1 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
3 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
4 on the entire header of the finally block in the fallthru case.
6 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
8 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
12 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
13 rs6000_emit_move_from_cr and call renamed function.
14 (rs6000_emit_prologue): Call renamed functions.
15 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
16 movesi_from_cr, remove volatile CRs.
18 2017-08-30 Jon Beniston <jon@beniston.com>
19 Richard Biener <rguenther@suse.de>
21 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
22 of VECTOR_MODE_P check.
23 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
26 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
28 * df.h (df_read_modify_subreg_p): Remove in favor of...
29 * rtl.h (read_modify_subreg_p): ...this new function. Take a
30 const_rtx instead of an rtx.
31 * cprop.c (local_cprop_find_used_regs): Update accordingly.
32 * df-problems.c (df_word_lr_mark_ref): Likewise.
33 * ira-lives.c (mark_pseudo_reg_live): Likewise.
34 (mark_pseudo_reg_dead): Likewise.
35 (mark_ref_dead): Likewise.
36 * reginfo.c (init_subregs_of_mode): Likewise.
37 * sched-deps.c (sched_analyze_1): Likewise.
38 * df-scan.c (df_def_record_1): Likewise.
39 (df_uses_record): Likewise.
40 (df_read_modify_subreg_p): Remove in favor of...
41 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
42 const_rtx instead of an rtx.
44 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
45 Alan Hayward <alan.hayward@arm.com>
46 David Sherwood <david.sherwood@arm.com>
48 * rtl.h (partial_subreg_p): New function.
49 * caller-save.c (save_call_clobbered_regs): Use it.
50 * calls.c (expand_call): Likewise.
51 * combine.c (combinable_i3pat): Likewise.
52 (simplify_set): Likewise.
53 (make_extraction): Likewise.
54 (make_compound_operation_int): Likewise.
55 (gen_lowpart_or_truncate): Likewise.
56 (force_to_mode): Likewise.
57 (make_field_assignment): Likewise.
58 (reg_truncated_to_mode): Likewise.
59 (record_truncated_value): Likewise.
60 (move_deaths): Likewise.
61 * cse.c (record_jump_cond): Likewise.
63 * cselib.c (cselib_lookup_1): Likewise.
64 * expmed.c (extract_bit_field_using_extv): Likewise.
65 * function.c (assign_parm_setup_reg): Likewise.
66 * ifcvt.c (noce_convert_multiple_sets): Likewise.
67 * ira-build.c (create_insn_allocnos): Likewise.
68 * lra-coalesce.c (merge_pseudos): Likewise.
69 * lra-constraints.c (match_reload): Likewise.
70 (simplify_operand_subreg): Likewise.
71 (curr_insn_transform): Likewise.
72 * lra-lives.c (process_bb_lives): Likewise.
73 * lra.c (new_insn_reg): Likewise.
74 (lra_substitute_pseudo): Likewise.
75 * regcprop.c (mode_change_ok): Likewise.
76 (maybe_mode_change): Likewise.
77 (copyprop_hardreg_forward_1): Likewise.
78 * reload.c (push_reload): Likewise.
79 (find_reloads): Likewise.
80 (find_reloads_subreg_address): Likewise.
81 * reload1.c (alter_reg): Likewise.
82 (eliminate_regs_1): Likewise.
83 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
85 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
87 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
88 back to if statements, including unpack.
90 2017-08-30 Martin Liska <mliska@suse.cz>
93 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
95 (func_checker::compare_asm_inputs_outputs): ... this function.
96 (func_checker::compare_gimple_asm): Use the function to compare
98 * ipa-icf-gimple.h: Rename the function.
100 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
101 Alan Hayward <alan.hayward@arm.com>
102 David Sherwood <david.sherwood@arm.com>
104 * coretypes.h (complex_mode): New type.
105 * gdbhooks.py (build_pretty_printer): Handle it.
106 * machmode.h (complex_mode): New class.
107 (complex_mode::includes_p): New function.
108 (is_complex_int_mode): Likewise.
109 (is_complex_float_mode): Likewise.
110 * genmodes.c (get_mode_class): Handle complex mode classes.
111 * function.c (expand_function_end): Use is_complex_int_mode.
113 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
114 Alan Hayward <alan.hayward@arm.com>
115 David Sherwood <david.sherwood@arm.com>
117 * coretypes.h (scalar_mode_pod): New typedef.
118 * gdbhooks.py (build_pretty_printer): Handle it.
119 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
120 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
121 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
122 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
125 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
126 Alan Hayward <alan.hayward@arm.com>
127 David Sherwood <david.sherwood@arm.com>
129 * machmode.h (mode_for_vector): Take a scalar_mode instead
131 * stor-layout.c (mode_for_vector): Likewise.
132 * explow.c (promote_mode): Use as_a <scalar_mode>.
133 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
135 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
136 Alan Hayward <alan.hayward@arm.com>
137 David Sherwood <david.sherwood@arm.com>
139 * target.def (preferred_simd_mode): Take a scalar_mode
140 instead of a machine_mode.
141 * targhooks.h (default_preferred_simd_mode): Likewise.
142 * targhooks.c (default_preferred_simd_mode): Likewise.
143 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
144 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
145 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
146 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
147 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
148 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
149 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
150 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
152 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
153 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
154 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
155 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
156 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
157 * doc/tm.texi: Regenerate.
158 * optabs-query.c (can_vec_mask_load_store_p): Return false for
161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
162 Alan Hayward <alan.hayward@arm.com>
163 David Sherwood <david.sherwood@arm.com>
165 * target.def (scalar_mode_supported_p): Take a scalar_mode
166 instead of a machine_mode.
167 * targhooks.h (default_scalar_mode_supported_p): Likewise.
168 * targhooks.c (default_scalar_mode_supported_p): Likewise.
169 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
170 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
171 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
172 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
173 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
174 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
175 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
176 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
177 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
178 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
179 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
180 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
182 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
183 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
184 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
185 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
186 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
188 * doc/tm.texi: Regenerate.
190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
191 Alan Hayward <alan.hayward@arm.com>
192 David Sherwood <david.sherwood@arm.com>
194 * coretypes.h (opt_scalar_mode): New typedef.
195 * gdbhooks.py (build_pretty_printers): Handle it.
196 * machmode.h (mode_iterator::get_2xwider): Add overload for
198 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
200 * expr.c (convert_mode_scalar): Likewise.
201 * omp-low.c (omp_clause_aligned_alignment): Likewise.
202 * optabs.c (expand_float): Likewise.
203 (expand_fix): Likewise.
204 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
206 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
207 Alan Hayward <alan.hayward@arm.com>
208 David Sherwood <david.sherwood@arm.com>
210 * optabs.c (expand_float): Explicitly check for scalars before
211 using a branching expansion.
212 (expand_fix): Likewise.
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 * expr.c (convert_mode): Split scalar handling out into...
219 (convert_mode_scalar): ...this new function. Treat the modes
222 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
223 Alan Hayward <alan.hayward@arm.com>
224 David Sherwood <david.sherwood@arm.com>
226 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
228 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
231 Alan Hayward <alan.hayward@arm.com>
232 David Sherwood <david.sherwood@arm.com>
234 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
235 rather than a machine_mode.
236 (fixed_from_string): Likewise.
237 (fixed_convert): Likewise.
238 (fixed_convert_from_int): Likewise.
239 (fixed_convert_from_real): Likewise.
240 (real_convert_from_fixed): Likewise.
241 * fixed-value.c (fixed_from_double_int): Likewise.
242 (fixed_from_string): Likewise.
243 (fixed_convert): Likewise.
244 (fixed_convert_from_int): Likewise.
245 (fixed_convert_from_real): Likewise.
246 (real_convert_from_fixed): Likewise.
247 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
249 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
250 Alan Hayward <alan.hayward@arm.com>
251 David Sherwood <david.sherwood@arm.com>
253 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
254 of separate mode class checks. Do not allow vector modes here.
255 (immed_wide_int_const): Use as_a <scalar_mode>.
256 * explow.c (trunc_int_for_mode): Likewise.
257 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
258 (wi::shwi): Likewise.
259 (wi::min_value): Likewise.
260 (wi::max_value): Likewise.
261 * dwarf2out.c (loc_descriptor): Likewise.
262 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
265 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
266 Alan Hayward <alan.hayward@arm.com>
267 David Sherwood <david.sherwood@arm.com>
269 * tree.h (SCALAR_TYPE_MODE): New macro.
270 * expr.c (expand_expr_addr_expr_1): Use it.
271 (expand_expr_real_2): Likewise.
272 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
273 (fold_convert_const_fixed_from_int): Likewise.
274 (fold_convert_const_fixed_from_real): Likewise.
275 (native_encode_fixed): Likewise
276 (native_encode_complex): Likewise
277 (native_encode_vector): Likewise.
278 (native_interpret_fixed): Likewise.
279 (native_interpret_real): Likewise.
280 (native_interpret_complex): Likewise.
281 (native_interpret_vector): Likewise.
282 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
283 (simd_clone_adjust_argument_types): Likewise.
284 (simd_clone_init_simd_arrays): Likewise.
285 (simd_clone_adjust): Likewise.
286 * stor-layout.c (layout_type): Likewise.
287 * tree.c (build_minus_one_cst): Likewise.
288 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
289 * tree-inline.c (estimate_move_cost): Likewise.
290 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
291 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
292 (vectorizable_reduction): Likewise.
293 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
294 (vect_recog_mixed_size_cond_pattern): Likewise.
295 (check_bool_pattern): Likewise.
296 (adjust_bool_pattern): Likewise.
297 (search_type_for_mask_1): Likewise.
298 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
299 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
300 (vectorizable_load): Likewise.
301 (vectorizable_store): Likewise.
302 * ubsan.c (ubsan_encode_value): Likewise.
303 * varasm.c (output_constant): Likewise.
305 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
306 Alan Hayward <alan.hayward@arm.com>
307 David Sherwood <david.sherwood@arm.com>
309 * coretypes.h (scalar_mode): New class.
310 * machmode.h (scalar_mode): Likewise.
311 (scalar_mode::includes_p): New function.
312 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
313 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
314 * genmodes.c (get_mode_class): Handle remaining scalar modes.
315 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
316 * expmed.c (store_bit_field_1): Likewise.
317 (extract_bit_field_1): Likewise.
318 * expr.c (write_complex_part): Likewise.
319 (read_complex_part): Likewise.
320 (emit_move_complex_push): Likewise.
321 (expand_expr_real_2): Likewise.
322 * function.c (assign_parm_setup_reg): Likewise.
323 (assign_parms_unsplit_complex): Likewise.
324 * optabs.c (expand_binop): Likewise.
325 * rtlanal.c (subreg_get_info): Likewise.
326 * simplify-rtx.c (simplify_immed_subreg): Likewise.
327 * varasm.c (output_constant_pool_2): Likewise.
329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
330 Alan Hayward <alan.hayward@arm.com>
331 David Sherwood <david.sherwood@arm.com>
333 * expmed.c (extract_high_half): Use scalar_int_mode and remove
335 (expmed_mult_highpart_optab): Likewise.
336 (expmed_mult_highpart): Likewise.
338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
339 Alan Hayward <alan.hayward@arm.com>
340 David Sherwood <david.sherwood@arm.com>
342 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
343 instead of a machine_mode.
344 (builtin_memset_read_str): Likewise.
345 * builtins.c (c_readstr): Likewise.
346 (builtin_memcpy_read_str): Likewise.
347 (builtin_strncpy_read_str): Likewise.
348 (builtin_memset_read_str): Likewise.
349 (builtin_memset_gen_str): Likewise.
350 (expand_builtin_signbit): Use scalar_int_mode for local variables.
351 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
352 instead of a machine_mode.
353 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
355 (make_extraction): Likewise.
356 (try_widen_shift_mode): Take and return scalar_int_modes instead
358 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
359 a scalar_int_mode instead of a machine_mode.
360 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
361 (avr_addr_space_pointer_mode): Likewise.
362 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
363 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
364 (msp430_unwind_word_mode): Likewise.
365 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
366 (spu_addr_space_pointer_mode): Likewise.
367 (spu_addr_space_address_mode): Likewise.
368 (spu_libgcc_cmp_return_mode): Likewise.
369 (spu_libgcc_shift_count_mode): Likewise.
370 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
371 (rl78_addr_space_pointer_mode): Likewise.
372 (fl78_unwind_word_mode): Likewise.
373 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
375 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
376 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
377 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
378 (mips_valid_pointer_mode): Likewise.
379 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
380 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
381 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
383 (ft32_addr_space_address_mode): Likewise.
384 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
385 scalar_int_mode instead of a machine_mode.
386 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
388 (m32c_addr_space_address_mode): Likewise.
389 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
390 (rs6000_eh_return_filter_mode): Likewise.
391 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
392 (rs6000_eh_return_filter_mode): Likewise.
393 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
394 (s390_libgcc_shift_count_mode): Likewise.
395 (s390_unwind_word_mode): Likewise.
396 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
398 * target.def (mode_rep_extended): Likewise.
399 (valid_pointer_mode): Likewise.
400 (addr_space.valid_pointer_mode): Likewise.
401 (eh_return_filter_mode): Return a scalar_int_mode rather than
403 (libgcc_cmp_return_mode): Likewise.
404 (libgcc_shift_count_mode): Likewise.
405 (unwind_word_mode): Likewise.
406 (addr_space.pointer_mode): Likewise.
