIT_DECL name.
21014 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
21017 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
21018 for long calls with APCS frame and VFP.
21020 2017-01-24 David Malcolm <dmalcolm@redhat.com>
21022 * cfg.c (original_copy_tables_initialized_p): New function.
21023 * cfg.h (original_copy_tables_initialized_p): New decl.
21024 * cfgrtl.c (relink_block_chain): Guard the call to
21025 free_original_copy_tables with a call to
21026 original_copy_tables_initialized_p.
21027 * cgraph.h (symtab_node::native_rtl_p): New decl.
21028 * cgraphunit.c (symtab_node::native_rtl_p): New function.
21029 (symtab_node::needed_p): Don't assert for early assembly output
21030 for __RTL functions.
21031 (cgraph_node::finalize_function): Set "force_output" for __RTL
21033 (cgraph_node::analyze): Bail out early for __RTL functions.
21034 (analyze_functions): Update assertion to support __RTL functions.
21035 (cgraph_node::expand): Bail out early for __RTL functions.
21036 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
21038 * function.h (struct function): Update comment for field
21040 * gimple-expr.c: Include "tree-pass.h".
21041 (gimple_has_body_p): Return false for __RTL functions.
21042 * Makefile.in (OBJS): Add run-rtl-passes.o.
21043 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
21045 (gcc::pass_manager::get_clean_slate): New accessor.
21046 * passes.c: Include "insn-addr.h".
21047 (should_skip_pass_p): Add logging. Update logic for running
21048 "expand" to be compatible with both __GIMPLE and __RTL. Guard
21049 property-provider override so it is only done for gimple passes.
21051 (skip_pass): New function.
21052 (execute_one_pass): Call skip_pass when skipping passes.
21053 * read-md.c (md_reader::read_char): Support filtering
21054 the input to a subset of line numbers.
21055 (md_reader::md_reader): Initialize fields
21056 m_first_line and m_last_line.
21057 (md_reader::read_file_fragment): New function.
21058 * read-md.h (md_reader::read_file_fragment): New decl.
21059 (md_reader::m_first_line): New field.
21060 (md_reader::m_last_line): New field.
21061 * read-rtl-function.c (function_reader::create_function): Only
21062 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
21063 curr_properties. Set DECL_INITIAL to a dummy block.
21064 (read_rtl_function_body_from_file_range): New function.
21065 * read-rtl-function.h (read_rtl_function_body_from_file_range):
21067 * run-rtl-passes.c: New file.
21068 * run-rtl-passes.h: New file.
21070 2017-01-24 Jeff Law <law@redhat.com>
21072 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
21075 2017-01-24 Bin Cheng <bin.cheng@arm.com>
21077 PR tree-optimization/79159
21078 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
21079 (record_nonwrapping_iv): Improve boundary using above function if no
21080 value range information.
21082 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
21083 Martin Jambor <mjambor@suse.cz>
21085 * brig-builtins.def: New file.
21086 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
21087 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
21088 (DEF_HSAIL_SAT_BUILTIN): Likewise.
21089 (DEF_HSAIL_INTR_BUILTIN): Likewise.
21090 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
21091 * builtin-types.def (BT_INT8): New.
21092 (BT_INT16): Likewise.
21093 (BT_UINT8): Likewise.
21094 (BT_UINT16): Likewise.
21095 (BT_FN_ULONG): Likewise.
21096 (BT_FN_UINT_INT): Likewise.
21097 (BT_FN_UINT_ULONG): Likewise.
21098 (BT_FN_UINT_LONG): Likewise.
21099 (BT_FN_UINT_PTR): Likewise.
21100 (BT_FN_ULONG_PTR): Likewise.
21101 (BT_FN_INT8_FLOAT): Likewise.
21102 (BT_FN_INT16_FLOAT): Likewise.
21103 (BT_FN_UINT32_FLOAT): Likewise.
21104 (BT_FN_UINT16_FLOAT): Likewise.
21105 (BT_FN_UINT8_FLOAT): Likewise.
21106 (BT_FN_UINT64_FLOAT): Likewise.
21107 (BT_FN_UINT16_UINT32): Likewise.
21108 (BT_FN_UINT32_UINT16): Likewise.
21109 (BT_FN_UINT16_UINT16_UINT16): Likewise.
21110 (BT_FN_INT_PTR_INT): Likewise.
21111 (BT_FN_UINT_PTR_UINT): Likewise.
21112 (BT_FN_LONG_PTR_LONG): Likewise.
21113 (BT_FN_ULONG_PTR_ULONG): Likewise.
21114 (BT_FN_VOID_UINT64_UINT64): Likewise.
21115 (BT_FN_UINT8_UINT8_UINT8): Likewise.
21116 (BT_FN_INT8_INT8_INT8): Likewise.
21117 (BT_FN_INT16_INT16_INT16): Likewise.
21118 (BT_FN_INT_INT_INT): Likewise.
21119 (BT_FN_UINT_FLOAT_UINT): Likewise.
21120 (BT_FN_FLOAT_UINT_UINT): Likewise.
21121 (BT_FN_ULONG_UINT_UINT): Likewise.
21122 (BT_FN_ULONG_UINT_PTR): Likewise.
21123 (BT_FN_ULONG_ULONG_ULONG): Likewise.
21124 (BT_FN_UINT_UINT_UINT): Likewise.
21125 (BT_FN_VOID_UINT_PTR): Likewise.
21126 (BT_FN_UINT_UINT_PTR: Likewise.
21127 (BT_FN_UINT32_UINT64_PTR): Likewise.
21128 (BT_FN_INT_INT_UINT_UINT): Likewise.
21129 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
21130 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
21131 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
21132 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
21133 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
21134 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
21135 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
21136 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
21137 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
21138 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
21139 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
21140 * doc/frontends.texi: List BRIG FE.
21141 * doc/install.texi (Testing): Add BRIG tesring requirements.
21142 * doc/invoke.texi (Overall Options): Mention BRIG.
21143 * doc/standards.texi (Standards): Doucment BRIG HSA version.
21145 2017-01-24 Richard Biener <rguenther@suse.de>
21147 PR translation/79208
21148 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
21150 2017-01-24 Martin Jambor <mjambor@suse.cz>
21153 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
21154 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
21155 and known_contexts.
21157 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
21159 PR middle-end/79123
21160 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
21161 casts from signed to unsigned really don't have a range.
21163 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
21165 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
21166 GMP_RNDx for compatiblity.
21168 2017-01-24 Martin Liska <mliska@suse.cz>
21171 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
21172 that would prevent us to call alloca with -1 as argument.
21174 2017-01-24 Jakub Jelinek <jakub@redhat.com>
21176 * dwarf2out.c (output_compilation_unit_header, output_file_names):
21177 Avoid -Wformat-security warning.
21179 2017-01-23 Andrew Pinski <apinski@cavium.com>
21181 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
21184 2017-01-23 Martin Sebor <msebor@redhat.com>
21186 PR middle-end/78703
21187 * gimple-ssa-sprintf.c (warn_level): New global.
21188 (format_integer): Use it here and throughout the rest of the file.
21189 Use the same switch to compute sign as base.
21190 (maybe_warn): New function.
21191 (format_directive): Factor out warnings into maybe_warn.
21192 Add debugging output. Use warn_level.
21193 (add_bytes): Use warn_level.
21194 (pass_sprintf_length::compute_format_length): Add debugging output.
21195 (try_substitute_return_value): Same.
21196 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
21198 PR middle-end/78703
21199 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
21200 (struct fmtresult, format_integer, format_floating): Adjust.
21201 (fmtresult::fmtresult): Set max correctly in two argument ctor.
21202 (get_string_length, format_string,format_directive): Same.
21203 (pass_sprintf_length::compute_format_length): Same.
21204 (try_substitute_return_value): Simplify slightly.
21206 PR middle-end/78703
21207 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
21208 (fmtresult::operator+=): Outlined.
21209 (struct fmtresult): Add ctors.
21210 (struct conversion_spec): Rename...
21211 (struct directive): ...to this. Add and remove data members.
21212 (directive::set_width, directive::set_precision): New functions.
21213 (format_percent): Use fmtresult ctor.
21214 (get_width_and_precision): Remove.
21215 (format_integer): Make naming changes. Avoid computing width and
21217 (format_floating): Same. Adjust indentation.
21218 (format_character, format_none): New functions.
21219 (format_string): Moved character handling to format_character.
21220 (format_directive): Remove arguments, change return type.
21221 (parse_directive): New function.
21222 (pass_sprintf_length::compute_format_length): Move directive
21223 parsing to parse_directive.
21225 2017-01-23 Jakub Jelinek <jakub@redhat.com>
21227 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
21228 (assign_assembler_name_if_needed): ... this.
21229 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
21230 (assign_assembler_name_if_needed): ... this.
21231 (free_lang_data_in_cgraph): Adjust callers.
21232 * cgraphunit.c (cgraph_node::analyze): Likewise.
21233 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
21236 2017-01-23 Richard Biener <rguenther@suse.de>
21238 PR tree-optimization/79088
21239 PR tree-optimization/79188
21240 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
21241 resetting loop bounds after last path deletion. Reset loop
21242 bounds of the target loop, make code match the comments.
21243 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
21244 Make sure loops need no fixups.
21246 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
21248 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
21249 exponent support with double type for first argument.
21250 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
21251 type returned by __builtin_vec_extract_sig,
21252 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
21253 functions from "vector int" to "vector unsigned int" or from
21254 "vector long long int" to "vector unsigned long long int".
21255 Changed type returned by __builtin_vec_extract_exp,
21256 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
21257 functions from "vector int" to "vector unsigned int" or from
21258 "vector long long int" to "vector unsigned long long int".
21259 Changed return type of __builtin_vec_test_data_class,
21260 __builtin_vec_test_data_class_sp, and
21261 __builtin_vec_test_data_class_dp from "vector int" to
21262 "vector bool int" or from "vector long long int" to "vector bool
21263 long long int" and changed second argument type from "unsigned
21264 int" to "int". Added new overloaded function forms "vector float
21265 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
21266 "vector float __builtin_vec_insert_exp_sp (vector float, vector
21267 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
21268 double, vector unsigned long long int)" and "vector double
21269 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
21270 long int)". Changed return type of
21271 __builtin_scalar_test_data_class and
21272 __builtin_scalar_test_data_class_sp and
21273 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
21274 int" and changed second argument from "unsigned int" to "int".