407 (addr_space.address_mode): Likewise.
408 * doc/tm.texi: Regenerate.
409 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
411 (do_jump): Use scalar_int_mode for local variables.
412 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
413 rather than a machine_mode.
414 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
415 (scompare_loc_descriptor_wide): Likewise.
416 (scompare_loc_descriptor_narrow): Likewise.
417 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
419 * except.c (sjlj_emit_dispatch_table): Likewise.
420 (expand_builtin_eh_copy_values): Likewise.
421 * explow.c (convert_memory_address_addr_space_1): Likewise.
422 Take a scalar_int_mode rather than a machine_mode.
423 (convert_memory_address_addr_space): Take a scalar_int_mode rather
425 (memory_address_addr_space): Use scalar_int_mode for local variables.
426 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
427 rather than a machine_mode.
428 * expmed.c (mask_rtx): Likewise.
429 (init_expmed_one_conv): Likewise.
430 (expand_mult_highpart_adjust): Likewise.
431 (extract_high_half): Likewise.
432 (expmed_mult_highpart_optab): Likewise.
433 (expmed_mult_highpart): Likewise.
434 (expand_smod_pow2): Likewise.
435 (expand_sdiv_pow2): Likewise.
436 (emit_store_flag_int): Likewise.
437 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
439 (extract_low_bits): Likewise.
440 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
442 * expr.c (pieces_addr::adjust): Likewise.
443 (can_store_by_pieces): Likewise.
444 (store_by_pieces): Likewise.
445 (clear_by_pieces_1): Likewise.
446 (expand_expr_addr_expr_1): Likewise.
447 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
448 (expand_expr_real_1): Likewise.
449 (try_casesi): Likewise.
450 * final.c (shorten_branches): Likewise.
451 * fold-const.c (fold_convert_const_int_from_fixed): Change the
452 type of "mode" to machine_mode.
453 * internal-fn.c (expand_arith_overflow_result_store): Take a
454 scalar_int_mode rather than a machine_mode.
455 (expand_mul_overflow): Use scalar_int_mode for local variables.
456 * loop-doloop.c (doloop_modify): Likewise.
457 (doloop_optimize): Likewise.
458 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
460 (expand_doubleword_shift_condmove): Likewise.
461 (expand_doubleword_shift): Likewise.
462 (expand_doubleword_clz): Likewise.
463 (expand_doubleword_popcount): Likewise.
464 (expand_doubleword_parity): Likewise.
465 (expand_absneg_bit): Use scalar_int_mode for local variables.
466 (prepare_float_lib_cmp): Likewise.
467 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
468 rather than a machine_mode.
469 (convert_memory_address_addr_space): Likewise.
470 (get_mode_bounds): Likewise.
471 (get_address_mode): Return a scalar_int_mode rather than a
473 * rtlanal.c (get_address_mode): Likewise.
474 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
476 * targhooks.c (default_mode_rep_extended): Likewise.
477 (default_valid_pointer_mode): Likewise.
478 (default_addr_space_valid_pointer_mode): Likewise.
479 (default_eh_return_filter_mode): Return a scalar_int_mode rather
481 (default_libgcc_cmp_return_mode): Likewise.
482 (default_libgcc_shift_count_mode): Likewise.
483 (default_unwind_word_mode): Likewise.
484 (default_addr_space_pointer_mode): Likewise.
485 (default_addr_space_address_mode): Likewise.
486 * targhooks.h (default_eh_return_filter_mode): Likewise.
487 (default_libgcc_cmp_return_mode): Likewise.
488 (default_libgcc_shift_count_mode): Likewise.
489 (default_unwind_word_mode): Likewise.
490 (default_addr_space_pointer_mode): Likewise.
491 (default_addr_space_address_mode): Likewise.
492 (default_mode_rep_extended): Take a scalar_int_mode rather than
494 (default_valid_pointer_mode): Likewise.
495 (default_addr_space_valid_pointer_mode): Likewise.
496 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
498 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
499 rather than a machine_mode.
500 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
502 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
503 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
506 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
507 Alan Hayward <alan.hayward@arm.com>
508 David Sherwood <david.sherwood@arm.com>
510 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
511 the mode argument to scalar_int_mode.
512 (do_jump_by_parts_zero_rtx): Likewise.
513 (do_jump_by_parts_equality_rtx): Likewise.
514 (do_jump_by_parts_greater): Take a mode argument.
515 (do_jump_by_parts_equality): Likewise.
516 (do_jump_1): Update calls accordingly.
518 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
519 Alan Hayward <alan.hayward@arm.com>
520 David Sherwood <david.sherwood@arm.com>
522 * is-a.h (safe_dyn_cast): New function.
523 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
524 (jump_table_for_label): Likewise.
525 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
526 instead of an rtx_insn *.
527 (shorten_branches): Use dyn_cast instead of LABEL_P and
528 JUMP_TABLE_DATA_P. Use jump_table_for_label and
529 rtx_jump_table_data::get_data_mode.
530 (final_scan_insn): Likewise.
532 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
533 Alan Hayward <alan.hayward@arm.com>
534 David Sherwood <david.sherwood@arm.com>
536 * combine.c (try_combine): Use is_a <scalar_int_mode> when
537 trying to combine a full-register integer set with a subreg
540 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
541 Alan Hayward <alan.hayward@arm.com>
542 David Sherwood <david.sherwood@arm.com>
544 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
545 that is always either address_mode or pointer_mode.
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 word_mode instead of innermode
552 when the two are known to be equal.
554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
555 Alan Hayward <alan.hayward@arm.com>
556 David Sherwood <david.sherwood@arm.com>
558 * simplify-rtx.c (simplify_const_unary_operation): Use
559 is_a <scalar_int_mode> instead of checking for a nonzero
560 precision. Forcibly convert op_mode to a scalar_int_mode
561 in that case. More clearly differentiate the operand and
562 result modes and use the former when deciding what the value
563 of a count-bits operation should be. Use is_int_mode instead
564 of checking for a MODE_INT. Remove redundant check for whether
565 this mode has a zero precision.
567 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
568 Alan Hayward <alan.hayward@arm.com>
569 David Sherwood <david.sherwood@arm.com>
571 * optabs.c (widen_leading): Change the type of the mode argument
572 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
573 (widen_bswap): Likewise.
574 (expand_parity): Likewise.
575 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
576 (expand_ffs): Likewise.
577 (epand_unop): Check for scalar integer modes before calling the
580 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
581 Alan Hayward <alan.hayward@arm.com>
582 David Sherwood <david.sherwood@arm.com>
584 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
586 (expand_expr_real_1): Update call accordingly.
588 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
589 Alan Hayward <alan.hayward@arm.com>
590 David Sherwood <david.sherwood@arm.com>
592 * expmed.c (store_bit_field_using_insv): Add op0_mode and
593 value_mode arguments. Use scalar_int_mode internally.
594 (store_bit_field_1): Rename the new integer mode from imode
595 to op0_mode and use it instead of GET_MODE (op0). Update calls
596 to store_split_bit_field, store_bit_field_using_insv and
597 store_fixed_bit_field.
598 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
599 Use scalar_int_mode internally. Use a bit count rather than a mode
600 when calculating the largest bit size for get_best_mode.
601 Update calls to store_split_bit_field and store_fixed_bit_field_1.
602 (store_fixed_bit_field_1): Add mode and value_mode arguments.
603 Remove assertion that OP0 has a scalar integer mode.
604 (store_split_bit_field): Add op0_mode and value_mode arguments.
605 Update calls to extract_fixed_bit_field.
606 (extract_bit_field_using_extv): Add an op0_mode argument.
607 Use scalar_int_mode internally.
608 (extract_bit_field_1): Rename the new integer mode from imode to
609 op0_mode and use it instead of GET_MODE (op0). Update calls to
610 extract_split_bit_field, extract_bit_field_using_extv and
611 extract_fixed_bit_field.
612 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
613 to extract_split_bit_field and extract_fixed_bit_field_1.
614 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
615 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
617 (extract_split_bit_field): Add an op0_mode argument. Update call
618 to extract_fixed_bit_field.
620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
621 Alan Hayward <alan.hayward@arm.com>
622 David Sherwood <david.sherwood@arm.com>
624 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
625 * explow.c (hard_function_value): Likewise.
626 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
627 convert_to_mode call outside the loop.
628 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
629 for the mode iterator. Require the mode specified by max_pieces
631 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
633 (copy_blkmode_to_reg): Likewise.
634 (set_storage_via_setmem): Likewise.
635 * optabs.c (prepare_cmp_insn): Likewise.
636 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
637 * stor-layout.c (finish_bitfield_representative): Likewise.
639 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
640 Alan Hayward <alan.hayward@arm.com>
641 David Sherwood <david.sherwood@arm.com>
643 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
644 * expr.c (convert_move): Use them.
645 (convert_modes): Likewise.
646 (store_expr_with_bounds): Likewise.
648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
649 Alan Hayward <alan.hayward@arm.com>
650 David Sherwood <david.sherwood@arm.com>
652 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
653 parameter for the mode of "x". Remove the "known_x", "known_mode"
654 and "known_ret" arguments. Change the type of the mode argument
656 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
657 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
658 (reg_num_sign_bit_copies_for_combine): Likewise.
659 * rtlanal.c (nonzero_bits1): Likewise.
660 (num_sign_bit_copies1): Likewise.
661 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
662 (reg_num_sign_bit_copies_general): Likewise.
663 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
664 (reg_nonzero_bits_general): Likewise.
666 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
667 Alan Hayward <alan.hayward@arm.com>
668 David Sherwood <david.sherwood@arm.com>
670 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
671 than in subroutines. Return 1 for non-integer modes.
672 (cached_num_sign_bit_copies): Change the type of the mode parameter
674 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
675 classes. Handle CONST_INT_P first and then check whether X also
676 has a scalar integer mode. Check the same thing for inner registers
677 of a SUBREG and for values that are being extended or truncated.
679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
680 Alan Hayward <alan.hayward@arm.com>
681 David Sherwood <david.sherwood@arm.com>
683 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
684 in subroutines. Return the mode mask for non-integer modes.
685 (cached_nonzero_bits): Change the type of the mode parameter
687 (nonzero_bits1): Likewise. Remove early exit for other mode
688 classes. Handle CONST_INT_P first and then check whether X
689 also has a scalar integer mode.
691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
692 Alan Hayward <alan.hayward@arm.com>
693 David Sherwood <david.sherwood@arm.com>
695 * expr.c (widest_int_mode_for_size): Make the comment match the code.
696 Return a scalar_int_mode and assert that the size is greater than
698 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
699 (op_by_pieces_d::op_by_pieces_d): Likewise.
700 (op_by_pieces_d::run): Likewise.
701 (can_store_by_pieces): Likewise.
703 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
704 Alan Hayward <alan.hayward@arm.com>
705 David Sherwood <david.sherwood@arm.com>
707 * combine.c (extract_left_shift): Add a mode argument and update
709 (make_compound_operation_int): Change the type of the mode parameter
710 to scalar_int_mode and update the call to extract_left_shift.
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 * combine.c (simplify_and_const_int): Change the type of the mode
717 parameter to scalar_int_mode.
718 (simplify_and_const_int_1): Likewise. Update recursive call.
720 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
721 Alan Hayward <alan.hayward@arm.com>
722 David Sherwood <david.sherwood@arm.com>
724 * combine.c (simplify_compare_const): Check that the mode is a
725 scalar_int_mode (rather than VOIDmode) before testing its
727 (simplify_comparison): Move COMPARISON_P handling out of the
728 loop and restrict the latter part of the loop to scalar_int_modes.
729 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
730 and when considering SUBREG_REGs. Use is_int_mode instead of
731 checking GET_MODE_CLASS against MODE_INT.
733 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
734 Alan Hayward <alan.hayward@arm.com>
735 David Sherwood <david.sherwood@arm.com>
737 * combine.c (try_widen_shift_mode): Move check for equal modes to...
738 (simplify_shift_const_1): ...here. Use scalar_int_mode for
739 shift_unit_mode and for modes involved in scalar shifts.
741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
742 Alan Hayward <alan.hayward@arm.com>
743 David Sherwood <david.sherwood@arm.com>
745 * combine.c (force_int_to_mode): New function, split out from...
746 (force_to_mode): ...here. Keep xmode up-to-date and use it
747 instead of GET_MODE (x).
749 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
750 Alan Hayward <alan.hayward@arm.com>
751 David Sherwood <david.sherwood@arm.com>
753 * optabs-query.h (extraction_insn::struct_mode): Change type to
754 opt_scalar_int_mode and update comment.
755 (extraction_insn::field_mode): Change type to scalar_int_mode.
756 (extraction_insn::pos_mode): Likewise.
757 * combine.c (make_extraction): Update accordingly.
758 * optabs-query.c (get_traditional_extraction_insn): Likewise.
759 (get_optab_extraction_insn): Likewise.
760 * recog.c (simplify_while_replacing): Likewise.
761 * expmed.c (narrow_bit_field_mem): Change the type of the mode
762 parameter to opt_scalar_int_mode.
764 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
765 Alan Hayward <alan.hayward@arm.com>
766 David Sherwood <david.sherwood@arm.com>
768 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
769 to a scalar_int_mode instead of a machine_mode.
770 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
771 (get_best_mode): Return a boolean and use a pointer argument to store
772 the selected mode. Replace the limit mode parameter with a bit limit.
773 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
774 for the values returned by bit_field_mode_iterator::next_mode.
775 (store_bit_field): Update call to get_best_mode.
776 (store_fixed_bit_field): Likewise.
777 (extract_fixed_bit_field): Likewise.
778 * expr.c (optimize_bitfield_assignment_op): Likewise.
779 * fold-const.c (optimize_bit_field_compare): Likewise.
780 (fold_truth_andor_1): Likewise.
781 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
782 Update for new type of m_mode.
783 (get_best_mode): As above.
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 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
790 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
791 (store_bit_field): Check is_a <scalar_int_mode> before calling
792 strict_volatile_bitfield_p.
793 (extract_bit_field): Likewise.
795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
796 Alan Hayward <alan.hayward@arm.com>
797 David Sherwood <david.sherwood@arm.com>
799 * target.def (cstore_mode): Return a scalar_int_mode.
800 * doc/tm.texi: Regenerate.
801 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
802 * targhooks.h (default_cstore_mode): Likewise.
803 * targhooks.c (default_cstore_mode): Likewise, using a forced
805 * expmed.c (emit_cstore): Expect the target of the cstore to be
808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
809 Alan Hayward <alan.hayward@arm.com>
810 David Sherwood <david.sherwood@arm.com>
812 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
814 (niter_desc): Likewise mode.
815 (iv_analyze): Add a mode parameter.
817 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
818 and change its type to scalar_int_mode.
819 * loop-iv.c: Update commentary at head of file.
820 (iv_constant): Pass the mode paraeter before the rtx it describes
821 and change its type to scalar_int_mode. Remove VOIDmode handling.
822 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
823 (iv_extend): Likewise.
824 (shorten_into_mode): Likewise.
825 (iv_add): Use scalar_int_mode.
827 (iv_shift): Likewise.
828 (canonicalize_iv_subregs): Likewise.
829 (get_biv_step_1): Pass the outer_mode parameter before the rtx
830 it describes and change its mode to scalar_int_mode. Also change
831 the type of the returned inner_mode to scalar_int_mode.
832 (get_biv_step): Likewise, turning outer_mode from a pointer
833 into a direct parameter. Update call to get_biv_step_1.
834 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
835 iv_constant and get_biv_step.
836 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
837 and change its type to scalar_int_mode. Don't initialise iv->mode
838 to VOIDmode and remove later checks for its still being VOIDmode.
839 Update calls to iv_analyze_op and iv_analyze_expr. Check
840 is_a <scalar_int_mode> when changing the mode under consideration.
841 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
842 Update call to iv_analyze_expr.
843 (iv_analyze_op): Add a mode parameter. Reject subregs whose
844 inner register is not also a scalar_int_mode. Update call to
846 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
847 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
848 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
849 separate mode class checks. Update calls to iv_analyze. Remove
850 fix-up of VOIDmodes after iv_analyze_biv.
851 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
852 don't have a scalar_int_mode. Update call to biv_p.
854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
855 Alan Hayward <alan.hayward@arm.com>
856 David Sherwood <david.sherwood@arm.com>
858 * cfgexpand.c (convert_debug_memory_address): Use
859 as_a <scalar_int_mode>.
860 * combine.c (expand_compound_operation): Likewise.
861 (make_extraction): Likewise.
862 (change_zero_ext): Likewise.
863 (simplify_comparison): Likewise.
864 * cse.c (cse_insn): Likewise.
865 * dwarf2out.c (minmax_loc_descriptor): Likewise.
866 (mem_loc_descriptor): Likewise.
867 (loc_descriptor): Likewise.
868 * expmed.c (init_expmed_one_mode): Likewise.
869 (synth_mult): Likewise.
870 (emit_store_flag_1): Likewise.
871 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
872 of a comparison with size.
873 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
874 (reduce_to_bit_field_precision): Likewise.
875 * function.c (expand_function_end): Likewise.
876 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
877 * loop-doloop.c (doloop_modify): Likewise.
878 * optabs.c (expand_binop): Likewise.
879 (expand_unop): Likewise.
880 (expand_copysign_absneg): Likewise.
881 (prepare_cmp_insn): Likewise.
882 (maybe_legitimize_operand): Likewise.
883 * recog.c (const_scalar_int_operand): Likewise.
884 * rtlanal.c (get_address_mode): Likewise.
885 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
886 (simplify_cond_clz_ctz): Likewise.
887 * tree-nested.c (get_nl_goto_field): Likewise.
888 * tree.c (build_vector_type_for_mode): Likewise.
889 * var-tracking.c (use_narrower_mode): Likewise.
891 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
892 Alan Hayward <alan.hayward@arm.com>
893 David Sherwood <david.sherwood@arm.com>
895 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
896 * builtins.c (expand_builtin_signbit): Use it.
897 * cfgexpand.c (expand_debug_expr): Likewise.
898 * dojump.c (do_jump): Likewise.
899 (do_compare_and_jump): Likewise.
900 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
901 * expmed.c (make_tree): Likewise.
902 * expr.c (expand_expr_real_2): Likewise.
903 (expand_expr_real_1): Likewise.
904 (try_casesi): Likewise.
905 * fold-const-call.c (fold_const_call_ss): Likewise.
906 * fold-const.c (unextend): Likewise.
907 (extract_muldiv_1): Likewise.
908 (fold_single_bit_test): Likewise.
909 (native_encode_int): Likewise.
910 (native_encode_string): Likewise.
911 (native_interpret_int): Likewise.
912 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
913 * internal-fn.c (expand_addsub_overflow): Likewise.
914 (expand_neg_overflow): Likewise.
915 (expand_mul_overflow): Likewise.
916 (expand_arith_overflow): Likewise.
917 * match.pd: Likewise.
918 * stor-layout.c (layout_type): Likewise.
919 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
920 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
921 * tree-ssanames.c (get_range_info): Likewise.
922 * tree-switch-conversion.c (array_value_type) Likewise.
923 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
924 (vect_recog_divmod_pattern): Likewise.
925 (vect_recog_mixed_size_cond_pattern): Likewise.
926 * tree-vrp.c (extract_range_basic): Likewise.
927 (simplify_float_conversion_using_ranges): Likewise.
928 * tree.c (int_fits_type_p): Likewise.
929 * ubsan.c (instrument_bool_enum_load): Likewise.
930 * varasm.c (mergeable_string_section): Likewise.
931 (narrowing_initializer_constant_valid_p): Likewise.
932 (output_constant): Likewise.
934 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
935 Alan Hayward <alan.hayward@arm.com>
936 David Sherwood <david.sherwood@arm.com>
938 * machmode.h (NARROWEST_INT_MODE): New macro.
939 * expr.c (alignment_for_piecewise_move): Use it instead of
940 GET_CLASS_NARROWEST_MODE (MODE_INT).
941 (push_block): Likewise.
942 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
944 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
946 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
947 Alan Hayward <alan.hayward@arm.com>
948 David Sherwood <david.sherwood@arm.com>
950 * postreload.c (move2add_valid_value_p): Change the type of the
951 mode parameter to scalar_int_mode.
952 (move2add_use_add2_insn): Add a mode parameter and use it instead
954 (move2add_use_add3_insn): Likewise.
955 (reload_cse_move2add): Update accordingly.
957 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
958 Alan Hayward <alan.hayward@arm.com>
959 David Sherwood <david.sherwood@arm.com>
961 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
963 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
964 * optabs.c (expand_unop): Likewise.
966 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
967 Alan Hayward <alan.hayward@arm.com>
968 David Sherwood <david.sherwood@arm.com>
970 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
971 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
972 (popcount_loc_descriptor): Likewise.
973 (bswap_loc_descriptor): Likewise.
974 (rotate_loc_descriptor): Likewise.
975 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
976 calling the functions above.
978 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
979 Alan Hayward <alan.hayward@arm.com>
980 David Sherwood <david.sherwood@arm.com>
982 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
984 (try_combine): Likewise.
985 (simplify_if_then_else): Likewise.
986 * cse.c (cse_insn): Likewise.
987 * dwarf2out.c (mem_loc_descriptor): Likewise.
988 * emit-rtl.c (gen_lowpart_common): Likewise.
989 * simplify-rtx.c (simplify_truncation): Likewise.
990 (simplify_binary_operation_1): Likewise.
991 (simplify_const_relational_operation): Likewise.
992 (simplify_ternary_operation): Likewise.
993 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
995 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
996 Alan Hayward <alan.hayward@arm.com>
997 David Sherwood <david.sherwood@arm.com>
999 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
1000 * reload.c (push_reload): Likewise.
1001 (find_reloads): Likewise.
1003 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1004 Alan Hayward <alan.hayward@arm.com>
1005 David Sherwood <david.sherwood@arm.com>
1007 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
1008 (make_compound_operation_int): Likewise.
1009 (change_zero_ext): Likewise.
1010 * expr.c (convert_move): Likewise.
1011 (convert_modes): Likewise.
1012 * fwprop.c (forward_propagate_subreg): Likewise.
1013 * loop-iv.c (get_biv_step_1): Likewise.
1014 * optabs.c (widen_operand): Likewise.
1015 * postreload.c (move2add_valid_value_p): Likewise.
1016 * recog.c (simplify_while_replacing): Likewise.
1017 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1018 (simplify_binary_operation_1): Likewise. Remove redundant
1019 mode equality check.
1021 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1022 Alan Hayward <alan.hayward@arm.com>
1023 David Sherwood <david.sherwood@arm.com>
1025 * combine.c (combine_simplify_rtx): Add checks for
1026 is_a <scalar_int_mode>.
1027 (simplify_if_then_else): Likewise.
1028 (make_field_assignment): Likewise.
1029 (simplify_comparison): Likewise.
1030 * ifcvt.c (noce_try_bitop): Likewise.
1031 * loop-invariant.c (canonicalize_address_mult): Likewise.
1032 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1034 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1035 Alan Hayward <alan.hayward@arm.com>
1036 David Sherwood <david.sherwood@arm.com>
1038 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
1039 is_a <scalar_int_mode> instead of != BLKmode.
1041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1042 Alan Hayward <alan.hayward@arm.com>
1043 David Sherwood <david.sherwood@arm.com>
1045 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
1046 instead of != VOIDmode.
1047 * combine.c (if_then_else_cond): Likewise.
1048 (change_zero_ext): Likewise.
1049 * dwarf2out.c (mem_loc_descriptor): Likewise.
1050 (loc_descriptor): Likewise.
1051 * rtlanal.c (canonicalize_condition): Likewise.
1052 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
1054 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1055 Alan Hayward <alan.hayward@arm.com>
1056 David Sherwood <david.sherwood@arm.com>
1058 * simplify-rtx.c (simplify_binary_operation_1): Use
1059 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
1061 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1062 Alan Hayward <alan.hayward@arm.com>
1063 David Sherwood <david.sherwood@arm.com>
1065 * wide-int.h (int_traits<unsigned char>) New class.
1066 (int_traits<unsigned short>) Likewise.
1067 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
1068 Use GET_MODE_UNIT_PRECISION and remove redundant test for
1070 * combine.c (set_nonzero_bits_and_sign_copies): Use
1071 is_a <scalar_int_mode>.
1072 (find_split_point): Likewise.
1073 (combine_simplify_rtx): Likewise.
1074 (simplify_logical): Likewise.
1075 (expand_compound_operation): Likewise.
1076 (expand_field_assignment): Likewise.
1077 (make_compound_operation): Likewise.
1078 (extended_count): Likewise.
1079 (change_zero_ext): Likewise.
1080 (simplify_comparison): Likewise.
1081 * dwarf2out.c (scompare_loc_descriptor): Likewise.
1082 (ucompare_loc_descriptor): Likewise.
1083 (minmax_loc_descriptor): Likewise.
1084 (mem_loc_descriptor): Likewise.
1085 (loc_descriptor): Likewise.
1086 * expmed.c (init_expmed_one_mode): Likewise.
1087 * lra-constraints.c (lra_constraint_offset): Likewise.
1088 * optabs.c (prepare_libcall_arg): Likewise.
1089 * postreload.c (move2add_note_store): Likewise.
1090 * reload.c (operands_match_p): Likewise.
1091 * rtl.h (load_extend_op): Likewise.
1092 * rtlhooks.c (gen_lowpart_general): Likewise.
1093 * simplify-rtx.c (simplify_truncation): Likewise.
1094 (simplify_unary_operation_1): Likewise.
1095 (simplify_binary_operation_1): Likewise.
1096 (simplify_const_binary_operation): Likewise.
1097 (simplify_const_relational_operation): Likewise.
1098 (simplify_subreg): Likewise.
1099 * stor-layout.c (bitwise_mode_for_mode): Likewise.
1100 * var-tracking.c (adjust_mems): Likewise.
1101 (prepare_call_arguments): Likewise.
1103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1104 Alan Hayward <alan.hayward@arm.com>
1105 David Sherwood <david.sherwood@arm.com>
1107 * machmode.h (is_int_mode): New fuction.
1108 * combine.c (find_split_point): Use it.
1109 (combine_simplify_rtx): Likewise.
1110 (simplify_if_then_else): Likewise.
1111 (simplify_set): Likewise.
1112 (simplify_shift_const_1): Likewise.
1113 (simplify_comparison): Likewise.
1114 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
1115 * cse.c (notreg_cost): Likewise.
1116 (cse_insn): Likewise.
1117 * cselib.c (cselib_lookup_1): Likewise.