21275 Changed type returned by __builtin_scalar_test_neg,
21276 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
21277 from "int" to "bool int". Added new overloaded function form
21278 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
21279 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
21280 exponent double-precision with floating point first argument.
21281 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
21282 documentation of scalar_test_data_class, scalar_test_neg,
21283 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
21284 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
21285 vec_test_data_class built-in functions to reflect refinements in
21286 their type signatures.
21288 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
21290 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
21292 (aarch64_elf_asm_destructor): Likewise.
21294 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
21296 PR rtl-optimization/78634
21297 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
21298 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
21299 * ifcvt.c (noce_try_cmove): Add missing cost check.
21301 PR rtl-optimization/71724
21302 * combine.c (if_then_else_cond): Look for situations where it is
21303 beneficial to undo the work of one of the recursive calls.
21305 2017-01-23 Bin Cheng <bin.cheng@arm.com>
21307 PR tree-optimization/70754
21308 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
21309 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
21310 combined stmt before it if not NULL.
21311 (combine_chains): Process refs reversely and compute dominance point
21314 2017-01-23 Martin Liska <mliska@suse.cz>
21316 PR tree-optimization/79196
21317 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
21318 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
21320 (strlen_optimize_stmt): Call the renamed function.
21322 2017-01-23 Michael Matz <matz@suse.de>
21324 PR tree-optimization/78384
21325 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
21327 2017-01-23 Richard Biener <rguenther@suse.de>
21329 PR tree-optimization/79186
21330 * tree-vrp.c (register_new_assert_for): Make sure we've seen
21331 both incoming edges before moving an assert.
21333 2017-01-23 Martin Jambor <mjambor@suse.cz>
21335 * ipa-prop.c (load_from_param_1): Removed.
21336 (load_from_unmodified_param): Bits from load_from_param_1 put back
21338 (load_from_param): Removed.
21339 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
21340 with stmt. Reverted back to use of load_from_unmodified_param.
21342 2017-01-23 Martin Jambor <mjambor@suse.cz>
21345 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
21346 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
21347 field a pointer to garbage collected vector, mark lattices and
21348 ipcp_orig_node with GTY((skip)).
21349 (ipa_get_param_count): Adjust to descriptors being a pointer.
21350 (ipa_get_param): Likewise.
21351 (ipa_get_type): Likewise.
21352 (ipa_get_param_move_cost): Likewise.
21353 (ipa_set_param_used): Likewise.
21354 (ipa_get_controlled_uses): Likewise.
21355 (ipa_set_controlled_uses): Likewise.
21356 (ipa_is_param_used): Likewise.
21357 (ipa_node_params_t): Move into garbage collector. New methods insert
21359 (ipa_node_params_sum): Annotate wth GTY(()).
21360 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
21362 (ipa_load_from_parm_agg): Adjust declaration.
21363 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
21364 * ipa-profile.c (ipa_profile): Likewise.
21365 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
21366 (ipa_populate_param_decls): Make descriptors parameter garbage
21368 (ipa_dump_param): Adjust to descriptors being a pointer.
21369 (ipa_alloc_node_params): Likewise.
21370 (ipa_initialize_node_params): Likewise.
21371 (load_from_param_1): Make descriptors parameter garbage collected.
21372 (load_from_unmodified_param): Likewise.
21373 (load_from_param): Likewise.
21374 (ipa_load_from_parm_agg): Likewise.
21375 (ipa_node_params::~ipa_node_params): Removed.
21376 (ipa_free_all_node_params): Remove call to delete operator.
21377 (ipa_node_params_t::insert): New.
21378 (ipa_node_params_t::remove): Likewise.
21379 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
21380 copy known_csts and known_contexts vectors.
21381 (ipa_read_node_info): Adjust to descriptors being a pointer.
21382 (ipcp_modif_dom_walker): Make m_descriptors field garbage
21384 (ipcp_transform_function): Make descriptors variable garbage
21387 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
21389 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
21390 * config/i386/avx512dqintrin.h: Ditto.
21391 * config/i386/avx512fintrin.h: Ditto.
21392 * config/i386/i386.c: Handle new builtins.
21393 * config/i386/i386-builtin.def: Add new builtins.
21394 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
21395 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
21397 2017-01-23 Jakub Jelinek <jakub@redhat.com>
21398 Martin Liska <mliska@suse.cz>
21400 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
21401 * asan.c (asan_expand_poison_ifn): Support stores and use
21402 appropriate ASAN report function.
21403 * internal-fn.c (expand_ASAN_POISON_USE): New function.
21404 * internal-fn.def (ASAN_POISON_USE): Declare.
21405 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
21406 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
21407 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
21408 ASAN_POISON calls w/o LHS.
21409 * tree-ssa.c (execute_update_addresses_taken): Create clobber
21410 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
21411 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
21412 * gimplify.c (asan_poison_variables): Add attribute
21413 use_after_scope_memory to variables that really needs to live
21415 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
21416 having the attribute.
21418 2017-01-23 Martin Liska <mliska@suse.cz>
21420 * asan.c (create_asan_shadow_var): New function.
21421 (asan_expand_poison_ifn): Likewise.
21422 * asan.h (asan_expand_poison_ifn): New declaration.
21423 * internal-fn.c (expand_ASAN_POISON): Likewise.
21424 * internal-fn.def (ASAN_POISON): New builtin.
21425 * sanopt.c (pass_sanopt::execute): Expand
21426 asan_expand_poison_ifn.
21427 * tree-inline.c (copy_decl_for_dup_finish): Make function
21429 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
21430 * tree-ssa.c (is_asan_mark_p): New function.
21431 (execute_update_addresses_taken): Rewrite local variables
21432 (identified just by use-after-scope as addressable) into SSA.
21434 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21436 * doc/install.texi (Specific): opensource.apple.com uses https
21437 now. Remove trailing slash.
21439 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21441 * README.Portability: Remove note on an Irix compatibility issue.
21443 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
21445 * gcov.c (INCLUDE_ALGORITHM): Define.
21446 (INCLUDE_VECTOR): Define.
21447 No longer include <vector> and <algorithm> directly.
21449 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
21451 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
21453 * doc/invoke.texi (Code Gen Options): Ditto.
21455 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
21458 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
21460 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
21462 rtl-optimization/79125
21463 * cprop.c (local_cprop_pass): Handle cases where we make an
21464 unconditional trap.
21466 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
21470 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
21471 read from, for big endian.
21473 2017-01-20 Jiong Wang <jiong.wang@arm.com>
21475 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
21476 register pauth builtins for LP64 only.
21478 2017-01-20 Marek Polacek <polacek@redhat.com>
21481 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
21484 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21486 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
21488 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
21489 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
21490 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
21492 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21495 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
21496 in big-endian mode when they are not a single duplicated value.
21498 2017-01-20 Richard Biener <rguenther@suse.de>
21500 * BASE-VER: Bump to 7.0.1.
21502 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21504 * omp-low.c (omplow_simd_context): New struct. Use it...
21505 (lower_rec_simd_input_clauses): ...here and...
21506 (lower_rec_input_clauses): ...here to hold common data. Adjust all
21507 references to idx, lane, max_vf, is_simt.
21509 2017-01-20 Graham Markall <graham.markall@embecosm.com>
21511 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
21514 2017-01-20 Martin Jambor <mjambor@suse.cz>
21516 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
21517 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
21519 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
21520 (GTFILES): Rename hsa.c to hsa-common.c.
21521 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
21522 * hsa-dump.c: Likewise.
21523 * hsa-gen.c: Likewise.
21524 * hsa-regalloc.c: Likewise.
21525 * ipa-hsa.c: Likewise.
21526 * omp-expand.c: Likewise.
21527 * omp-low.c: Likewise.
21528 * toplev.c: Likewise.
21530 2017-01-20 Marek Polacek <polacek@redhat.com>
21533 * doc/invoke.texi: Document -Wduplicated-branches.
21534 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
21535 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
21536 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
21537 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
21538 return 0 only when not OEP_LEXICOGRAPHIC.
21539 (fold_build_cleanup_point_expr): Use the expression
21540 location when building CLEANUP_POINT_EXPR.
21541 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
21542 * tree.c (add_expr): Handle error_mark_node.
21544 2017-01-20 Martin Liska <mliska@suse.cz>
21547 * tree-profile.c (init_ic_make_global_vars): Do not call
21549 (gimple_init_gcov_profiler): Likewise.
21551 2017-01-20 Martin Liska <mliska@suse.cz>
21554 * cgraph.h (maybe_create_reference): Remove argument and
21556 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
21558 * ipa-cp.c (create_specialized_node): Likewise.
21559 * symtab.c (symtab_node::maybe_create_reference): Handle
21560 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
21562 2017-01-20 Martin Liska <mliska@suse.cz>
21564 * read-rtl-function.c (function_reader::create_function): Use
21565 build_decl instread of build_decl_stat.
21567 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
21569 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
21570 * config/i386/avx512dqintrin.h: Ditto.
21571 * config/i386/avx512fintrin.h: Ditto.
21572 * config/i386/i386-builtin-types.def: Add new types.
21573 * config/i386/i386.c: Handle new types.
21574 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
21575 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
21576 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
21577 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
21578 (__builtin_ia32_kshiftridi): New.
21579 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
21581 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
21585 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
21586 define to rs6000_init_stack_protect_guard.
21587 (rs6000_init_stack_protect_guard): New function.
21589 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
21590 Yunqiang Su <yunqiang.su@imgtec.com>
21592 * config.gcc (supported_defaults): Add madd4.
21593 (with_madd4): Add validation.
21594 (all_defaults): Add madd4.
21595 * config/mips/mips.opt (mmadd4): New option.
21596 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
21598 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
21600 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
21601 (ISA_HAS_FUSED_MADD4): Likewise.
21602 * doc/invoke.texi (-mmadd4): Document the new option.
21603 * doc/install.texi (--with-madd4): Document the new option.
21605 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21607 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
21608 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
21609 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
21610 (aarch64_init_pauth_hint_builtins): New.
21611 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
21612 (aarch64_expand_builtin): Expand new builtins.
21614 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21616 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
21617 * combine-stack-adj.c (no_unhandled_cfa): Handle
21618 REG_CFA_TOGGLE_RA_MANGLE.
21619 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
21620 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
21621 info for return address signing.
21622 (aarch64_expand_epilogue): Likewise.
21624 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21626 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
21627 * config/aarch64/aarch64-protos.h
21628 (aarch64_return_address_signing_enabled): New declaration.
21629 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
21631 (aarch64_expand_prologue): Sign return address before it's pushed onto
21633 (aarch64_expand_epilogue): Authenticate return address fetched from
21635 (aarch64_override_options): Sanity check for ILP32 and ISA level.
21636 (aarch64_attributes): New function attributes for "sign-return-address".
21637 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
21638 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
21639 ("*do_return"): Generate combined instructions according to key index.
21640 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
21641 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
21643 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
21644 * config/aarch64/aarch64.opt (msign-return-address=): New.
21645 * doc/extend.texi (AArch64 Function Attributes): Documents
21646 "sign-return-address=".
21647 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
21649 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
21651 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
21652 overall option summary.
21654 2017-01-19 Jiong Wang <jiong.wang@arm.com>
21656 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
21657 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
21658 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
21659 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
21661 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
21663 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
21664 -mpower9-minmax by default for -mcpu=power9.
21665 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
21666 128-bit floating point.
21668 2017-01-20 Alan Modra <amodra@gmail.com>
21670 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
21671 optimizing for size.
21673 2017-01-20 Alan Modra <amodra@gmail.com>
21676 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
21677 for strcmp and strncmp from corresponding builtin decl.
21679 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
21681 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
21682 instead of i386/rtems-64.h.
21683 * config/i386/rtems-64.h: Remove.
21685 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
21689 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
21691 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
21693 2017-01-19 Tamar Christina <tamar.christina@arm.com>
21695 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
21696 Change int to HOST_WIDE_INT.
21697 * config/aarch64/aarch64-protos.h
21698 (aarch64_simd_gen_const_vector_dup): Likewise.
21699 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
21701 2017-01-19 David Malcolm <dmalcolm@redhat.com>
21703 * langhooks-def.h (lhd_type_for_size): New decl.
21704 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
21705 * langhooks.c (lhd_type_for_size): New function, taken from
21708 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
21710 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
21711 define_bypass for CR latency.
21712 (power9-cracked-alu): Update bypass latency and remove power9-branch.
21713 (power9-alu2): Add define_bypass for CR latency.
21715 (power9-mul): Update insn latency.
21716 (power9-mul-compare): Update insn latency, bypass latency and remove
21719 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21721 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
21723 * config/aarch64/aarch64.md
21724 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
21725 aarch64_nopcrelative_literal_loads.
21726 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
21728 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
21730 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
21731 TARGET_LOONGSON_3A.
21732 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
21734 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
21737 * config.gcc (supported_defaults): Add lxc1-sxc1.
21738 (with_lxc1_sxc1): Add validation.
21739 (all_defaults): Add lxc1-sxc1.
21740 * config/mips/mips.opt (mlxc1-sxc1): New option.
21741 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
21743 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
21744 __mips_no_lxc1_sxc1.
21745 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
21746 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
21747 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
21749 2017-01-19 Richard Biener <rguenther@suse.de>
21751 PR tree-optimization/72488
21752 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
21753 sure to restore SSA info.
21754 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
21756 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
21758 PR rtl-optimization/79121
21759 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
21760 of the inner type when shifting an extended value.
21762 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
21765 * symtab.c (symtab_node::equal_address_to): Fix comparing of
21766 interposable aliases.
21768 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
21771 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
21772 Use the evmergelohi instruction.
21773 (mov_si<mode>_e500_subreg4_2_le): Likewise.
21774 (mov_sitf_e500_subreg8_2_be): Likewise.
21775 (mov_sitf_e500_subreg12_2_le): Likewise.
21776 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
21777 (mov_si<mode>_e500_subreg4_2_be): Likewise.
21778 (mov_sitf_e500_subreg8_2_le): Likewise.
21779 (mov_sitf_e500_subreg12_2_be): Likewise.
21781 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21783 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
21784 attribute from vecsimple to vecperm.
21785 (altivec_vbpermq2): Likewise.
21787 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21790 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
21792 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21793 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
21794 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
21795 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
21796 case where N arg is SIZE_MAX.
21797 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
21798 (cmpstrsi): Add pattern.
21800 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21802 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
21803 __builtin_vec_revb builtins.
21804 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
21805 built-in functions to support generation of the ISA 3.0 XXBR<x>
21806 vector byte reverse instructions.
21807 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
21808 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
21809 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
21810 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
21811 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
21812 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
21813 (P9V_BUILTIN_VEC_REVB): Likewise.
21814 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
21815 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
21816 (p9_xxbrq_v16qi): Likewise.
21817 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
21818 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
21819 (p9_xxbrh_v8hi): Likewise.
21820 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
21821 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
21822 vec_revb built-in functions.
21824 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
21826 PR rtl-optimization/78952
21827 * config/i386/i386.md (any_extract): New code iterator.
21828 (*insvqi_2): Use any_extract for source operand.
21829 (*insvqi_3): Use any_shiftrt for source operand.
21831 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
21833 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
21835 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
21837 2017-01-18 Matthias Klose <doko@ubuntu.com>
21839 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
21841 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21843 * config/rs6000/altivec.h (vec_bperm): Change #define.
21844 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
21845 (altivec_vbpermq2): New define_insn.
21846 (altivec_vbpermd): Likewise.
21847 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
21848 function interface.
21849 (VBPERMD): Likewise.
21850 (VBPERM): New polymorphic function interface.
21851 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
21852 Add entries for P9V_BUILTIN_VEC_VBPERM.
21853 * doc/extend.texi: Add interfaces for vec_bperm.
21855 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21857 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
21858 first letter of error messages.
21859 (s390_resolve_overloaded_builtin): Likewise.
21860 * config/s390/s390.c (s390_expand_builtin): Likewise.
21861 (s390_invalid_arg_for_unprototyped_fn): Likewise.
21862 (s390_valid_target_attribute_inner_p): Likewise.
21863 * config/s390/s390.md ("tabort"): Likewise.
21865 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
21867 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
21868 (ISA_AVOID_DIV_HILO): New macro.
21869 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
21870 (ISA_HAS_DDIV): Likewise.
21872 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21874 * doc/invoke.texi (fabi-version): Correct number of occurrences.
21876 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21878 * doc/invoke.texi (fabi-version): Spelling fix.
21880 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21883 * doc/invoke.texi (fabi-version): Mention mangling fix for
21886 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
21889 * doc/invoke.texi (fabi-version): Document discriminator mangling.
21891 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
21894 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
21895 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
21897 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
21899 (stack_protect_test): Ditto.
21900 * config/rs6000/rs6000.opt (mstack-protector-guard=,
21901 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
21903 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
21904 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
21905 -mstack-protector-guard-offset=.
21906 (RS/6000 and PowerPC Options): Ditto.
21908 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
21910 * config/i386/i386.h (MASK_CLASS_P): New define.
21911 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
21912 there are no registers from different register sets also when
21913 mask registers are used. Update function comment.
21914 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
21915 to (*k/*r) and (*k/*km) alternatives.
21917 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
21919 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
21920 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
21921 (EH_RETURN_HANDLER_RTX): New define.
21922 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
21923 Force frame pointer in EH return functions.
21924 (aarch64_expand_epilogue): Add barrier for eh_return.
21925 (aarch64_final_eh_return_addr): Remove.
21926 (aarch64_eh_return_handler_rtx): New function.
21927 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
21929 (aarch64_eh_return_handler_rtx): New prototype.
21931 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21933 * config/rs6000/altivec.h (vec_rlmi): New #define.
21934 (vec_vrlnm): Likewise.
21935 (vec_rlnm): Likewise.
21936 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
21937 (UNSPEC_VRLNM): Likewise.
21938 (VIlong): New mode iterator.
21939 (altivec_vrl<VI_char>mi): New define_insn.
21940 (altivec_vrl<VI_char>nm): Likewise.
21941 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
21943 (VRLDNM): Likewise.
21944 (RLNM): New polymorphic function entry.
21945 (VRLWMI): New monomorphic function entry.
21946 (VRLDMI): Likewise.
21947 (RLMI): New polymorphic function entry.
21948 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
21949 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
21950 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
21953 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21956 * dwarf2out.c (field_byte_offset): Restore the
21957 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
21958 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
21959 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
21962 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
21965 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
21967 2017-01-17 Jakub Jelinek <jakub@redhat.com>
21970 * dwarf2out.c (add_data_member_location_attribute): For constant
21971 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
21972 instead of DW_AT_data_member_location, DW_AT_bit_offset and
21973 DW_AT_byte_size attributes.
21975 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
21977 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
21978 after forcing to constant memory when the code model is medium.
21980 2017-01-17 Julia Koval <julia.koval@intel.com>
21983 * config/i386/avx512fintrin.h
21984 (_mm512_i32gather_ps): Change __addr type to void const*.
21985 (_mm512_mask_i32gather_ps): Ditto.
21986 (_mm512_i32gather_pd): Ditto.
21987 (_mm512_mask_i32gather_pd): Ditto.
21988 (_mm512_i64gather_ps): Ditto.
21989 (_mm512_mask_i64gather_ps): Ditto.
21990 (_mm512_i64gather_pd): Ditto.
21991 (_mm512_mask_i64gather_pd): Ditto.
21992 (_mm512_i32gather_epi32): Ditto.
21993 (_mm512_mask_i32gather_epi32): Ditto.
21994 (_mm512_i32gather_epi64): Ditto.
21995 (_mm512_mask_i32gather_epi64): Ditto.
21996 (_mm512_i64gather_epi32): Ditto.
21997 (_mm512_mask_i64gather_epi32): Ditto.
21998 (_mm512_i64gather_epi64): Ditto.
21999 (_mm512_mask_i64gather_epi64): Ditto.
22000 (_mm512_i32scatter_ps): Change __addr type to void*.
22001 (_mm512_mask_i32scatter_ps): Ditto.
22002 (_mm512_i32scatter_pd): Ditto.
22003 (_mm512_mask_i32scatter_pd): Ditto.
22004 (_mm512_i64scatter_ps): Ditto.
22005 (_mm512_mask_i64scatter_ps): Ditto.
22006 (_mm512_i64scatter_pd): Ditto.
22007 (_mm512_mask_i64scatter_pd): Ditto.
22008 (_mm512_i32scatter_epi32): Ditto.
22009 (_mm512_mask_i32scatter_epi32): Ditto.
22010 (_mm512_i32scatter_epi64): Ditto.
22011 (_mm512_mask_i32scatter_epi64): Ditto.