1118 * dojump.c (do_jump_1): Likewise.
1119 (do_compare_rtx_and_jump): Likewise.
1120 * dse.c (get_call_args): Likewise.
1121 * dwarf2out.c (rtl_for_decl_init): Likewise.
1122 (native_encode_initializer): Likewise.
1123 * expmed.c (emit_store_flag_1): Likewise.
1124 (emit_store_flag): Likewise.
1125 * expr.c (convert_modes): Likewise.
1126 (store_field): Likewise.
1127 (expand_expr_real_1): Likewise.
1128 * fold-const.c (fold_read_from_constant_string): Likewise.
1129 * gimple-ssa-sprintf.c (get_format_string): Likewise.
1130 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
1131 * optabs.c (expand_binop): Likewise.
1132 (expand_unop): Likewise.
1133 (expand_abs_nojump): Likewise.
1134 (expand_one_cmpl_abs_nojump): Likewise.
1135 * simplify-rtx.c (mode_signbit_p): Likewise.
1136 (val_signbit_p): Likewise.
1137 (val_signbit_known_set_p): Likewise.
1138 (val_signbit_known_clear_p): Likewise.
1139 (simplify_relational_operation_1): Likewise.
1140 * tree.c (vector_type_mode): Likewise.
1142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1143 Alan Hayward <alan.hayward@arm.com>
1144 David Sherwood <david.sherwood@arm.com>
1146 * machmode.h (smallest_mode_for_size): Fix formatting.
1147 (smallest_int_mode_for_size): New function.
1148 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
1149 instead of smallest_mode_for_size.
1150 * combine.c (make_extraction): Likewise.
1151 * config/arc/arc.c (arc_expand_movmem): Likewise.
1152 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
1153 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
1154 * config/s390/s390.c (s390_expand_insv): Likewise.
1155 * config/sparc/sparc.c (assign_int_registers): Likewise.
1156 * config/spu/spu.c (spu_function_value): Likewise.
1157 (spu_function_arg): Likewise.
1158 * coverage.c (get_gcov_type): Likewise.
1159 (get_gcov_unsigned_t): Likewise.
1160 * dse.c (find_shift_sequence): Likewise.
1161 * expmed.c (store_bit_field_1): Likewise.
1162 * expr.c (convert_move): Likewise.
1163 (store_field): Likewise.
1164 * internal-fn.c (expand_arith_overflow): Likewise.
1165 * optabs-query.c (get_best_extraction_insn): Likewise.
1166 * optabs.c (expand_twoval_binop_libfunc): Likewise.
1167 * stor-layout.c (layout_type): Likewise.
1168 (initialize_sizetypes): Likewise.
1169 * targhooks.c (default_get_mask_mode): Likewise.
1170 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
1172 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1173 Alan Hayward <alan.hayward@arm.com>
1174 David Sherwood <david.sherwood@arm.com>
1176 * machmode.h (opt_mode::else_blk): New function.
1177 (int_mode_for_mode): Declare.
1178 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
1179 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
1181 * cfgexpand.c (expand_debug_expr): Likewise.
1182 * combine.c (gen_lowpart_or_truncate): Likewise.
1183 (gen_lowpart_for_combine): Likewise.
1184 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
1185 * config/avr/avr.c (avr_to_int_mode): Likewise.
1186 (avr_out_plus_1): Likewise.
1187 (avr_out_plus): Likewise.
1188 (avr_out_round): Likewise.
1189 * config/i386/i386.c (ix86_split_to_parts): Likewise.
1190 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
1191 (s390_expand_vcond): Likewise.
1192 * config/spu/spu.c (spu_split_immediate): Likewise.
1193 (spu_expand_mov): Likewise.
1194 * dse.c (get_stored_val): Likewise.
1195 * expmed.c (store_bit_field_1): Likewise.
1196 (convert_extracted_bit_field): Use int_mode_for_mode instead of
1198 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
1199 (extract_low_bits): Likewise.
1200 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
1201 handling rather than repeating the check.
1202 (emit_group_store): Likewise.
1203 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
1204 * optabs.c (expand_absneg_bit): Likewise.
1205 (expand_copysign_absneg): Likewise.
1206 (expand_copysign_bit): Likewise.
1207 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
1208 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
1209 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
1210 * var-tracking.c (prepare_call_arguments): Likewise.
1211 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
1212 int_mode_for_mode instead of mode_for_size.
1213 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
1215 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1216 Alan Hayward <alan.hayward@arm.com>
1217 David Sherwood <david.sherwood@arm.com>
1219 * machmode.h (int_mode_for_size): New function.
1220 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
1221 instead of mode_for_size.
1222 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
1224 * combine.c (expand_field_assignment): Use int_mode_for_size
1225 instead of mode_for_size.
1226 (make_extraction): Likewise.
1227 (simplify_shift_const_1): Likewise.
1228 (simplify_comparison): Likewise.
1229 * dojump.c (do_jump): Likewise.
1230 * dwarf2out.c (mem_loc_descriptor): Likewise.
1231 * emit-rtl.c (init_derived_machine_modes): Likewise.
1232 * expmed.c (flip_storage_order): Likewise.
1233 (convert_extracted_bit_field): Likewise.
1234 * expr.c (copy_blkmode_from_reg): Likewise.
1235 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
1236 * internal-fn.c (expand_mul_overflow): Likewise.
1237 * lower-subreg.c (simple_move): Likewise.
1238 * optabs-libfuncs.c (init_optabs): Likewise.
1239 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1240 * tree.c (vector_type_mode): Likewise.
1241 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
1242 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
1243 * tree-vect-generic.c (expand_vector_parallel): Likewise.
1244 * tree-vect-stmts.c (vectorizable_load): Likewise.
1245 (vectorizable_store): Likewise.
1247 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1248 Alan Hayward <alan.hayward@arm.com>
1249 David Sherwood <david.sherwood@arm.com>
1251 * coretypes.h (pod_mode): New type.
1252 (scalar_int_mode_pod): New typedef.
1253 * machmode.h (pod_mode): New class.
1254 (int_n_data_t::m): Change type to scalar_int_mode_pod.
1255 * genmodes.c (emit_mode_int_n): Update accordingly.
1256 * lower-subreg.h (target_lower_subreg): Change type to
1257 scalar_int_mode_pod.
1258 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
1259 scalar_int_mode_pod.
1261 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1262 Alan Hayward <alan.hayward@arm.com>
1263 David Sherwood <david.sherwood@arm.com>
1265 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
1266 machine_mode to scalar_int_mode.
1267 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
1268 (rs6000_option_override_internal): Remove cast to int.
1269 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
1270 machine_mode to scalar_int_mode.
1271 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
1272 (rs6000_option_override_internal): Remove cast to int.
1273 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
1274 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
1276 * config/s390/s390.c (s390_expand_builtin): Likewise.
1277 * coretypes.h (scalar_int_mode): New type.
1278 (opt_scalar_int_mode): New typedef.
1279 * machmode.h (scalar_int_mode): New class.
1280 (scalar_int_mode::includes_p): New function.
1281 (byte_mode): Change type to scalar_int_mode.
1282 (word_mode): Likewise.
1283 (ptr_mode): Likewise.
1284 * emit-rtl.c (byte_mode): Likewise.
1285 (word_mode): Likewise.
1286 (ptr_mode): Likewise.
1287 (init_derived_machine_modes): Update accordingly.
1288 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
1289 and MODE_PARTIAL_INT.
1290 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
1291 opt_scalar_int_mode.
1293 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1294 Alan Hayward <alan.hayward@arm.com>
1295 David Sherwood <david.sherwood@arm.com>
1297 * target.def (libgcc_floating_mode_supported_p): Take a
1299 * doc/tm.texi: Regenerate.
1300 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
1302 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
1303 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
1306 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1307 Alan Hayward <alan.hayward@arm.com>
1308 David Sherwood <david.sherwood@arm.com>
1310 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
1311 * doc/tm.texi: Regenerate.
1312 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
1313 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
1314 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
1315 * targhooks.h (default_floatn_mode): Likewise.
1316 * targhooks.c (default_floatn_mode): Likewise.
1317 * tree.c (build_common_tree_nodes): Update accordingly.
1319 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1320 Alan Hayward <alan.hayward@arm.com>
1321 David Sherwood <david.sherwood@arm.com>
1323 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
1324 (mode_iterator::iterate_p): Likewise.
1325 (mode_iterator::get_wider): Likewise.
1326 * expr.c (init_expr_target): Use opt_scalar_float_mode.
1328 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1329 Alan Hayward <alan.hayward@arm.com>
1330 David Sherwood <david.sherwood@arm.com>
1332 * coretypes.h (opt_scalar_float_mode): New typedef.
1333 * machmode.h (float_mode_for_size): New function.
1334 * emit-rtl.c (double_mode): Delete.
1335 (init_emit_once): Use float_mode_for_size.
1336 * stor-layout.c (layout_type): Likewise.
1337 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
1339 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1340 Alan Hayward <alan.hayward@arm.com>
1341 David Sherwood <david.sherwood@arm.com>
1343 * output.h (assemble_real): Take a scalar_float_mode.
1344 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
1345 * config/arm/arm.md (consttable_4): Likewise.
1346 (consttable_8): Likewise.
1347 (consttable_16): Likewise.
1348 * config/mips/mips.md (consttable_float): Likewise.
1349 * config/s390/s390.c (s390_output_pool_entry): Likewise.
1350 * varasm.c (assemble_real): Take a scalar_float_mode.
1351 (output_constant_pool_2): Update accordingly.
1352 (output_constant): Likewise.
1354 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1355 Alan Hayward <alan.hayward@arm.com>
1356 David Sherwood <david.sherwood@arm.com>
1358 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
1359 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
1360 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
1361 (native_encode_real): Likewise.
1362 (native_interpret_real): Likewise.
1363 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
1364 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1367 Alan Hayward <alan.hayward@arm.com>
1368 David Sherwood <david.sherwood@arm.com>
1370 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
1371 <scalar_float_mode>. Simplify.
1372 (gen_extend_conv_libfunc): Likewise.
1374 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1375 Alan Hayward <alan.hayward@arm.com>
1376 David Sherwood <david.sherwood@arm.com>
1378 * coretypes.h (scalar_float_mode): New type.
1379 * machmode.h (mode_traits::from_int): Use machine_mode if
1380 USE_ENUM_MODES is defined.
1381 (is_a): New function.
1383 (dyn_cast): Likewise.
1384 (scalar_float_mode): New class.
1385 (scalar_float_mode::includes_p): New function.
1386 (is_float_mode): Likewise.
1387 * gdbhooks.py (MachineModePrinter): New class.
1388 (build_pretty_printer): Use it for scalar_float_mode.
1389 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
1390 (format_helper::format_helper): Turn into a template.
1391 * genmodes.c (get_mode_class): New function.
1392 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
1393 or machine_mode if none.
1394 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
1395 as_a <scalar_float_mode>.
1396 * dwarf2out.c (mem_loc_descriptor): Likewise.
1397 (insert_float): Likewise.
1398 (add_const_value_attribute): Likewise.
1399 * simplify-rtx.c (simplify_immed_subreg): Likewise.
1400 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
1401 (expand_unop): Update accordingly.
1402 (expand_abs_nojump): Likewise.
1403 (expand_copysign_absneg): Take a scalar_float_mode.
1404 (expand_copysign_bit): Likewise.
1405 (expand_copysign): Update accordingly.
1407 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1408 Alan Hayward <alan.hayward@arm.com>
1409 David Sherwood <david.sherwood@arm.com>
1411 * coretypes.h (opt_mode): New class.
1412 * machmode.h (opt_mode): Likewise.
1413 (opt_mode::else_void): New function.
1414 (opt_mode::require): Likewise.
1415 (opt_mode::exists): Likewise.
1416 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
1417 (GET_MODE_2XWIDER_MODE): Likewise.
1418 (mode_iterator::get_wider): Update accordingly.
1419 (mode_iterator::get_2xwider): Likewise.
1420 (mode_iterator::get_known_wider): Likewise, turning into a template.
1421 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
1422 forcing a wider mode to exist.
1423 * config/cr16/cr16.h (LONG_REG_P): Likewise.
1424 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1425 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
1426 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
1427 * lower-subreg.c (init_lower_subreg): Likewise.
1428 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
1429 on the final iteration.
1430 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
1431 a wider mode exists before asking for a move pattern.
1432 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
1433 forcing a wider mode to exist.
1434 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
1435 returning false if no such mode exists.
1436 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
1437 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
1438 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
1439 Avoid checking for a MODE_INT if we already know the mode is not a
1441 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
1442 forcing a wider mode to exist.
1443 (expmed_mult_highpart_optab): Likewise.
1444 (expmed_mult_highpart): Likewise.
1445 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
1447 * lto-streamer-in.c (lto_input_mode_table): Likewise.
1448 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
1449 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
1450 * internal-fn.c (expand_mul_overflow): Update use of
1451 GET_MODE_2XWIDER_MODE.
1452 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1453 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
1454 GET_MODE_WIDER_MODE.
1455 (convert_plusminus_to_widen): Likewise.
1456 * tree-switch-conversion.c (array_value_type): Likewise.
1457 * var-tracking.c (emit_note_insn_var_location): Likewise.
1458 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
1459 Return false inside rather than outside the loop if no wider mode
1461 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
1462 and GET_MODE_2XWIDER_MODE
1463 (can_compare_p): Use else_void.