22012 (_mm512_i64scatter_epi32): Ditto.
22013 (_mm512_mask_i64scatter_epi32): Ditto.
22014 (_mm512_i64scatter_epi64): Ditto.
22015 (_mm512_mask_i64scatter_epi64): Ditto.
22016 * config/i386/avx512pfintrin.h
22017 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
22018 (_mm512_mask_prefetch_i32gather_ps): Ditto.
22019 (_mm512_mask_prefetch_i64gather_pd): Ditto.
22020 (_mm512_mask_prefetch_i64gather_ps): Ditto.
22021 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
22022 (_mm512_prefetch_i32scatter_ps): Ditto.
22023 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
22024 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
22025 (_mm512_prefetch_i64scatter_pd): Ditto.
22026 (_mm512_prefetch_i64scatter_ps): Ditto.
22027 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
22028 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
22029 * config/i386/avx512vlintrin.h
22030 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
22031 (_mm_mmask_i32gather_ps): Ditto.
22032 (_mm256_mmask_i32gather_pd): Ditto.
22033 (_mm_mmask_i32gather_pd): Ditto.
22034 (_mm256_mmask_i64gather_ps): Ditto.
22035 (_mm_mmask_i64gather_ps): Ditto.
22036 (_mm256_mmask_i64gather_pd): Ditto.
22037 (_mm_mmask_i64gather_pd): Ditto.
22038 (_mm256_mmask_i32gather_epi32): Ditto.
22039 (_mm_mmask_i32gather_epi32): Ditto.
22040 (_mm256_mmask_i32gather_epi64): Ditto.
22041 (_mm_mmask_i32gather_epi64): Ditto.
22042 (_mm256_mmask_i64gather_epi32): Ditto.
22043 (_mm_mmask_i64gather_epi32): Ditto.
22044 (_mm256_mmask_i64gather_epi64): Ditto.
22045 (_mm_mmask_i64gather_epi64): Ditto.
22046 (_mm256_i32scatter_ps): Change __addr type to void*.
22047 (_mm256_mask_i32scatter_ps): Ditto.
22048 (_mm_i32scatter_ps): Ditto.
22049 (_mm_mask_i32scatter_ps): Ditto.
22050 (_mm256_i32scatter_pd): Ditto.
22051 (_mm256_mask_i32scatter_pd): Ditto.
22052 (_mm_i32scatter_pd): Ditto.
22053 (_mm_mask_i32scatter_pd): Ditto.
22054 (_mm256_i64scatter_ps): Ditto.
22055 (_mm256_mask_i64scatter_ps): Ditto.
22056 (_mm_i64scatter_ps): Ditto.
22057 (_mm_mask_i64scatter_ps): Ditto.
22058 (_mm256_i64scatter_pd): Ditto.
22059 (_mm256_mask_i64scatter_pd): Ditto.
22060 (_mm_i64scatter_pd): Ditto.
22061 (_mm_mask_i64scatter_pd): Ditto.
22062 (_mm256_i32scatter_epi32): Ditto.
22063 (_mm256_mask_i32scatter_epi32): Ditto.
22064 (_mm_i32scatter_epi32): Ditto.
22065 (_mm_mask_i32scatter_epi32): Ditto.
22066 (_mm256_i32scatter_epi64): Ditto.
22067 (_mm256_mask_i32scatter_epi64): Ditto.
22068 (_mm_i32scatter_epi64): Ditto.
22069 (_mm_mask_i32scatter_epi64): Ditto.
22070 (_mm256_i64scatter_epi32): Ditto.
22071 (_mm256_mask_i64scatter_epi32): Ditto.
22072 (_mm_i64scatter_epi32): Ditto.
22073 (_mm_mask_i64scatter_epi32): Ditto.
22074 (_mm256_i64scatter_epi64): Ditto.
22075 (_mm256_mask_i64scatter_epi64): Ditto.
22076 (_mm_i64scatter_epi64): Ditto.
22077 (_mm_mask_i64scatter_epi64): Ditto.
22078 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
22079 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
22080 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
22081 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
22082 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
22083 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
22084 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
22085 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
22086 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
22087 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
22088 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
22089 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
22090 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
22091 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
22092 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
22093 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
22094 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
22095 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
22096 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
22097 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
22098 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
22099 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
22100 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
22101 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
22102 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
22103 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
22104 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
22105 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
22106 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
22107 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
22108 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
22109 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
22110 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
22111 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
22112 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
22113 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
22114 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
22115 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
22116 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
22117 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
22118 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
22119 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
22120 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
22121 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
22122 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
22123 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
22124 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
22125 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
22126 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
22127 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
22128 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
22129 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
22130 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
22131 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
22132 definitions accordingly.
22134 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
22135 Kuan-Lin Chen <kuanlinchentw@gmail.com>
22138 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
22141 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
22144 * ira-conflicts.c (ira_build_conflicts): Update total conflict
22145 hard regs for inner regno.
22147 2017-01-17 Martin Liska <mliska@suse.cz>
22150 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
22151 assumption and add comment.
22153 2017-01-17 Nathan Sidwell <nathan@acm.org>
22155 * ipa-visibility.c (localize_node): New function, broken out of ...
22156 (function_and_variable_visibility): ... here. Call it.
22158 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
22160 PR middle-end/77445
22161 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
22162 correctly set frequency of oudgoing edge.
22163 (duplicate_thread_path): Fix profile updating.
22165 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22168 * configure.ac: Add GCC_BASE_VER.
22169 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
22170 version from BASE-VER file.
22171 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
22172 (gcc.o): Depend on $(BASEVER).
22173 * common.opt (dumpfullversion): New option.
22174 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
22175 * doc/invoke.texi: Document -dumpfullversion.
22176 * doc/install.texi: Document --with-gcc-major-version-only.
22177 * configure: Regenerated.
22179 2017-01-17 Richard Biener <rguenther@suse.de>
22181 PR tree-optimization/71433
22182 * tree-vrp.c (register_new_assert_for): Merge same asserts
22183 on all incoming edges.
22184 (process_assert_insertions_for): Handle insertions at the
22187 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
22189 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
22190 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
22192 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
22195 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
22198 2017-01-17 Alan Modra <amodra@gmail.com>
22201 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
22202 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
22203 symbolic stack limit when pic.
22205 2017-01-16 Martin Sebor <msebor@redhat.com>
22207 PR tree-optimization/78608
22208 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
22210 2017-01-16 Jeff Law <law@redhat.com>
22213 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
22214 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
22215 for several include directories that may be relative to sysroot.
22216 * config/i386/x-mingw32 (gplus_includedir): Define.
22217 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
22218 (native_system_includedir): Likewise.
22219 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
22220 override if TARGET_SYSTEM_ROOT is defined.
22221 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
22223 PR tree-optimization/79090
22224 PR tree-optimization/33562
22225 PR tree-optimization/61912
22226 PR tree-optimization/77485
22227 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
22228 and computed trims into the dump file.
22230 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
22232 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
22234 2017-01-16 Jakub Jelinek <jakub@redhat.com>
22237 * gimplify.c (gimplify_init_constructor): If want_value and
22238 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
22242 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
22243 sequence. Formatting fixes.
22244 (doloop_optimize): Formatting fixes.
22247 * gcc.c (debug_level_greater_than_spec_func): New function.
22248 (static_spec_functions): Add debug-level-gt spec function.
22249 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
22251 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22252 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22253 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
22254 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
22255 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
22256 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
22258 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
22260 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
22261 QImode fixups to general and mask registers only.
22263 2017-01-16 Carl Love <cel@us.ibm.com>
22265 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
22266 for built-in functions
22267 vector signed char vec_nabs (vector signed char)
22268 vector signed short vec_nabs (vector signed short)
22269 vector signed int vec_nabs (vector signed int)
22270 vector signed long long vec_nabs (vector signed long long)
22271 vector float vec_nabs (vector float)
22272 vector double vec_nabs (vector double)
22273 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
22275 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
22276 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
22277 * doc/extend.texi: Update the documentation file for the new built-in
22280 2017-01-16 Martin Sebor <msebor@redhat.com>
22282 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
22285 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22287 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
22288 UNSPEC_VSX__XXSPLTD to require special splat handling.
22290 2017-01-16 David Malcolm <dmalcolm@redhat.com>
22293 * system.h: Poison strndup.
22295 2017-01-16 Alan Modra <amodra@gmail.com>
22298 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
22301 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
22303 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
22305 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
22307 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
22308 call recog here. Assert that INSN_CODE (insn) is non-negative.
22310 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22313 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
22315 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
22316 in the currently scheduled RTL fragment.
22318 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22320 PR rtl-optimization/78751
22321 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
22324 2017-01-14 Jeff Law <law@redhat.com>
22326 PR tree-optimization/79090
22327 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
22328 variable length stores.
22329 (compute_trims): Delete dead assignment to *trim_tail.
22330 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
22333 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
22335 PR rtl-optimization/78626
22336 PR rtl-optimization/78727
22337 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
22338 of a block, and split such blocks after everything else is finished.
22340 2017-01-14 Alan Modra <amodra@gmail.com>
22343 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
22344 target legitimate_combined_insn.
22345 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
22346 (rs6000_legitimate_combined_insn): New function.
22347 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
22349 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
22350 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
22351 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
22353 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
22355 * doc/frontends.texi (G++ and GCC): Remove references to Java.
22357 2017-01-13 Jeff Law <law@redhat.com>
22359 PR tree-optimization/33562
22360 PR tree-optimization/61912
22361 PR tree-optimization/77485
22362 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
22364 (delete_dead_assignment): Likewise.
22365 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
22366 statement to delete_dead_call and delete_dead_assignment.
22368 2017-01-13 David Malcolm <dmalcolm@redhat.com>
22371 * substring-locations.c (format_warning_va): Strengthen case 1 so
22372 that both endpoints of the substring must be within the format
22373 range for just the substring to be printed.
22375 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
22377 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
22378 * config/i386/i386.c (ix86_target_string): Add missing options
22379 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
22380 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
22381 flags_other and ix86_target_other to flags2_other. Display unknown
22383 (ix86_valid_target_attribute_inner_p): Add missing options and
22384 reorder options by implied ISAs, as in ix86_target_string.
22386 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22388 * hash-table.h (hash_table::too_empty_p): New function.
22389 (hash_table::expand): Use it.
22390 (hash_table::traverse): Likewise.
22391 (hash_table::empty_slot): Use sizeof (value_type) instead of
22392 sizeof (PTR) to convert bytes to elements. Shrink the table
22393 if the current size is excessive for the current number of
22396 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22398 * ira-costs.c (record_reg_classes): Break from the inner loop
22399 early once alt_fail is known to be true. Update outer loop
22400 handling accordingly.
22402 2017-01-13 Jeff Law <law@redhat.com>
22404 * tree-ssa-dse.c (decrement_count): New function.
22405 (increment_start_addr, maybe_trim_memstar_call): Likewise.
22406 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
22407 when we know the partially dead statement is a mem* function.
22409 PR tree-optimization/61912
22410 PR tree-optimization/77485
22411 * tree-ssa-dse.c: Include expr.h.
22412 (maybe_trim_constructor_store): New function.
22413 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
22415 PR tree-optimization/33562
22416 PR tree-optimization/61912
22417 PR tree-optimization/77485
22418 * doc/invoke.texi: Document new dse-max-object-size param.
22419 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
22420 * tree-ssa-dse.c: Include params.h.
22421 (dse_store_status): New enum.
22422 (initialize_ao_ref_for_dse): New, partially extracted from
22424 (valid_ao_ref_for_dse, normalize_ref): New.
22425 (setup_live_bytes_from_ref, compute_trims): Likewise.
22426 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
22427 (maybe_trim_partially_dead_store): Likewise.
22428 (maybe_trim_complex_store): Likewise.
22429 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
22430 Track what bytes live from the original store. Return tri-state
22431 for dead, partially dead or live.
22432 (dse_dom_walker): Add constructor, destructor and new private members.
22433 (delete_dead_call, delete_dead_assignment): New extracted from
22435 (dse_optimize_stmt): Make a member of dse_dom_walker.
22436 Use initialize_ao_ref_for_dse.
22438 PR tree-optimization/33562
22439 PR tree-optimization/61912
22440 PR tree-optimization/77485
22441 * sbitmap.h (bitmap_count_bits): Prototype.
22442 (bitmap_clear_range, bitmap_set_range): Likewise.
22443 * sbitmap.c (bitmap_clear_range): New function.
22444 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
22446 2017-01-13 Martin Liska <mliska@suse.cz>
22449 * function.c (set_cfun): Add new argument force.
22450 * function.h (set_cfun): Likewise.
22451 * ipa-inline-transform.c (inline_call): Use the function when
22452 strict alising from is dropped for function we inline to.
22454 2017-01-13 Richard Biener <rguenther@suse.de>
22456 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
22457 for dumping GIMPLE INTEGER_CSTs.
22459 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22461 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
22462 to 201112L since C++17.
22464 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22467 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
22468 if -fsanitize=kernel-address is present.
22470 2017-01-13 Richard Biener <rguenther@suse.de>
22472 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
22473 as _Literal ( type ) number in case usual suffixes do not
22474 preserve all information.
22476 2017-01-13 Richard Biener <rguenther@suse.de>
22478 PR tree-optimization/77283
22479 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
22480 and ssa-iterators.h.
22481 (is_feasible_trace): Implement a cost model based on joiner
22484 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22487 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
22488 char or short to __float128/_Float128 directly.
22490 2017-01-12 Martin Sebor <msebor@redhat.com>
22492 to -Wformat-overflow.
22493 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
22494 (min_bytes_remaining): Same.
22495 (get_string_length): Same.
22496 (format_string): Same.
22497 (format_directive): Same.
22499 (pass_sprintf_length::handle_gimple_call): Same.
22501 2017-01-12 Jakub Jelinek <jakub@redhat.com>
22503 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
22504 info.nowrite calls with no lhs that can't throw. Return bool
22505 whether gsi_remove has been called or not.
22506 (pass_sprintf_length::handle_gimple_call): Return bool whether
22507 try_substitute_return_value called gsi_remove. Formatting fix.
22508 (pass_sprintf_length::execute): Don't use gsi_remove if
22509 handle_gimple_call returned true.
22512 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
22513 be removed due to side-effects, don't remove following barrier nor
22514 turn the successor edge into fallthru edge.
22516 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22519 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
22520 element-reversing loads and stores as not swappable.
22522 2017-01-12 Nathan Sidwell <nathan@acm.org>
22523 Nicolai Stange <nicstange@gmail.com>
22525 * combine.c (try_combine): Don't ignore result of overlap checking
22526 loop. Combine overlap & asm check into single loop.
22528 2017-01-12 Richard Biener <rguenther@suse.de>
22530 * tree-pretty-print.c (dump_generic_node): Provide -gimple
22531 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
22533 2017-01-12 Richard Biener <rguenther@suse.de>
22535 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
22536 and TS_TARGET_OPTION directly derive from TS_BASE.
22537 * tree-core.h (tree_optimization_option): Derive from tree_base.
22538 (tree_target_option): Likewise.
22540 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
22542 * config/i386/i386.c (memory_address_length): Increase len
22543 only when rip_relative_addr_p returns false.
22545 2017-01-11 Julia Koval <julia.koval@intel.com>
22547 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
22548 (OPTION_MASK_ISA_SGX_SET): New.
22549 (ix86_handle_option): Handle OPT_msgx.
22550 * config.gcc: Added sgxintrin.h.
22551 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
22552 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
22553 * config/i386/i386.c (ix86_target_string): Add -msgx.
22555 (ix86_option_override_internal): Handle new options.
22556 (ix86_valid_target_attribute_inner_p): Add sgx.
22557 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
22558 * config/i386/i386.opt: Add msgx.
22559 * config/i386/sgxintrin.h: New file.
22560 * config/i386/x86intrin.h: Add sgxintrin.h.
22562 2017-01-11 Jakub Jelinek <jakub@redhat.com>
22565 * fold-const.c (maybe_nonzero_address): Return 1 for function
22567 (tree_single_nonzero_warnv_p): Don't handle function local objects
22571 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
22574 2017-01-11 David Malcolm <dmalcolm@redhat.com>
22577 * opts.c: Include "spellcheck.h"
22578 (struct string_fragment): New struct.
22579 (struct edit_distance_traits<const string_fragment &>): New
22581 (get_closest_sanitizer_option): New function.
22582 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
22584 2017-01-11 Jakub Jelinek <jakub@redhat.com>
22586 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
22588 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
22589 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
22590 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
22591 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
22592 for initial die_offset if dwarf_split_debug_info.
22593 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
22594 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
22596 (output_skeleton_debug_sections): Formatting fix. Use
22597 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
22598 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
22600 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
22602 * config/arm/cortex-a53.md: Add bypasses for
22603 cortex_a53_r2f_cvt.
22604 (cortex_a53_r2f): Only use for transfers.
22605 (cortex_a53_f2r): Likewise.
22606 (cortex_a53_r2f_cvt): Add reservation for conversions.
22607 (cortex_a53_f2r_cvt): Likewise.
22609 2017-01-11 Tamar Christina <tamar.christina@arm.com>
22611 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
22612 to all inlined functions, change static to extern.
22614 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
22617 * config/arm/arm.c (legitimize_pic_address): Handle reference to
22619 (arm_assemble_integer): Likewise.
22621 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
22623 * config.gcc: Use new awk script to check CPU, FPU and architecture
22624 parameters for --with-... options.
22625 * config/arm/parsecpu.awk: New file
22626 * config/arm/arm-cpus.in: New file.
22627 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
22629 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
22631 * config/arm/t-arm: Update dependency rules.
22632 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
22633 of processing .def files.
22634 * config/arm/genopt.sh: Deleted.
22635 * config/arm/gentune.sh: Deleted.
22636 * config/arm/arm-cores.def: Deleted.
22637 * config/arm/arm-arches.def: Deleted.
22638 * config/arm/arm-fpus.def: Deleted.
22639 * config/arm/arm-tune.md: Regenerated.
22640 * config/arm/arm-tables.opt: Regenerated.
22641 * config/arm/arm-cpu.h: New generated file.
22642 * config/arm/arm-cpu-data.h: New generated file.
22643 * config/arm/arm-cpu-cdata.h: New generated file.
22645 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
22648 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
22650 (input_varpool_node): Unpack dynamically_initialized bit.
22652 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
22654 PR rtl-optimization/79032
22655 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
22656 the alignment of the adjusted memory reference against that of MODE,
22657 instead of the alignment of the original memory reference.
22659 2017-01-11 Martin Jambor <mjambor@suse.cz>
22661 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
22663 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
22664 decorated functions.
22666 2017-01-11 Richard Biener <rguenther@suse.de>
22668 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
22669 set range/nonnull info for PHI results. Do not set it on
22670 stmts marked for removal.
22672 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
22674 * expr.c (store_field): In the bitfield case, fetch the return value
22675 from the registers before applying a single big-endian adjustment.
22676 Always do a final load for a BLKmode value not larger than a word.
22678 2017-01-10 David Malcolm <dmalcolm@redhat.com>
22681 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
22682 that we correctly handle column numbers greater than
22683 LINE_MAP_MAX_COLUMN_NUMBER.
22685 2017-01-10 Martin Sebor <msebor@redhat.com>
22687 PR middle-end/78245
22688 * gimple-ssa-sprintf.c (get_destination_size): Call
22689 {init,fini}object_sizes.
22690 * tree-object-size.c (addr_object_size): Adjust.
22691 (pass_through_call): Adjust.
22692 (pass_object_sizes::execute): Adjust.
22693 * tree-object-size.h (fini_object_sizes): Declare.
22695 2017-01-10 Martin Sebor <msebor@redhat.com>
22697 PR tree-optimization/78775
22698 * builtins.c (get_size_range): Move...
22699 * calls.c: ...to here.
22700 (alloc_max_size): Accept zero argument.
22701 (operand_signed_p): Remove.
22702 (maybe_warn_alloc_args_overflow): Call get_size_range.
22703 * calls.h (get_size_range): Declare.
22705 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
22707 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
22708 from TI's devices.csv file as of September 2016.
22709 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
22711 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
22713 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
22714 * doc/invoke.texi: Likewise.
22715 * doc/md.texi: Likewise.
22716 * doc/objc.texi: Likewise.
22718 2017-01-10 Joshua Conner <joshconner@google.com>
22720 * config/arm/fuchsia-elf.h: New file.
22721 * config/fuchsia.h: New file.