1464 * gdbhooks.py (OptMachineModePrinter): New class.
1465 (build_pretty_printer): Use it for opt_mode.
1467 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1468 Alan Hayward <alan.hayward@arm.com>
1469 David Sherwood <david.sherwood@arm.com>
1471 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
1472 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
1474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1475 Alan Hayward <alan.hayward@arm.com>
1476 David Sherwood <david.sherwood@arm.com>
1478 * machmode.h (mode_traits): New structure.
1479 (get_narrowest_mode): New function.
1480 (mode_iterator::start): Likewise.
1481 (mode_iterator::iterate_p): Likewise.
1482 (mode_iterator::get_wider): Likewise.
1483 (mode_iterator::get_known_wider): Likewise.
1484 (mode_iterator::get_2xwider): Likewise.
1485 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
1486 (FOR_EACH_MODE): Likewise.
1487 (FOR_EACH_MODE_FROM): Likewise.
1488 (FOR_EACH_MODE_UNTIL): Likewise.
1489 (FOR_EACH_WIDER_MODE): Likewise.
1490 (FOR_EACH_2XWIDER_MODE): Likewise.
1491 * builtins.c (expand_builtin_strlen): Use new mode iterators.
1492 * combine.c (simplify_comparison): Likewise
1493 * config/i386/i386.c (type_natural_mode): Likewise.
1494 * cse.c (cse_insn): Likewise.
1495 * dse.c (find_shift_sequence): Likewise.
1496 * emit-rtl.c (init_derived_machine_modes): Likewise.
1497 (init_emit_once): Likewise.
1498 * explow.c (hard_function_value): Likewise.
1499 * expmed.c (extract_fixed_bit_field_1): Likewise.
1500 (extract_bit_field_1): Likewise.
1501 (expand_divmod): Likewise.
1502 (emit_store_flag_1): Likewise.
1503 * expr.c (init_expr_target): Likewise.
1504 (convert_move): Likewise.
1505 (alignment_for_piecewise_move): Likewise.
1506 (widest_int_mode_for_size): Likewise.
1507 (emit_block_move_via_movmem): Likewise.
1508 (copy_blkmode_to_reg): Likewise.
1509 (set_storage_via_setmem): Likewise.
1510 (compress_float_constant): Likewise.
1511 * omp-low.c (omp_clause_aligned_alignment): Likewise.
1512 * optabs-query.c (get_best_extraction_insn): Likewise.
1513 * optabs.c (expand_binop): Likewise.
1514 (expand_twoval_unop): Likewise.
1515 (expand_twoval_binop): Likewise.
1516 (widen_leading): Likewise.
1517 (widen_bswap): Likewise.
1518 (expand_parity): Likewise.
1519 (expand_unop): Likewise.
1520 (prepare_cmp_insn): Likewise.
1521 (prepare_float_lib_cmp): Likewise.
1522 (expand_float): Likewise.
1523 (expand_fix): Likewise.
1524 (expand_sfix_optab): Likewise.
1525 * postreload.c (move2add_use_add2_insn): Likewise.
1526 * reg-stack.c (reg_to_stack): Likewise.
1527 * reginfo.c (choose_hard_reg_mode): Likewise.
1528 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
1529 * stor-layout.c (mode_for_size): Likewise.
1530 (smallest_mode_for_size): Likewise.
1531 (mode_for_vector): Likewise.
1532 (finish_bitfield_representative): Likewise.
1533 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
1534 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
1535 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
1536 * var-tracking.c (prepare_call_arguments): Likewise.
1538 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1539 Alan Hayward <alan.hayward@arm.com>
1540 David Sherwood <david.sherwood@arm.com>
1542 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
1543 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
1544 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
1545 * machmode.h (mode_size): Move earlier in file.
1546 (mode_precision): Likewise.
1547 (mode_inner): Likewise.
1548 (mode_nunits): Likewise.
1549 (mode_unit_size): Likewise.
1550 (unit_unit_precision): Likewise.
1551 (mode_wider): Likewise.
1552 (mode_2xwider): Likewise.
1553 (machine_mode): New class.
1554 (mode_to_bytes): New function.
1555 (mode_to_bits): Likewise.
1556 (mode_to_precision): Likewise.
1557 (mode_to_inner): Likewise.
1558 (mode_to_unit_size): Likewise.
1559 (mode_to_unit_precision): Likewise.
1560 (mode_to_nunits): Likewise.
1561 (GET_MODE_SIZE): Use mode_to_bytes.
1562 (GET_MODE_BITSIZE): Use mode_to_bits.
1563 (GET_MODE_PRECISION): Use mode_to_precision.
1564 (GET_MODE_INNER): Use mode_to_inner.
1565 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
1566 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
1567 (GET_MODE_NUNITS): Use mode_to_nunits.
1568 * system.h (ALWAYS_INLINE): New macro.
1569 * config/powerpcspe/powerpcspe-c.c
1570 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
1571 int for arg1_mode and arg2_mode.
1573 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1574 Alan Hayward <alan.hayward@arm.com>
1575 David Sherwood <david.sherwood@arm.com>
1577 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
1578 Prefix mode names with E_ in case statements.
1579 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1580 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
1581 (aarch64_split_simd_move): Likewise.
1582 (aarch64_gen_storewb_pair): Likewise.
1583 (aarch64_gen_loadwb_pair): Likewise.
1584 (aarch64_gen_store_pair): Likewise.
1585 (aarch64_gen_load_pair): Likewise.
1586 (aarch64_get_condition_code_1): Likewise.
1587 (aarch64_constant_pool_reload_icode): Likewise.
1588 (get_rsqrte_type): Likewise.
1589 (get_rsqrts_type): Likewise.
1590 (get_recpe_type): Likewise.
1591 (get_recps_type): Likewise.
1592 (aarch64_gimplify_va_arg_expr): Likewise.
1593 (aarch64_simd_container_mode): Likewise.
1594 (aarch64_emit_load_exclusive): Likewise.
1595 (aarch64_emit_store_exclusive): Likewise.
1596 (aarch64_expand_compare_and_swap): Likewise.
1597 (aarch64_gen_atomic_cas): Likewise.
1598 (aarch64_emit_bic): Likewise.
1599 (aarch64_emit_atomic_swap): Likewise.
1600 (aarch64_emit_atomic_load_op): Likewise.
1601 (aarch64_evpc_trn): Likewise.
1602 (aarch64_evpc_uzp): Likewise.
1603 (aarch64_evpc_zip): Likewise.
1604 (aarch64_evpc_ext): Likewise.
1605 (aarch64_evpc_rev): Likewise.
1606 (aarch64_evpc_dup): Likewise.
1607 (aarch64_gen_ccmp_first): Likewise.
1608 (aarch64_gen_ccmp_next): Likewise.
1609 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
1610 (alpha_emit_xfloating_libcall): Likewise.
1611 (emit_insxl): Likewise.
1612 (alpha_arg_type): Likewise.
1613 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
1614 (arc_preferred_simd_mode): Likewise.
1615 (arc_secondary_reload): Likewise.
1616 (get_arc_condition_code): Likewise.
1617 (arc_print_operand): Likewise.
1618 (arc_legitimate_constant_p): Likewise.
1619 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1620 * config/arc/arc.md (casesi_load): Likewise.
1621 (casesi_compact_jump): Likewise.
1622 * config/arc/predicates.md (proper_comparison_operator): Likewise.
1623 (cc_use_register): Likewise.
1624 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1625 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
1626 (arm_init_iwmmxt_builtins): Likewise.
1627 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
1628 (neon_expand_vector_init): Likewise.
1629 (arm_attr_length_move_neon): Likewise.
1630 (maybe_get_arm_condition_code): Likewise.
1631 (arm_emit_vector_const): Likewise.
1632 (arm_preferred_simd_mode): Likewise.
1633 (arm_output_iwmmxt_tinsr): Likewise.
1634 (thumb1_output_casesi): Likewise.
1635 (thumb2_output_casesi): Likewise.
1636 (arm_emit_load_exclusive): Likewise.
1637 (arm_emit_store_exclusive): Likewise.
1638 (arm_expand_compare_and_swap): Likewise.
1639 (arm_evpc_neon_vuzp): Likewise.
1640 (arm_evpc_neon_vzip): Likewise.
1641 (arm_evpc_neon_vrev): Likewise.
1642 (arm_evpc_neon_vtrn): Likewise.
1643 (arm_evpc_neon_vext): Likewise.
1644 (arm_validize_comparison): Likewise.
1645 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
1646 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
1647 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
1648 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
1649 (c6x_preferred_simd_mode): Likewise.
1650 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
1651 (epiphany_rtx_costs): Likewise.
1652 * config/epiphany/predicates.md (proper_comparison_operator):
1654 * config/frv/frv.c (condexec_memory_operand): Likewise.
1655 (frv_emit_move): Likewise.
1656 (output_move_single): Likewise.
1657 (output_condmove_single): Likewise.
1658 (frv_hard_regno_mode_ok): Likewise.
1659 (frv_matching_accg_mode): Likewise.
1660 * config/h8300/h8300.c (split_adds_subs): Likewise.
1661 (h8300_rtx_costs): Likewise.
1662 (h8300_print_operand): Likewise.
1663 (compute_mov_length): Likewise.
1664 (output_logical_op): Likewise.
1665 (compute_logical_op_length): Likewise.
1666 (compute_logical_op_cc): Likewise.
1667 (h8300_shift_needs_scratch_p): Likewise.
1668 (output_a_shift): Likewise.
1669 (compute_a_shift_length): Likewise.
1670 (compute_a_shift_cc): Likewise.
1671 (expand_a_rotate): Likewise.
1672 (output_a_rotate): Likewise.
1673 * config/i386/i386.c (classify_argument): Likewise.
1674 (function_arg_advance_32): Likewise.
1675 (function_arg_32): Likewise.
1676 (function_arg_64): Likewise.
1677 (function_value_64): Likewise.
1678 (ix86_gimplify_va_arg): Likewise.
1679 (ix86_legitimate_constant_p): Likewise.
1680 (put_condition_code): Likewise.
1681 (split_double_mode): Likewise.
1682 (ix86_avx256_split_vector_move_misalign): Likewise.
1683 (ix86_expand_vector_logical_operator): Likewise.
1684 (ix86_split_idivmod): Likewise.
1685 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
1686 (ix86_build_const_vector): Likewise.
1687 (ix86_build_signbit_mask): Likewise.
1688 (ix86_match_ccmode): Likewise.
1689 (ix86_cc_modes_compatible): Likewise.
1690 (ix86_expand_branch): Likewise.
1691 (ix86_expand_sse_cmp): Likewise.
1692 (ix86_expand_sse_movcc): Likewise.
1693 (ix86_expand_int_sse_cmp): Likewise.
1694 (ix86_expand_vec_perm_vpermi2): Likewise.
1695 (ix86_expand_vec_perm): Likewise.
1696 (ix86_expand_sse_unpack): Likewise.
1697 (ix86_expand_int_addcc): Likewise.
1698 (ix86_split_to_parts): Likewise.
1699 (ix86_vectorize_builtin_gather): Likewise.
1700 (ix86_vectorize_builtin_scatter): Likewise.
1701 (avx_vpermilp_parallel): Likewise.
1702 (inline_memory_move_cost): Likewise.
1703 (ix86_tieable_integer_mode_p): Likewise.
1704 (x86_maybe_negate_const_int): Likewise.
1705 (ix86_expand_vector_init_duplicate): Likewise.
1706 (ix86_expand_vector_init_one_nonzero): Likewise.
1707 (ix86_expand_vector_init_one_var): Likewise.
1708 (ix86_expand_vector_init_concat): Likewise.
1709 (ix86_expand_vector_init_interleave): Likewise.
1710 (ix86_expand_vector_init_general): Likewise.
1711 (ix86_expand_vector_set): Likewise.
1712 (ix86_expand_vector_extract): Likewise.
1713 (emit_reduc_half): Likewise.
1714 (ix86_emit_i387_round): Likewise.
1715 (ix86_mangle_type): Likewise.
1716 (ix86_expand_round_sse4): Likewise.
1717 (expand_vec_perm_blend): Likewise.
1718 (canonicalize_vector_int_perm): Likewise.
1719 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
1720 (expand_vec_perm_1): Likewise.
1721 (expand_vec_perm_interleave3): Likewise.
1722 (expand_vec_perm_even_odd_pack): Likewise.
1723 (expand_vec_perm_even_odd_1): Likewise.
1724 (expand_vec_perm_broadcast_1): Likewise.
1725 (ix86_vectorize_vec_perm_const_ok): Likewise.
1726 (ix86_expand_vecop_qihi): Likewise.
1727 (ix86_expand_mul_widen_hilo): Likewise.
1728 (ix86_expand_sse2_abs): Likewise.
1729 (ix86_expand_pextr): Likewise.
1730 (ix86_expand_pinsr): Likewise.
1731 (ix86_preferred_simd_mode): Likewise.
1732 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
1733 * config/i386/sse.md (*andnot<mode>3): Likewise.
1734 (<mask_codefor><code><mode>3<mask_name>): Likewise.
1735 (*<code><mode>3): Likewise.
1736 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
1737 (ia64_expand_atomic_op): Likewise.
1738 (ia64_arg_type): Likewise.
1739 (ia64_mode_to_int): Likewise.
1740 (ia64_scalar_mode_supported_p): Likewise.
1741 (ia64_vector_mode_supported_p): Likewise.