22722 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
22723 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
22725 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
22727 2016-01-10 Richard Biener <rguenther@suse.de>
22729 PR tree-optimization/79034
22730 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
22731 Propagate out degenerate PHIs in the joiner.
22733 2017-01-10 Martin Liska <mliska@suse.cz>
22735 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
22736 (sort_congruence_classes_by_decl_uid): Likewise.
22737 (sort_congruence_class_groups_by_decl_uid): Likewise.
22738 (sem_item_optimizer::merge_classes): Sort class, groups in these
22739 classes and members in the groups by DECL_UID of declarations.
22740 This would make merge operations stable.
22742 2017-01-10 Martin Liska <mliska@suse.cz>
22744 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
22745 usage of m_classes_vec.
22746 (sem_item_optimizer::~sem_item_optimizer): Likewise.
22747 (sem_item_optimizer::get_group_by_hash): Likewise.
22748 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
22749 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
22750 (sem_item_optimizer::verify_classes): Likewise.
22751 (sem_item_optimizer::process_cong_reduction): Likewise.
22752 (sem_item_optimizer::dump_cong_classes): Likewise.
22753 (sem_item_optimizer::merge_classes): Likewise.
22754 * ipa-icf.h (congruence_class_hash): Rename from
22755 congruence_class_group_hash. Remove declaration of m_classes_vec.
22757 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
22759 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
22760 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
22761 * config.gcc: Add avx512vpopcntdqintrin.h.
22762 * config/i386/avx512vpopcntdqintrin.h: New.
22763 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
22764 * config/i386/i386-builtin-types.def: Add new types.
22765 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
22766 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
22767 __builtin_ia32_vpopcountq_v8di_mask): New.
22768 * config/i386/i386-c.c (ix86_target_macros_internal): Define
22769 __AVX512VPOPCNTDQ__.
22770 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
22771 (PTA_AVX512VPOPCNTDQ): Define.
22772 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
22773 TARGET_AVX512VPOPCNTDQ_P): Define.
22774 * config/i386/i386.opt: Add mavx512vpopcntdq.
22775 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
22776 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
22778 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22780 PR middle-end/77484
22781 * predict.def (PRED_CALL): Set to 67.
22783 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
22785 * expr.c (store_field): In the bitfield case, if the value comes from
22786 a function call and is of an aggregate type returned in registers, do
22787 not modify the field mode; extract the value in all cases if the mode
22788 is BLKmode and the size is not larger than a word.
22790 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
22793 * config/i386/cpuid.h: Fix undefined behavior.
22795 2017-01-04 Jeff Law <law@redhat.com>
22797 PR tree-optimization/79007
22798 PR tree-optimization/67955
22799 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
22800 conservative for pt.null when flag_non_call_exceptions is on.
22802 2017-01-09 Jakub Jelinek <jakub@redhat.com>
22804 PR translation/79019
22805 PR translation/79020
22806 * params.def (PARAM_INLINE_MIN_SPEEDUP,
22807 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
22808 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
22810 * config/avr/avr.opt (maccumulate-args): Likewise.
22811 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
22812 * common.opt (freport-bug): Likewise.
22813 * cif-code.def (CIF_FINAL_ERROR): Likewise.
22814 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
22815 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
22816 translatable string.
22817 * config/i386/i386.c (function_value_32): Likewise.
22818 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
22819 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
22821 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
22822 * common/config/msp430/msp430-common.c (msp430_handle_option):
22824 * symtab.c (symtab_node::verify_base): Likewise.
22825 * opts.c (set_debug_level): Likewise.
22826 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
22827 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
22828 missing whitespace to translatable strings.
22829 * config/avr/avr.md (bswapsi2): Fix typo in comment.
22830 * config/sh/superh.h: Likewise.
22831 * config/i386/xopintrin.h: Likewise.
22832 * config/i386/znver1.md: Likewise.
22833 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
22834 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
22835 * double-int.h (struct double_int): Likewise.
22836 * double-int.c (div_and_round_double): Likewise.
22837 * wide-int.cc: Likewise.
22838 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
22839 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
22840 * cfgcleanup.c (crossjumps_occured): Renamed to ...
22841 (crossjumps_occurred): ... this.
22842 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
22845 PR tree-optimization/78899
22846 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
22847 returning bool return struct loop *, NULL for failure and the new
22849 (versionable_outer_loop_p): Don't version outer loop if it has
22850 dont_vectorized bit set.
22851 (tree_if_conversion): When versioning outer loop, ensure
22852 tree_if_conversion is performed also on the inner loop of the
22853 non-vectorizable outer loop copy.
22854 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
22855 LOOP_VECTORIZED in inner loop of the scalar outer loop and
22856 prevent vectorization of it.
22857 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
22858 the outer loop vectorization of the non-scalar version is attempted
22859 before vectorization of the inner loop in scalar version. If
22860 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
22861 vectorization of its inner loop.
22862 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
22863 has 2 inner loops, rename also on edges from bb whose single pred
22864 is outer_loop->header. Fix typo in function comment.
22866 2017-01-09 Martin Sebor <msebor@redhat.com>
22869 * asan.c (asan_emit_stack_protection): Increase local buffer size
22870 to avoid snprintf truncation warning.
22872 2017-01-09 Andrew Pinski <apinski@cavium.com>
22874 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
22875 to reference thunderx2t99 for the tuning structure
22876 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
22878 (thunderx2t99_extra_costs): This.
22879 * config/aarch64/aarch64-tune.md: Regenerate.
22880 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
22881 (vulcan_addrcost_table): This.
22882 (vulcan_regmove_cost): Rename to ...
22883 (thunderx2t99_regmove_cost): This.
22884 (vulcan_vector_cost): Rename to ...
22885 (thunderx2t99_vector_cost): this.
22886 (vulcan_branch_cost): Rename to ...
22887 (thunderx2t99_branch_cost): This.
22888 (vulcan_tunings): Rename to ...
22889 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
22890 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
22892 2017-01-09 Martin Jambor <mjambor@suse.cz>
22896 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
22897 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
22898 (propagate_vr_accross_jump_function): Use the above function for all
22899 value range computations for pass-through jump functions and type
22900 converasion from explicit value range values.
22901 (ipcp_propagate_stage): Do not attempt to deduce types of formal
22902 parameters from TYPE_ARG_TYPES.
22903 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
22904 (ipa_write_node_info): Stream type of the actual argument.
22905 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
22907 2017-01-09 Martin Liska <mliska@suse.cz>
22910 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
22911 (lookup_compiler): Do not show error message with have_E.
22913 2017-01-09 Jakub Jelinek <jakub@redhat.com>
22915 PR tree-optimization/78938
22916 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
22917 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
22918 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
22919 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
22922 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22924 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
22927 2017-01-09 Richard Biener <rguenther@suse.de>
22929 PR tree-optimization/78997
22930 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
22931 name condition properly.
22933 2017-01-09 Richard Biener <rguenther@suse.de>
22936 * dwarf2out.c (is_cxx): New overload with context.
22937 (is_naming_typedef_decl): Use it.
22939 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22941 * invoke.texi (Option Summary): Correct spacing in option lists
22942 and add line breaks to fix over-long lines.
22944 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22946 PR middle-end/17660
22948 * extend.texi (Common Variable Attributes): Add xref to GCC
22949 Internals manual to explain mode attribute keywords.
22951 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
22954 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
22955 and Preprocessor Options.
22956 (Options for Linking): Document -pthread here....
22957 (RS/6000 and PowerPC Options): ...not here.
22958 (Solaris 2 Options): ...or here.
22959 * doc/cppopts.texi: Document -pthread.
22961 2017-01-08 Martin Sebor <msebor@redhat.com>
22963 PR middle-end/77708
22964 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
22965 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
22966 New member functions.
22967 (format_directive): Used them.
22969 (pass_sprintf_length::handle_gimple_call): Same.
22970 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
22971 to avoid truncation for any argument.
22972 (extract_affine_mul): Same.
22973 * tree.c (get_file_function_name): Same.
22975 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
22977 PR middle-end/77484
22978 * predict.def (PRED_INDIR_CALL): Set to 86.
22980 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22982 PR preprocessor/54124
22983 * doc/cppopts.texi: Reformat -d subtable to list the full name
22984 of the options. Add cross-reference to the docs for the general
22985 compiler -d options.
22986 * doc/invoke.texi (Developer Options): Add cross-reference to the
22987 preprocessor-specific -d option documentation.
22989 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22991 PR preprocessor/13498
22992 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
22993 redudant material, and reflect new command-line options.
22994 (System Headers): Likewise.
22996 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
22998 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
22999 -isystem, and -idirafter. Copy-edit.
23000 * doc/cppopts.texi: Copy-edit. Remove contradiction about
23001 default for -ftrack-macro-expansion. Delete obsolete and
23002 badly-formatted implementation details about -fdebug-cpp output.
23003 * doc/cppwarnopts.texi: Copy-edit.
23005 2017-01-07 David Malcolm <dmalcolm@redhat.com>
23008 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
23009 that the transition from a max line width >= 1<<10 to narrower
23010 lines works correctly.
23012 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
23014 * doc/options.texi (PerFunction): New.
23015 * opt-functions.awk (switch_flags): Map both Optimization and
23016 PerFunction to CL_OPTIMIZATION.
23017 * opth-gen.awk: Test for PerFunction flag along with
23019 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
23020 it only when the latter is present. Skip those that don't in
23021 the hash function generator.
23022 * common.opt (fvar-tracking): Mark as PerFunction instead of
23024 (fvar-tracking-assignments): Likewise.
23025 (fvar-tracking-assignments-toggle): Likewise.
23026 (fvar-tracking-uninit): Likewise.
23028 2017-01-07 Jakub Jelinek <jakub@redhat.com>
23030 PR translation/79018
23031 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
23034 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
23037 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
23038 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
23039 TARGET_LONG_JUMP_TABLE_OFFSETS.
23040 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
23041 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
23042 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
23043 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
23044 * config/m68k/m68k.md (tablejump expander): Likewise.
23045 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
23046 TARGET_LONG_JUMP_TABLE_OFFSETS.
23047 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
23048 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
23050 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
23051 David Holsgrove <david.holsgrove@xilinx.com>
23053 * common/config/microblaze/microblaze-common.c
23054 (TARGET_EXCEPT_UNWIND_INFO): Remove.
23055 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
23057 * config/microblaze/microblaze.c (microblaze_must_save_register)
23058 (microblaze_expand_epilogue, microblaze_return_addr): Handle
23060 (microblaze_eh_return): New function.
23061 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
23062 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
23063 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
23064 * config/microblaze/microblaze.md (eh_return): New pattern.
23066 2017-01-06 Jakub Jelinek <jakub@redhat.com>
23068 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
23069 GCC_DIAGNOSTIC_STRINGIFY): Define.
23071 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
23073 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23075 * config/arm/arm.md (<mcrr>): New.
23077 * config/arm/arm.c (arm_arch5te): New.
23078 (arm_option_override): Set arm_arch5te.
23079 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
23081 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
23082 (arm_mcrr_qualifiers): ... this. New.
23083 (MRRC_QUALIFIERS): Define to...
23084 (arm_mrrc_qualifiers): ... this. New.
23085 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
23087 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
23088 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
23089 (MRRCI, mrrc, MRRC): New.
23090 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
23091 VUNSPEC_MRRC2): New.
23093 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23095 * config/arm/arm.md (<mcr>): New.
23097 * config/arm/arm.c (arm_coproc_builtin_available): Add
23098 support for mcr, mrc, mcr2 and mrc2.
23099 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
23100 (arm_mcr_qualifiers): ... this. New.
23101 (MRC_QUALIFIERS): Define to ...
23102 (arm_mrc_qualifiers): ... this. New.
23103 (MCR_QUALIFIERS): Define to ...
23104 (arm_mcr_qualifiers): ... this. New.
23105 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
23107 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
23108 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
23109 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
23110 VUNSPEC_MRC2): New.
23112 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23114 * config/arm/arm.md (*ldc): New.
23118 * config/arm/arm.c (arm_coproc_builtin_available): Add
23119 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
23120 (arm_coproc_ldc_stc_legitimate_address): New.
23121 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
23122 'qualifier_const_pointer'.
23123 (LDC_QUALIFIERS): Define to...
23124 (arm_ldc_qualifiers): ... this. New.
23125 (STC_QUALIFIERS): Define to...
23126 (arm_stc_qualifiers): ... this. New.
23127 * config/arm/arm-protos.h
23128 (arm_coproc_ldc_stc_legitimate_address): New.
23129 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
23130 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
23131 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
23132 stc2, stcl, stc2l): New.
23133 * config/arm/constraints.md (Uz): New.
23134 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
23135 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
23136 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
23137 VUNSPEC_STC2L): New.
23139 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23141 * config/arm/arm.md (<cdp>): New.
23142 * config/arm/arm.c (neon_const_bounds): Rename this ...
23143 (arm_const_bounds): ... this.
23144 (arm_coproc_builtin_available): New.
23145 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
23146 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
23147 (CDP_QUALIFIERS): Define to...
23148 (arm_cdp_qualifiers): ... this. New.
23150 (arm_expand_builtin_args): Add case for 6 arguments.
23151 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
23152 (arm_const_bounds): ... this.
23153 (arm_coproc_builtin_available): New.
23154 * config/arm/arm_acle.h (__arm_cdp): New.
23156 * config/arm/arm_acle_builtins.def (cdp): New.
23158 * config/arm/iterators.md (CDPI,CDP,cdp): New.
23159 * config/arm/neon.md: Rename all 'neon_const_bounds' to
23160 'arm_const_bounds'.
23161 * config/arm/types.md (coproc): New.
23162 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
23163 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
23164 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
23165 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
23167 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23169 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
23170 (UBINOP_QUALIFIERS): New.
23172 (acle_builtin_data): New. Change comment.
23173 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
23174 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
23175 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
23176 arm_acle_builtins.def.
23177 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
23178 (arm_init_acle_builtins): New.
23179 (CRC32_BUILTIN): Remove.
23180 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
23181 crc32cb, crc32ch and crc32cw.
23182 (arm_init_crc32_builtins): Remove.
23183 (arm_init_builtins): Use arm_init_acle_builtins rather
23184 than arm_init_crc32_builtins.
23185 (arm_expand_acle_builtin): New.
23186 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
23187 * config/arm/arm_acle_builtins.def: New.
23189 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23191 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
23192 (arm_builtin_datum): ... this.
23193 (arm_init_neon_builtin): Rename to ...
23194 (arm_init_builtin): ... this. Add a new parameters PREFIX
23195 and USE_SIG_IN_NAME.
23196 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
23197 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
23198 'arm_builtin_datum'.
23199 (arm_init_vfp_builtins): Likewise.
23200 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
23201 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
23202 (arm_expand_neon_args): Rename to ...
23203 (arm_expand_builtin_args): ... this. Rename builtin_arg
23204 enum values and differentiate between ARG_BUILTIN_MEMORY
23205 and ARG_BUILTIN_NEON_MEMORY.
23206 (arm_expand_neon_builtin_1): Rename to ...
23207 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
23208 values, arm_expand_builtin_args and add bool parameter NEON.
23209 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
23210 (arm_expand_vfp_builtin): Likewise.
23211 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
23213 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23215 PR middle-end/77484
23216 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
23217 * predict.c (tree_estimate_probability_bb): Reverse direction of
23218 polymorphic call predictor.
23220 2017-01-06 David Malcolm <dmalcolm@redhat.com>
23222 * passes.c (execute_one_pass): Split out pass-skipping logic into...
23223 (determine_pass_name_match): ...this new function and...
23224 (should_skip_pass_p): ...this new function.
23226 2017-01-06 Nathan Sidwell <nathan@acm.org>
23228 * ipa-visibility.c (function_and_variable_visibility): Reformat
23229 comments and long lines. Remove extrneous if.
23230 * symtab.c (symtab_node::make_decl_local): Fix code format.
23231 (symtab_node::set_section_for_node): Fix comment typo.
23233 2017-01-06 Martin Liska <mliska@suse.cz>
23236 * lra-constraints.c: Rename invariant to lra_invariant.
23237 * predict.c (set_even_probabilities): Initialize e to NULL.
23239 2017-01-05 Martin Sebor <msebor@redhat.com>
23241 PR tree-optimization/78910
23242 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
23243 (format_integer): Correct off-by-one error in the handling
23244 of precision with negative numbers in signed conversions..
23246 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
23248 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
23250 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23252 PR tree-optimization/71016
23253 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
23254 factor_out_conditional_conversion. Formatting fix.
23255 (factor_out_conditional_conversion): Add cond_stmt argument.
23256 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
23257 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
23260 2017-01-05 David Malcolm <dmalcolm@redhat.com>
23262 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
23263 read-rtl-function.o, and selftest-rtl.o.
23264 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
23265 (selftest::aarch64_test_loading_full_dump): New function.
23266 (selftest::aarch64_run_selftests): New function.
23267 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
23268 selftest::aarch64_run_selftests.
23269 * config/i386/i386.c
23270 (selftest::ix86_test_loading_dump_fragment_1): New function.
23271 (selftest::ix86_test_loading_call_insn): New function.
23272 (selftest::ix86_test_loading_full_dump): New function.
23273 (selftest::ix86_test_loading_unspec): New function.
23274 (selftest::ix86_run_selftests): Call the new functions.
23275 * emit-rtl.c (maybe_set_max_label_num): New function.
23276 * emit-rtl.h (maybe_set_max_label_num): New decl.
23277 * function.c (instantiate_decls): Guard call to
23278 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
23279 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
23281 * gensupport.c (gen_reader::gen_reader): Pass "false"
23282 for new "compact" param of rtx_reader.
23283 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
23284 rather than an empty string for NULL strings.
23285 * read-md.c: Potentially include config.h rather than bconfig.h.
23286 Wrap include of errors.h with #ifdef GENERATOR_FILE.
23287 (have_error): New global, copied from errors.c.
23288 (md_reader::read_name): Rename to...
23289 (md_reader::read_name_1): ...this, adding "out_loc" param,
23290 and converting "missing name or number" to returning false, rather
23292 (md_reader::read_name): Reimplement in terms of read_name_1.
23293 (md_reader::read_name_or_nil): New function.
23294 (md_reader::read_string): Handle "(nil)" by returning NULL.
23295 (md_reader::md_reader): Add new param "compact".
23296 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
23297 (md_reader::read_file): New method.
23298 * read-md.h (md_reader::md_reader): Add new param "compact".
23299 (md_reader::read_file): New method.
23300 (md_reader::is_compact): New accessor.
23301 (md_reader::read_name): Convert return type from void to file_location.
23302 (md_reader::read_name_or_nil): New decl.
23303 (md_reader::read_name_1): New decl.
23304 (md_reader::m_compact): New field.
23305 (noop_reader::noop_reader): Pass "false" for new "compact" param
23307 (rtx_reader::rtx_reader): Add new "compact" param.
23308 (rtx_reader::read_rtx_operand): Make virtual and convert return
23309 type from void to rtx.
23310 (rtx_reader::read_until): New decl.
23311 (rtx_reader::handle_any_trailing_information): New virtual function.
23312 (rtx_reader::postprocess): New virtual function.
23313 (rtx_reader::finalize_string): New virtual function.
23314 (rtx_reader::m_in_call_function_usage): New field.
23315 (rtx_reader::m_reuse_rtx_by_id): New field.
23316 * read-rtl-function.c: New file.
23317 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
23318 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
23319 (selftest::verify_three_block_rtl_cfg): New decl.
23320 * read-rtl-function.h: New file.
23321 * read-rtl.c: Potentially include config.h rather than bconfig.h.
23322 For host, include function.h, memmodel.h, and emit-rtl.h.
23323 (one_time_initialization): New function.
23324 (struct compact_insn_name): New struct.
23325 (compact_insn_names): New array.
23326 (find_code): Handle insn codes in compact dumps.
23327 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
23328 (bind_subst_iter_and_attr): Likewise.
23329 (add_condition_to_string): Likewise.
23330 (add_condition_to_rtx): Likewise.
23331 (apply_attribute_uses): Likewise.
23332 (add_current_iterators): Likewise.
23333 (apply_iterators): Likewise.
23334 (initialize_iterators): Guard usage of apply_subst_iterator with
23335 #ifdef GENERATOR_FILE.
23336 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
23337 (md_reader::read_mapping): Likewise.
23338 (add_define_attr_for_define_subst): Likewise.
23339 (add_define_subst_attr): Likewise.
23340 (read_subst_mapping): Likewise.
23341 (check_code_iterator): Likewise.
23342 (rtx_reader::read_rtx): Likewise. Move one-time initialization
23344 (one_time_initialization): New function.
23345 (rtx_reader::read_until): New method.
23346 (read_flags): New function.
23347 (parse_reg_note_name): New function.