1742 (expand_vec_perm_broadcast): Likewise.
1743 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
1744 (iq2000_function_arg_advance): Likewise.
1745 (iq2000_function_arg): Likewise.
1746 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
1747 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
1748 (m68k_libcall_value): Likewise.
1749 (m68k_function_value): Likewise.
1750 (sched_attr_op_type): Likewise.
1751 * config/mcore/mcore.c (mcore_output_move): Likewise.
1752 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
1754 (microblaze_function_arg): Likewise.
1755 * config/mips/mips.c (mips16_build_call_stub): Likewise.
1756 (mips_print_operand): Likewise.
1757 (mips_mode_ok_for_mov_fmt_p): Likewise.
1758 (mips_vector_mode_supported_p): Likewise.
1759 (mips_preferred_simd_mode): Likewise.
1760 (mips_expand_vpc_loongson_even_odd): Likewise.
1761 (mips_expand_vec_unpack): Likewise.
1762 (mips_expand_vi_broadcast): Likewise.
1763 (mips_expand_vector_init): Likewise.
1764 (mips_expand_vec_reduc): Likewise.
1765 (mips_expand_msa_cmp): Likewise.
1766 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
1767 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
1768 (cc_flags_for_mode): Likewise.
1769 * config/msp430/msp430.c (msp430_print_operand): Likewise.
1770 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
1771 (nds32_output_casesi_pc_relative): Likewise.
1772 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1773 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
1774 (nvptx_gen_unpack): Likewise.
1775 (nvptx_gen_pack): Likewise.
1776 (nvptx_gen_shuffle): Likewise.
1777 (nvptx_gen_wcast): Likewise.
1778 (nvptx_preferred_simd_mode): Likewise.
1779 * config/pa/pa.c (pa_secondary_reload): Likewise.
1780 * config/pa/predicates.md (base14_operand): Likewise.
1781 * config/powerpcspe/powerpcspe-c.c
1782 (altivec_resolve_overloaded_builtin): Likewise.
1783 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
1785 (rs6000_preferred_simd_mode): Likewise.
1786 (output_vec_const_move): Likewise.
1787 (rs6000_expand_vector_extract): Likewise.
1788 (rs6000_split_vec_extract_var): Likewise.
1789 (reg_offset_addressing_ok_p): Likewise.
1790 (rs6000_legitimate_offset_address_p): Likewise.
1791 (rs6000_legitimize_address): Likewise.
1792 (rs6000_emit_set_const): Likewise.
1793 (rs6000_const_vec): Likewise.
1794 (rs6000_emit_move): Likewise.
1795 (spe_build_register_parallel): Likewise.
1796 (rs6000_darwin64_record_arg_recurse): Likewise.
1797 (swap_selector_for_mode): Likewise.
1798 (spe_init_builtins): Likewise.
1799 (paired_init_builtins): Likewise.
1800 (altivec_init_builtins): Likewise.
1801 (do_load_for_compare): Likewise.
1802 (rs6000_generate_compare): Likewise.
1803 (rs6000_expand_float128_convert): Likewise.
1804 (emit_load_locked): Likewise.
1805 (emit_store_conditional): Likewise.
1806 (rs6000_output_function_epilogue): Likewise.
1807 (rs6000_handle_altivec_attribute): Likewise.
1808 (rs6000_function_value): Likewise.
1809 (emit_fusion_gpr_load): Likewise.
1810 (emit_fusion_p9_load): Likewise.
1811 (emit_fusion_p9_store): Likewise.
1812 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
1813 (fusion_gpr_mem_load): Likewise.
1814 (fusion_addis_mem_combo_load): Likewise.
1815 (fusion_addis_mem_combo_store): Likewise.
1816 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
1817 (fusion_gpr_mem_load): Likewise.
1818 (fusion_addis_mem_combo_load): Likewise.
1819 (fusion_addis_mem_combo_store): Likewise.
1820 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1822 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
1823 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
1824 (rs6000_preferred_simd_mode): Likewise.
1825 (output_vec_const_move): Likewise.
1826 (rs6000_expand_vector_extract): Likewise.
1827 (rs6000_split_vec_extract_var): Likewise.
1828 (reg_offset_addressing_ok_p): Likewise.
1829 (rs6000_legitimate_offset_address_p): Likewise.
1830 (rs6000_legitimize_address): Likewise.
1831 (rs6000_emit_set_const): Likewise.
1832 (rs6000_const_vec): Likewise.
1833 (rs6000_emit_move): Likewise.
1834 (rs6000_darwin64_record_arg_recurse): Likewise.
1835 (swap_selector_for_mode): Likewise.
1836 (paired_init_builtins): Likewise.
1837 (altivec_init_builtins): Likewise.
1838 (rs6000_expand_float128_convert): Likewise.
1839 (emit_load_locked): Likewise.
1840 (emit_store_conditional): Likewise.
1841 (rs6000_output_function_epilogue): Likewise.
1842 (rs6000_handle_altivec_attribute): Likewise.
1843 (rs6000_function_value): Likewise.
1844 (emit_fusion_gpr_load): Likewise.
1845 (emit_fusion_p9_load): Likewise.
1846 (emit_fusion_p9_store): Likewise.
1847 * config/rx/rx.c (rx_gen_move_template): Likewise.
1848 (flags_from_mode): Likewise.
1849 * config/s390/predicates.md (s390_alc_comparison): Likewise.
1850 (s390_slb_comparison): Likewise.
1851 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
1852 (s390_vector_mode_supported_p): Likewise.
1853 (s390_cc_modes_compatible): Likewise.
1854 (s390_match_ccmode_set): Likewise.
1855 (s390_canonicalize_comparison): Likewise.
1856 (s390_emit_compare_and_swap): Likewise.
1857 (s390_branch_condition_mask): Likewise.
1858 (s390_rtx_costs): Likewise.
1859 (s390_secondary_reload): Likewise.
1860 (__SECONDARY_RELOAD_CASE): Likewise.
1861 (s390_expand_cs): Likewise.
1862 (s390_preferred_simd_mode): Likewise.
1863 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
1864 * config/sh/sh.c (sh_print_operand): Likewise.
1865 (dump_table): Likewise.
1866 (sh_secondary_reload): Likewise.
1867 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1868 * config/sh/sh.md (casesi_worker_1): Likewise.
1869 (casesi_worker_2): Likewise.
1870 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
1871 (fcc_comparison_operator): Likewise.
1872 * config/sparc/sparc.c (sparc_expand_move): Likewise.
1873 (emit_soft_tfmode_cvt): Likewise.
1874 (sparc_preferred_simd_mode): Likewise.
1875 (output_cbranch): Likewise.
1876 (sparc_print_operand): Likewise.
1877 (sparc_expand_vec_perm_bmask): Likewise.
1878 (vector_init_bshuffle): Likewise.
1879 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
1880 (spu_vector_mode_supported_p): Likewise.
1881 (spu_expand_insv): Likewise.
1882 (spu_emit_branch_or_set): Likewise.
1883 (spu_handle_vector_attribute): Likewise.
1884 (spu_builtin_splats): Likewise.
1885 (spu_builtin_extract): Likewise.
1886 (spu_builtin_promote): Likewise.
1887 (spu_expand_sign_extend): Likewise.
1888 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
1889 (tilegx_simd_int): Likewise.
1890 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
1891 (tilepro_simd_int): Likewise.
1892 * config/v850/v850.c (const_double_split): Likewise.
1893 (v850_print_operand): Likewise.
1894 (ep_memory_offset): Likewise.
1895 * config/vax/vax.c (vax_rtx_costs): Likewise.
1896 (vax_output_int_move): Likewise.
1897 (vax_output_int_add): Likewise.
1898 (vax_output_int_subtract): Likewise.
1899 * config/visium/predicates.md (visium_branch_operator): Likewise.
1900 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
1901 (visium_print_operand_address): Likewise.
1902 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
1903 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
1904 (xtensa_expand_conditional_branch): Likewise.
1905 (xtensa_copy_incoming_a7): Likewise.
1906 (xtensa_output_literal): Likewise.
1907 * dfp.c (decimal_real_maxval): Likewise.
1908 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
1910 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1911 Alan Hayward <alan.hayward@arm.com>
1912 David Sherwood <david.sherwood@arm.com>
1914 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
1915 (mode_nunits_inline): Likewise.
1916 (mode_inner_inline): Likewise.
1917 (mode_unit_size_inline): Likewise.
1918 (mode_unit_precision_inline): Likewise.
1919 (emit_insn_modes_h): Likewise. Also emit a #define of the
1921 (emit_mode_wider): Add an E_ prefix to mode names.
1922 (emit_mode_complex): Likewise.
1923 (emit_mode_inner): Likewise.
1924 (emit_mode_adjustments): Likewise.
1925 (emit_mode_int_n): Likewise.
1926 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
1927 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
1928 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
1929 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
1930 (CRC32_BUILTIN, ENTRY): Likewise.
1931 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
1932 (aarch64_pop_regs): Likewise.
1933 (aarch64_process_components): Likewise.
1934 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
1935 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
1936 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
1937 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
1938 * config/arm/arm.c (arm_init_libfuncs): Likewise.
1939 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
1941 * config/i386/i386-builtin.def (pcmpestr): Likewise.
1942 (pcmpistr): Likewise.
1943 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
1944 * config/mmix/mmix.c (mmix_output_condition): Likewise.
1945 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
1947 * config/rl78/rl78.c (mduc_regs): Likewise.
1948 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
1949 (htm_expand_builtin): Likewise.
1950 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
1951 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
1952 * config/xtensa/xtensa.c (print_operand): Likewise.
1953 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
1954 (NUM_MODE_VECTOR_INT): Likewise.
1955 * genoutput.c (null_operand): Likewise.
1956 (output_operand_data): Likewise.
1957 * genrecog.c (print_parameter_value): Likewise.
1958 * lra.c (debug_operand_data): Likewise.
1960 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1962 * dwarf2out.c (scompare_loc_descriptor_wide)
1963 (scompare_loc_descriptor_narrow): New functions, split out from...
1964 (scompare_loc_descriptor): ...here.
1965 * expmed.c (emit_store_flag_int): New function, split out from...
1966 (emit_store_flag): ...here.
1968 2017-08-30 Richard Biener <rguenther@suse.de>
1970 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
1971 (dwarf2out_early_finish): Move setting of AT_pubnames from
1972 early debug output to early finish.
1974 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
1976 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
1977 and -mdata-region to the assembler.
1979 2017-08-30 Richard Biener <rguenther@suse.de>
1981 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
1983 (gen_subprogram_die): Add DW_AT_object_pointer only early.
1984 (dwarf2out_early_global_decl): Only generate a DIE for the
1985 abstract origin if it doesn't already exist or is a declaration DIE.
1986 (resolve_addr): Do not add the linkage name twice when
1987 generating a stub DIE for the DW_TAG_GNU_call_site target.
1989 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
1991 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
1992 Use machine_mode rather than int for arg1_mode.
1994 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
1997 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
1998 that the second argument of the built-in functions to unpack
1999 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
2000 switch statement instead a lot of if statements.
2001 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
2002 Allow 64-bit values to be in Altivec registers as well as
2003 traditional floating point registers.
2004 (pack<mode>, FMOVE128_VSX iterator): Likewise.
2006 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
2008 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
2009 MAX_REGS_PER_ADDRESS == 1.
2011 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
2013 * config/i386/i386.opt (flag_fentry): Do not init to -1.
2014 * config/i386/i386.c (ix86_option_override_internal): Simplify
2015 setting of opts->x_flag_entry.
2017 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2018 Jakub Jelinek <jakub@redhat.com>
2019 Richard Biener <rguenther@suse.de>
2021 PR tree-optimization/81503
2022 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
2023 folded constant fits in the target type; reorder tests for clarity.
2025 2017-08-29 Martin Liska <mliska@suse.cz>
2027 * passes.def: Include pass_lower_switch.
2028 * stmt.c (dump_case_nodes): Remove and move to
2029 tree-switch-conversion.
2030 (case_values_threshold): Likewise.
2031 (expand_switch_as_decision_tree_p): Likewise.
2032 (emit_case_decision_tree): Likewise.
2033 (expand_case): Likewise.
2034 (balance_case_nodes): Likewise.
2035 (node_has_low_bound): Likewise.
2036 (node_has_high_bound): Likewise.
2037 (node_is_bounded): Likewise.
2038 (emit_case_nodes): Likewise.
2039 (struct simple_case_node): New struct.
2040 (add_case_node): Remove.
2041 (emit_case_dispatch_table): Use vector instead of case_list.
2042 (reset_out_edges_aux): Remove.
2043 (compute_cases_per_edge): Likewise.
2044 (expand_case): Build list of simple_case_node.
2045 (expand_sjlj_dispatch_table): Use it.
2046 * tree-switch-conversion.c (struct case_node): Moved from
2047 stmt.c and adjusted.
2048 (emit_case_nodes): Likewise.
2049 (node_has_low_bound): Likewise.
2050 (node_has_high_bound): Likewise.
2051 (node_is_bounded): Likewise.
2052 (case_values_threshold): Likewise.
2053 (reset_out_edges_aux): Likewise.
2054 (compute_cases_per_edge): Likewise.
2055 (add_case_node): Likewise.
2056 (dump_case_nodes): Likewise.
2057 (balance_case_nodes): Likewise.
2058 (expand_switch_as_decision_tree_p): Likewise.
2059 (emit_jump): Likewise.
2060 (emit_case_decision_tree): Likewise.
2061 (try_switch_expansion): Likewise.
2062 (do_jump_if_equal): Likewise.
2063 (emit_cmp_and_jump_insns): Likewise.
2064 (fix_phi_operands_for_edge): New function.
2065 (record_phi_operand_mapping): Likewise.
2066 (class pass_lower_switch): New pass.
2067 (pass_lower_switch::execute): New function.
2068 (make_pass_lower_switch): Likewise.
2069 (conditional_probability):
2070 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
2071 * tree-pass.h: Add make_pass_lower_switch.
2073 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
2076 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
2079 2017-08-29 Richard Biener <rguenther@suse.de>
2081 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
2082 we do not add a DW_AT_inline attribute twice.
2083 (gen_subprogram_die): Remove code setting DW_AT_inline on
2084 DECL_ABSTRACT_P nodes.
2086 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
2088 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
2089 calls to internal functions.
2090 (gimplify_modify_expr): Likewise.
2091 * tree-call-cdce.c (use_internal_fn): Likewise.
2092 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
2093 (convert_to_divmod): Set the nothrow flag.
2094 * tree-if-conv.c (predicate_mem_writes): Likewise.
2095 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
2096 (vectorizable_call): Likewise.
2097 (vectorizable_store): Likewise.
2098 (vectorizable_load): Likewise.
2099 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
2100 (vect_recog_mask_conversion_pattern): Likewise.
2102 2017-08-29 Martin Liska <mliska@suse.cz>
2105 * gcc.c (driver_handle_option): Add new argument.
2106 * opts-common.c (handle_option): Pass
2107 target_option_override_hook.
2108 * opts-global.c (lang_handle_option): Add new option.
2109 (set_default_handlers): Add new argument.
2110 (decode_options): Likewise.
2111 * opts.c (target_handle_option): Likewise.
2112 (common_handle_option): Call target_option_override_hook.
2113 * opts.h (struct cl_option_handler_func): Add hook for
2114 target option override.
2115 (struct cl_option_handlers): Likewise.
2116 (set_default_handlers): Add new argument.
2117 (decode_options): Likewise.
2118 (common_handle_option): Likewise.
2119 (target_handle_option): Likewise.
2120 * toplev.c (toplev::main): Pass targetm.target_option.override
2123 2017-08-29 Richard Biener <rguenther@suse.de>
2124 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
2126 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
2127 life to the active subtree.
2129 2017-08-28 Jeff Law <law@redhat.com>
2131 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
2132 derive_equivalences.
2133 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
2135 (edge_info::derive_equivalences): New private member function
2137 * tree-ssa-dom.c (class edge_info): Changed from a struct
2138 to a class. Add ctor/dtor, methods and data members.
2139 (edge_info::edge_info): Renamed from allocate_edge_info.
2140 Initialize additional members.
2141 (edge_info::~edge_info): New.
2142 (free_dom_edge_info): Delete the edge info.
2143 (record_edge_info): Use new class & associated member functions.
2144 Tighten forms for testing for edge equivalences.
2145 (record_temporary_equivalences): Iterate over the simple
2146 equivalences rather than assuming there's only one per edge.
2147 (cprop_into_successor_phis): Iterate over the simple
2148 equivalences rather than assuming there's only one per edge.
2149 (optimize_stmt): Use operand_equal_p rather than pointer
2150 equality for mini-DSE code.
2152 2017-08-28 Nathan Sidwell <nathan@acm.org>
2154 * gcc.c (execute): Fold SIGPIPE handling into switch
2155 statement. Adjust internal error message.
2157 2017-08-28 Richard Biener <rguenther@suse.de>
2160 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
2161 Do nothing for removed DIEs.
2163 2017-08-28 Richard Biener <rguenther@suse.de>
2165 PR tree-optimization/81977
2166 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
2169 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
2172 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
2173 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
2174 using targetm.gen_mem_thread_fence.
2176 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
2179 * config/i386/i386.md (*<btsc><mode>): Change operand 2
2180 predicate to register_operand. Reorder operands.
2181 (*btr<mode>): Ditto.
2182 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
2183 (*btr<mode>_mask): Ditto.
2185 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
2187 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
2188 * config/rs6000/xmmintrin.h: New file.
2189 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
2191 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2194 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
2195 parameter and_insn and return it.
2196 (recombine_lvx_pattern): Insert a copy to ensure availability of
2197 the base register of the copied masking operation at the point of
2198 the instruction replacement.
2199 (recombine_stvx_pattern): Likewise.
2201 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
2203 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
2204 undocumented switches.
2205 (-mpower9-dform-vector): Likewise.
2206 (-mpower9-dform): Likewise.
2207 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
2208 comments to delete references to -mpower9-dform* switches.
2209 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
2210 Delete reference to -mpower9-dform* switches, test for
2211 -mpower9-vector instead.
2212 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
2213 (OTHER_P9_VECTOR_MASKS): Likewise.
2214 (POWERPC_MASKS): Likewise.
2215 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
2216 tests against -mpower9-dform* to -mpower9-vector. Delete code
2217 that checked for -mpower9-dform* consistancy with other options.
2218 Add test for -mpower9-misc to enable other power9 switches.
2219 (rs6000_init_hard_regno_mode_ok): Likewise.
2220 (rs6000_option_override_internal): Likewise.
2221 (rs6000_emit_prologue): Likewise.
2222 (rs6000_emit_epilogue): Likewise.
2223 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
2224 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
2225 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
2227 (emit_fusion_p9_store): Likewise.
2228 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
2229 resetting these macros if the assembler does not support ISA 3.0
2231 (TARGET_P9_DFORM_VECTOR): Likewise.
2232 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
2233 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
2235 2017-08-25 Alan Modra <amodra@gmail.com>
2237 PR rtl-optimization/81747
2238 * cse.c (cse_extended_basic_block): Don't attempt to record
2239 equivalences for degenerate conditional branches that branch
2240 to their fall-through.
2242 2017-08-24 Martin Sebor <msebor@redhat.com>
2245 * gimple-fold.c (size_must_be_zero_p): New function.
2246 (gimple_fold_builtin_memory_op): Call it.
2248 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
2250 * config/rs6000/mm_malloc.h: New file.
2252 2017-08-24 Bin Cheng <bin.cheng@arm.com>
2254 PR tree-optimization/81913
2255 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
2256 analysis when either IVs in condition can wrap.
2258 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
2260 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
2261 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
2263 2017-08-24 Richard Biener <rguenther@suse.de>
2266 * targhooks.c (default_target_can_inline_p): Properly
2267 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
2268 is present and always compare.
2269 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
2270 infer -mfpmath=sse from TARGET_SSE_P.
2271 (ix86_can_inline_p): Properly use target_option_default_node when
2272 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
2274 2017-08-24 Richard Biener <rguenther@suse.de>
2277 * dwarf2out.c (output_die): Handle flag_generate_offload like
2279 (output_comp_unit): Likewise.
2280 (gen_array_type_die): Likewise.
2281 (dwarf2out_early_finish): Likewise.
2282 (note_variable_value_in_expr): Likewise.
2283 (dwarf2out_finish): Likewise. Adjust assert.
2284 * cgraphunit.c (symbol_table::compile): Move setting of
2285 flag_generate_offload earlier ...
2286 (symbol_table::finalize_compilation_unit): ... here, before
2287 early debug finalization.
2289 2017-08-24 Richard Biener <rguenther@suse.de>
2291 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
2292 and ipa-fnsummary.h.
2293 (ix86_can_inline_p): When ix86_fpmath flags do not match
2294 check whether the callee uses FP math at all.
2296 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
2299 * tree-ssanames.c (get_nonzero_bits): Use element_precision
2300 instead of TYPE_PRECISION.
2302 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
2303 Alan Hayward <alan.hayward@arm.com>
2304 David Sherwood <david.sherwood@arm.com>
2306 * combine.c (make_extraction): Use subreg_offset_from_lsb.
2308 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
2310 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
2312 (ix86_frame::stack_realign_allocate): New field.
2313 (struct machine_frame_state): Modify comments.
2314 (machine_frame_state::sp_realigned_fp_end): New field.
2315 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
2317 (sp_valid_at): Add assertion to assure no attempt to access invalid
2318 offset of a realigned stack.
2319 (fp_valid_at): Likewise.
2320 (choose_baseaddr): Modify comments.
2321 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
2322 ix86_expand_prologue.
2323 (ix86_expand_prologue): Modify stack realignment and allocation.
2324 (ix86_expand_epilogue): Modify comments.
2325 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
2326 avx2_runtime, avx512f, and avx512f_runtime.
2328 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
2330 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
2331 (mstackrealign): Do not init to -1.
2332 * config/i386/i386.c (ix86_option_override_internal):
2333 Check opts_set, not opts when setting default value of
2334 opts->x_ix86_force_align_arg_pointer.
2336 2017-08-23 Richard Biener <rguenther@suse.de>
2338 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
2339 lang_hooks.decl_printable_name.
2340 * print-rtl-function.c (print_rtx_function): Likewise.
2341 * tree-pretty-print.c (dump_function_header): Likewise.
2343 2017-08-23 Richard Biener <rguenther@suse.de>
2346 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
2347 -g0 at compile-time.
2349 2017-08-23 Tamar Christina <tamar.christina@arm.com>
2352 * doc/sourcebuild.texi (Other hardware attributes):
2355 2017-08-23 Tamar Christina <tamar.christina@arm.com>
2358 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
2359 Fix single-use check.
2361 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
2363 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
2365 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
2367 * doc/install.texi: Modify to add more details on running selected
2370 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
2372 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
2373 is combined with -mabi=ms.
2374 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
2377 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2379 PR tree-optimization/81488
2380 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
2381 and cached_basis fields.
2382 (MAX_SPREAD): New constant.
2383 (alloc_cand_and_find_basis): Initialize new fields.
2384 (clear_visited): New function.
2385 (create_phi_basis_1): Rename from create_phi_basis, set visited
2386 and cached_basis fields.
2387 (create_phi_basis): New wrapper function.
2388 (phi_add_costs_1): Rename from phi_add_costs, add spread
2389 parameter, set visited field, short-circuit when limits reached.
2390 (phi_add_costs): New wrapper function.
2391 (record_phi_increments_1): Rename from record_phi_increments, set
2393 (record_phi_increments): New wrapper function.
2394 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
2395 (phi_incr_cost): New wrapper function.
2396 (all_phi_incrs_profitable_1): Rename from
2397 all_phi_incrs_profitable, set visited field.
2398 (all_phi_incrs_profitable): New wrapper function.
2400 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
2401 Alan Hayward <alan.hayward@arm.com>
2402 David Sherwood <david.sherwood@arm.com>
2404 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
2405 that takes the outer and inner modes.
2406 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
2407 comparison as the canonical test for a paradoxical subreg.
2408 * combine.c (simplify_set): Use paradoxical_subreg_p.
2409 (make_extraction): Likewise.
2410 (force_to_mode): Likewise.
2411 (rtx_equal_for_field_assignment_p): Likewise.
2412 (gen_lowpart_for_combine): Likewise.
2413 (simplify_comparison): Likewise.
2414 * cse.c (equiv_constant): Likewise.
2415 * expmed.c (store_bit_field_1): Likewise.
2416 * final.c (alter_subreg): Likewise.
2417 * fwprop.c (propagate_rtx): Likewise.
2418 (forward_propagate_subreg): Likewise.
2419 * ira-conflicts.c (ira_build_conflicts): Likewise.
2420 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
2421 * lra-constraints.c (curr_insn_transform): Likewise.
2422 (split_reg): Likewise.
2423 * lra-eliminations.c (move_plus_up): Likewise.
2424 (lra_eliminate_regs_1): Likewise.
2425 * recog.c (general_operand): Likewise.
2426 * ree.c (combine_reaching_defs): Likewise.
2427 * reload.c (push_reload): Likewise.
2428 (find_reloads): Likewise.
2429 * reload1.c (elimination_effects): Likewise.
2430 (compute_reload_subreg_offset): Likewise.
2431 (choose_reload_regs): Likewise.
2432 * rtlanal.c (subreg_lsb_1): Likewise.
2433 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2434 (simplify_subreg): Likewise.
2435 * var-tracking.c (track_loc_p): Likewise.
2436 * emit-rtl.c (byte_lowpart_offset): Likewise.
2437 (paradoxical_subreg_p): Delete out-of-line definition.
2439 2017-08-22 Jeff Law <law@redhat.com>
2441 PR tree-optimization/81741
2442 PR tree-optimization/71947
2443 * tree-ssa-dom.c: Include tree-inline.h.
2444 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
2445 equivalences if one is more expensive to compute than the other.
2446 * tree-ssa-scopedtables.h (class const_or_copies): Make
2447 record_const_or_copy_raw method private.
2448 (class avail_exprs_stack): New method simplify_binary_operation.
2449 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
2450 avail_exprs_stack::simplify_binary_operation as needed.
2451 (avail_exprs_stack::simplify_binary_operation): New function.
2453 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
2455 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
2456 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
2457 (DOT_SYMBOLS): Likewise.
2458 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
2459 (RELOCATABLE_NEEDS_FIXUP): Likewise.
2460 (RS6000_ABI_NAME): Likewise.
2461 (TARGET_CMODEL): Likewise.
2462 (TOC_SECTION_ASM_OP): Likewise.