23348 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
23349 Handle reuse_rtx ids.
23350 Wrap iterator lookup within #ifdef GENERATOR_FILE.
23351 Add parsing support for RTL dumps, mirroring the special-cases in
23352 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
23353 values, and calling handle_any_trailing_information.
23354 (rtx_reader::read_rtx_operand): Convert return type from void
23355 to rtx, returning return_rtx. Handle case 'e'. Call
23356 finalize_string on XSTR and XTMPL fields.
23357 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
23358 "(nil)" values were omitted. Call the postprocess vfunc on the
23360 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
23361 class ctor. Initialize m_in_call_function_usage. Call
23362 one_time_initialization.
23363 * rtl-tests.c (selftest::test_uncond_jump): Call
23364 set_new_first_and_last_insn.
23365 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
23366 * selftest-rtl.c: New file.
23367 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
23368 (selftest::get_insn_by_uid): New decl.
23369 * selftest-run-tests.c (selftest::run_tests): Call
23370 read_rtl_function_c_tests.
23371 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
23372 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
23375 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
23377 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
23378 operands in a special way. Assert that pos+len <= mode precision.
23380 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23382 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
23383 3 argument Alias with unlimited for the negative form.
23384 (fno-vect-cost-model): Removed.
23386 2017-01-05 Martin Liska <mliska@suse.cz>
23388 * hsa-gen.c (gen_hsa_divmod): New function.
23389 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
23391 2017-01-05 Martin Liska <mliska@suse.cz>
23394 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
23397 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23399 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
23400 small constant length operands.
23402 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23404 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
23405 between loop iterations.
23407 2017-01-05 Martin Liska <mliska@suse.cz>
23410 * gimplify.c (gimplify_decl_expr): Compare to
23411 asan_poisoned_variables instread of checking flags.
23412 (gimplify_target_expr): Likewise.
23413 (gimplify_expr): Likewise.
23414 (gimplify_function_tree): Conditionally initialize
23415 asan_poisoned_variables.
23417 2017-01-04 Jeff Law <law@redhat.com>
23419 PR tree-optimizatin/78812
23420 * rtl.h (contains_mem_rtx_p): Prototype.
23421 * ifcvt.c (containts_mem_rtx_p): Move from here to...
23422 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
23423 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
23424 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
23425 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
23427 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23429 * input.c (assert_char_at_range): Default-initialize actual_range.
23431 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23433 * df-scan.c (df_ref_create_structure): Make regno unsigned,
23434 to match the caller.
23436 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23438 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
23439 insns after final jump in test to emit dummy move.
23441 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23443 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
23444 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
23446 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23448 * multiple_target.c (create_dispatcher_calls): Init e_next.
23449 * tree-ssa-loop-split.c (split_loop): Init border.
23450 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
23453 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
23458 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
23459 (altivec_register_operand): Do not return true if the operand
23460 contains a SUBREG mixing SImode and SFmode.
23461 (vsx_register_operand): Likewise.
23462 (vsx_reg_sfsubreg_ok): New predicate.
23463 (vfloat_operand): Do not return true if the operand contains a
23464 SUBREG mixing SImode and SFmode.
23465 (vint_operand): Likewise.
23466 (vlogical_operand): Likewise.
23467 (gpc_reg_operand): Likewise.
23468 (int_reg_operand): Likewise.
23469 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
23470 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
23471 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
23473 (rs6000_emit_move_si_sf_subreg): New helper function.
23474 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
23475 fixup SUBREGs involving SImode and SFmode.
23476 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
23477 numbers for the new peephole2 optimization.
23478 (peephole2 for SFmode unions): New peephole2 to optimize cases in
23479 the GLIBC math library that do AND/IOR/XOR operations on single
23480 precision floating point.
23481 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
23482 target macros to say whether we need to avoid SUBREGs mixing
23484 (TARGET_ALLOW_SF_SUBREG): Likewise.
23485 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
23486 (UNSPEC_SI_FROM_SF): Likewise.
23487 (iorxor): Change spacing.
23488 (and_ior_xor): New iterator for AND, IOR, and XOR.
23489 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
23490 (movdi_from_sf_zero_ext): Likewise.
23491 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
23492 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
23493 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
23494 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
23495 (fms<mode>4): Likewise.
23496 (fnma<mode>4): Likewise.
23497 (fnms<mode>4): Likewise.
23498 (nfma<mode>4): Likewise.
23499 (nfms<mode>4): Likewise.
23501 2017-01-04 Marek Polacek <polacek@redhat.com>
23504 * doc/invoke.texi: Document -Wpointer-compare.
23506 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23508 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
23511 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
23512 descriptions for -gdwarf-5 and emit them as uleb128 instead of
23515 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
23518 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
23519 documentation of the powerpc_popcntb_ok attribute.
23520 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23521 code to issue warning messages if a requested CPU configuration is
23522 not supported by the binary (assembler and loader) toolchain.
23523 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
23524 made to define a built-in function that has been disabled.
23525 (paired_init_builtins): Add assertion to prevent ICE if attempt is
23526 made to define a built-in function that has been disabled.
23527 (altivec_init_builtins): Add comment explaining why definition
23528 of the DST built-in functions is not preceded by an assertion
23529 check. Add assertions to prevent ICE if attempts are made to
23530 define an altivec predicate or an abs* built-in function that has
23532 (htm_init_builtins): Add comment explaining why definition of the
23533 htm built-in functions is not preceded by an assertion check.
23535 2017-01-04 Jeff Law <law@redhat.com>
23537 PR tree-optimizatin/67955
23538 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
23539 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
23540 the points-to solution does not include pt_null. Use DECL_PT_UID
23543 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
23545 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
23546 Use gen_int_mode instead of gen_lopwart for const_int operands.
23548 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23550 PR tree-optimization/71563
23551 * match.pd: Simplify X << Y into X if Y is known to be 0 or
23552 out of range value - has low bits known to be zero.
23554 2017-01-04 Alan Modra <amodra@gmail.com>
23556 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
23557 * configure: Regenerate.
23558 * config.in: Regenerate.
23560 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23563 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
23564 a substring of the message, but strcmp with the whole message. Ifdef
23565 ENABLE_NLS, translate the message first using dgettext.
23567 2017-01-03 Jeff Law <law@redhat.com>
23569 PR tree-optimizatin/78856
23570 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
23571 (mark_threaded_blocks): Remove code to truncate thread paths that
23572 cross multiple loop headers. Instead invalidate the cached loop
23573 iteration information and handle case of a thread path walking
23574 into an irreducible region.
23576 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
23579 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
23580 assertions. Add support for doing the signbit if the IEEE 128-bit
23581 floating point value is in a GPR.
23582 * config/rs6000/rs6000.md (Fsignbit): Delete.
23583 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
23584 Update the length attribute if the value is in a GPR.
23585 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
23586 the sign or zero extension instruction, since the value is always 0/1.
23587 (signbit<mode>2_dm2): Delete using <Fsignbit>.
23590 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
23591 extracting SImode to a GPR register so that we can generate a
23592 store, limit the vector to be in a traditional Altivec register
23593 for the vextuwrx instruction.
23595 2017-01-03 Ian Lance Taylor <iant@google.com>
23597 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
23599 2017-01-03 Martin Sebor <msebor@redhat.com>
23601 PR tree-optimization/78696
23602 * gimple-ssa-sprintf.c (format_floating): Correct handling of
23603 precision. Use MPFR for %f for greater fidelity. Correct handling
23605 (pass_sprintf_length::compute_format_length): Set width and precision
23606 specified by asrerisk to void_node for vararg functions.
23607 (try_substitute_return_value): Adjust dump output.
23609 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
23611 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
23613 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
23615 * doc/invoke.texi (SPARC options): Document -mlra as the default.
23616 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
23617 -mlra/-mno-lra was passed to the compiler.
23619 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
23621 PR rtl-optimization/65618
23622 * emit-rtl.c (try_split): Move initialization of "before" and
23623 "after" to just before the call to emit_insn_after_setloc.
23625 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
23627 * doc/md.texi (Standard Names): Remove reference to Java frontend.
23629 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
23631 * dwarf2out.c (gen_enumeration_type_die): When
23632 -gno-strict-dwarf, add a DW_AT_encoding attribute.
23634 2017-01-03 Jakub Jelinek <jakub@redhat.com>
23636 PR tree-optimization/78965
23637 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
23638 Change first argument from const call_info & to call_info &. For %n
23639 set info.nowrite to false.
23641 PR middle-end/78901
23642 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
23643 possibly throwing calls.
23645 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
23646 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
23647 and fns handling, rather than in a separate case SSA_NAME.
23649 2017-01-02 Jeff Law <law@redhat.com>
23651 * config/darwin-driver.c (darwin_driver_init): Const-correctness
23652 fixes for first_period and second_period variables.
23654 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
23657 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
23658 (*insvqi_1): New insn pattern.
23659 (*insvqi_1_mem_rex64): Ditto.
23660 (*insvqi_2): Ditto.
23661 (*insvqi_3): Rename from *insvqi.
23663 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
23665 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
23667 * doc/cfg.texi (Edges): Remove reference to Java.
23668 (Maintaining the CFG): Ditto.
23670 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23672 PR middle-end/77674
23673 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
23674 transparent aliases.
23676 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23678 PR middle-end/77484
23679 * predict.def (PRED_CALL): Update hitrate.
23680 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
23681 * predict.c (tree_estimate_probability_bb): Split CALL predictor
23682 into direct/indirect/polymorphic variants.
23684 2017-01-01 Jakub Jelinek <jakub@redhat.com>
23686 Update copyright years.
23688 * gcc.c (process_command): Update copyright notice dates.
23689 * gcov-dump.c (print_version): Ditto.
23690 * gcov.c (print_version): Ditto.
23691 * gcov-tool.c (print_version): Ditto.
23692 * gengtype.c (create_file): Ditto.
23693 * doc/cpp.texi: Bump @copying's copyright year.
23694 * doc/cppinternals.texi: Ditto.
23695 * doc/gcc.texi: Ditto.
23696 * doc/gccint.texi: Ditto.
23697 * doc/gcov.texi: Ditto.
23698 * doc/install.texi: Ditto.
23699 * doc/invoke.texi: Ditto.
23701 Copyright (C) 2017 Free Software Foundation, Inc.
23703 Copying and distribution of this file, with or without modification,
23704 are permitted in any medium without royalty provided the copyright
23705 notice and this notice are preserved.