2463 (SET_CMODEL): New macro.
2464 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
2466 2017-08-22 Richard Biener <rguenther@suse.de>
2468 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
2469 to allow for free-lang-data replacements similar to verify_type_variant.
2471 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
2472 Alan Hayward <alan.hayward@arm.com>
2473 David Sherwood <david.sherwood@arm.com>
2475 * config/aarch64/aarch64.md (casesi): Use DImode rather than
2476 VOIDmode for the LABEL_REF.
2478 2017-08-22 Richard Biener <rguenther@suse.de>
2480 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
2482 2017-08-22 Richard Biener <rguenther@suse.de>
2484 * common.opt (feliminate-dwarf2-dups): Ignore.
2485 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
2486 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
2487 same_die_p_wrap, compute_section_prefix,
2488 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
2489 (comdat_symbol_id, comdat_symbol_number): Likewise.
2490 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
2492 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
2493 (output_die): Mark unreachable path unreachable.
2494 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
2495 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
2496 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
2497 (dwarf2out_early_finish): Likewise.
2499 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
2501 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
2503 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
2506 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
2507 not VAR_P. Filter attribute warnings with OPT_Wattributes.
2508 (avr_attribute_table) <io, io_low, address>: Initialize
2509 .decl_required with true.
2511 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
2513 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
2514 undocumented debugging options.
2515 (-mvsx-scalar-double): Likewise.
2516 (-mallow-df-permute): Likewise.
2517 (-mvectorize-builtins): Likewise.
2518 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
2519 (rs6000_builtin_vectorized_function): Likewise.
2520 (rs6000_builtin_md_vectorized_function): Likewise.
2521 (rs6000_opt_vars): Likewise.
2523 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
2526 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
2527 (*btrq_imm): Rename from *btrq.
2528 (*btcq_imm): Rename from *btcq.
2529 (btsc): New code attribute.
2530 (*<btsc><mode>): New insn pattern.
2531 (*btr<mode>): Ditto.
2532 (*<btsc><mode>_mask): New insn_and_split pattern.
2533 (*btr<mode>_mask): Ditto.
2535 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2536 Alan Hayward <alan.hayward@arm.com>
2537 David Sherwood <david.sherwood@arm.com>
2539 * function.c (pad_below): Simplify padding calculation.
2541 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2542 Alan Hayward <alan.hayward@arm.com>
2543 David Sherwood <david.sherwood@arm.com>
2545 * target.def (function_prologue): Remove frame size argument.
2546 (function_epilogue): Likewise.
2547 * doc/tm.texi: Regenerate.
2548 * final.c (final_start_function): Update call to function_prologue.
2549 (final_end_function): Update call to function_epilogue.
2550 (default_function_pro_epilogue): Remove frame size argument.
2551 * output.h (default_function_pro_epilogue): Likewise.
2552 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
2553 (arm_output_function_prologue): Likewise.
2554 * config/frv/frv.c (frv_function_prologue): Likewise.
2555 (frv_function_epilogue): Likewise.
2556 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
2557 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
2558 (ia64_output_function_epilogue): Likewise.
2559 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
2560 (m32r_output_function_epilogue): Likewise.
2561 * config/microblaze/microblaze.c (microblaze_function_prologue)
2562 (microblaze_function_epilogue): Likewise.
2563 * config/mips/mips.c (mips_output_function_prologue): Likewise.
2564 (mips_output_function_epilogue): Likewise.
2565 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
2566 (mmix_target_asm_function_epilogue): Likewise.
2567 * config/msp430/msp430.c (msp430_start_function): Likewise.
2568 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
2569 (nds32_asm_function_epilogue): Likewise.
2570 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
2571 * config/pa/pa.c (pa_output_function_prologue): Likewise.
2572 (pa_output_function_epilogue): Likewise.
2573 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
2574 (rs6000_output_function_epilogue): Likewise.
2575 * config/rl78/rl78.c (rl78_start_function): Likewise.
2576 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
2577 (rs6000_output_function_epilogue): Likewise.
2578 * config/rx/rx.c (rx_output_function_prologue): Likewise.
2579 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
2580 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
2581 (sparc_asm_function_epilogue): Likewise.
2583 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2585 * tree.h (type_has_mode_precision_p): New function.
2586 * convert.c (convert_to_integer_1): Use it.
2587 * expr.c (expand_expr_real_2): Likewise.
2588 (expand_expr_real_1): Likewise.
2589 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
2590 * match.pd: Likewise.
2591 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
2592 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
2593 * tree-tailcall.c (process_assignment): Likewise.
2594 * tree-vect-loop.c (vectorizable_reduction): Likewise.
2595 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
2596 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
2597 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
2598 (vectorizable_assignment): Likewise.
2599 (vectorizable_shift): Likewise.
2600 (vectorizable_operation): Likewise.
2601 * tree-vrp.c (register_edge_assert_for_2): Likewise.
2603 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
2605 * match.pd: Add pow (C, x) simplification.
2607 2017-08-21 Richard Biener <rguenther@suse.de>
2609 PR tree-optimization/81900
2610 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
2611 for blocks with abnormal predecessors.
2612 (compute_antic): Do not set visited flag prematurely.
2614 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
2617 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
2619 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2621 * stor-layout.h (vector_type_mode): Move to...
2622 * tree.h (vector_type_mode): ...here.
2623 * stor-layout.c (vector_type_mode): Move to...
2624 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
2626 2017-08-21 Richard Biener <rguenther@suse.de>
2628 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
2629 register_external_die hooks.
2630 (debug_false_tree_charstarstar_uhwistar): Declare.
2631 (debug_nothing_tree_charstar_uhwi): Likewise.
2632 * debug.c (do_nothing_debug_hooks): Adjust.
2633 (debug_false_tree_charstarstar_uhwistar): New do nothing.
2634 (debug_nothing_tree_charstar_uhwi): Likewise.
2635 * dbxout.c (dbx_debug_hooks): Adjust.
2636 (xcoff_debug_hooks): Likewise.
2637 * sdbout.c (sdb_debug_hooks): Likewise.
2638 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
2639 * dwarf2out.c (macinfo_label_base): New global.
2640 (dwarf2out_register_external_die): New function for the
2641 register_external_die hook.
2642 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
2643 (dwarf2_debug_hooks): Use them.
2644 (dwarf2_lineno_debug_hooks): Adjust.
2645 (struct die_struct): Add with_offset flag.
2646 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
2647 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
2648 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
2649 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
2650 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
2651 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
2652 defining section names for the early LTO debug variants.
2653 (reset_indirect_string): New helper.
2654 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
2655 (print_dw_val): Add support for offsetted symbol references.
2656 (get_ultimate_context): Split out from is_cxx.
2657 (is_cxx): Use get_ultimate_context.
2658 (is_fortran): Add decl overload.
2659 (compute_comp_unit_symbol): Split out worker from
2660 compute_section_prefix.
2661 (compute_section_prefix): Call compute_comp_unit_symbol and
2662 set comdat_type_p here.
2663 (output_die): Skip DIE symbol output for the LTO added one.
2664 Handle DIE symbol references with offset.
2665 (output_comp_unit): Guard section name mangling properly.
2666 For LTO debug sections emit a symbol at the section beginning
2667 which we use to refer to its DIEs.
2668 (add_abstract_origin_attribute): For DIEs registered via
2669 dwarf2out_register_external_die directly refer to the early
2670 DIE rather than indirectly through the shadow one we created.
2671 Remove obsolete call to dwarf2out_abstract_function for
2672 non-function/block origins.
2673 (gen_array_type_die): When generating early LTO debug do
2674 not emit DW_AT_string_length.
2675 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
2676 late when in LTO. As suggested place a gcc_unreachable for
2677 the DECL_ABSTRACT_P case.
2678 (gen_subprogram_die): Avoid another specification DIE
2679 for early built declarations/definitions for the late LTO case.
2680 (gen_variable_die): Add type references for late duplicated VLA dies
2682 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
2683 we have the abstract instance already.
2684 (process_scope_var): Adjust decl DIE contexts in LTO which
2685 first puts them in limbo.
2686 (gen_decl_die): Do not generate type DIEs late apart from
2687 types for VLAs or for decls we do not yet have a DIE. Do not
2688 call dwarf2out_abstract_function late.
2689 (dwarf2out_early_global_decl): Make sure to create DIEs
2690 for abstract instances of a decl first.
2691 (dwarf2out_late_global_decl): Adjust comment.
2692 (output_macinfo_op): With multiple macro sections use
2693 macinfo_label_base to distinguish labels.
2694 (output_macinfo): Likewise. Update macinfo_label_base.
2695 Pass in the line info label.
2696 (note_variable_value_in_expr): When generating LTO resolve
2697 all variable values here by generating DIEs as needed.
2698 (init_sections_and_labels): Add early LTO debug flag parameter
2699 and generate different sections and names if set. Add generation
2700 counter for the labels so we can have multiple of them.
2701 (reset_dies): Helper to allow DIEs to be output multiple times.
2702 (dwarf2out_finish): When outputting DIEs to the fat part of an
2703 LTO object first reset DIEs.
2704 (dwarf2out_early_finish): Output early DIEs when generating LTO.
2705 (modified_type_die): Check for decl_ultimate_origin being self
2707 (gen_type_die_with_usage): Likewise.
2708 (gen_typedef_die): Allow decl_ultimate_origin being self.
2709 (set_decl_abstract_flags): Remove.
2710 (set_block_abstract_flags): Likewise.
2711 (dwarf2out_abstract_function): Treat the early generated DIEs
2712 as the abstract copy and only add DW_AT_inline and
2713 DW_AT_artificial here and call set_decl_origin_self.
2714 If the DIE has an abstract origin don't do anything.
2715 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
2716 if we have none yet (Go fails to build one, PR78628).
2717 (variably_modified_type_p): Prevent endless recursion for Ada
2718 cyclic pointer types.
2719 * lto-streamer-in.c: Include debug.h.
2720 (dref_queue): New global.
2721 (lto_read_tree_1): Stream in DIE references.
2722 (lto_input_tree): Register DIE references.
2723 (input_function): Stream DECL_DEBUG_ARGS.
2724 * lto-streamer-out.c: Include debug.h.
2725 (lto_write_tree_1): Output DIE references.
2726 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
2727 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
2728 (output_function): Stream DECL_DEBUG_ARGS.
2729 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
2730 Stream DECL_ABSTRACT_ORIGIN.
2731 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
2732 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
2733 DECL_CONTEXT for file-scope decls.
2734 * lto-streamer.h (struct dref_entry): Declare.
2735 (dref_queue): Likewise.
2736 * cfgexpand.c (pass_expand::execute): Do not call the
2737 outlining_inline_function hook here.
2738 * lto-wrapper.c (debug_obj): New global.
2739 (tool_cleanup): Unlink it if required.
2740 (debug_objcopy): New function.
2741 (run_gcc): Handle early debug sections in the IL files by
2742 extracting them to separate files, partially linkin them and
2743 feeding the result back as result to the linker.
2744 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
2745 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
2746 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
2747 sections into a separate segment.
2748 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
2750 (darwin_asm_dwarf_section): Likewise.
2751 (darwin_asm_output_dwarf_offset): Likewise.
2752 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
2754 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2755 Alan Hayward <alan.hayward@arm.com>
2756 David Sherwood <david.sherwood@arm.com>
2758 * read-md.h (md_reader::record_potential_iterator_use): Replace
2759 pointer argument with an rtx and an index.
2760 * read-rtl.c (iterator_group::apply_iterator): Likewise.
2761 (apply_mode_iterator): Likewise.
2762 (apply_code_iterator): Likewise.
2763 (apply_int_iterator): Likewise.
2764 (apply_subst_iterator): Likewise.
2765 (record_iterator_use): Likewise.
2766 (record_attribute_use): Likewise.
2767 (md_reader::record_potential_iterator_use): Likewise. Update calls
2768 to record_iterator_use and apply_iterator.
2769 (iterator_use): Replace ptr with x and index.
2770 (attribute_use): Likewise.
2771 (apply_attribute_uses): Update calls to apply_iterator.
2772 (apply_iterators): Likewise. Update initialization of iterator_use.
2773 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
2774 and record_potential_iterator_use.
2775 (rtx_reader::read_rtx_operand): Likewise.
2777 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
2778 Alan Hayward <alan.hayward@arm.com>
2779 David Sherwood <david.sherwood@arm.com>
2781 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
2784 2017-08-21 Richard Biener <rguenther@suse.de>
2787 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
2788 at struct end conservatively when comparing common bases.
2790 2017-08-21 Richard Biener <rguenther@suse.de>
2792 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
2793 (mem_ref_in_stmt): Remove.
2794 (determine_max_movement): Use ref index to get at the reference.
2795 (invariantness_dom_walker::before_dom_children): Deal with
2796 lim data already initialized.
2797 (gather_mem_refs_stmt): Initialize lim data and record ref index.
2799 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2801 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
2802 (TARGET_ISA_ROUND): Ditto.
2803 (TARGET_ROUND): Ditto.
2804 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
2805 * config/i386/i386.md: Ditto.
2806 * config/i386/sse.md: Ditto.
2807 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
2808 with OPTION_MASK_ISA_SSE4_1.
2810 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
2813 * doc/extend.texi (x86 Built-in Functions): Correct the name of
2814 __builtin_ia32_lzcnt_u16.
2816 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
2819 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
2820 (rs6000_set_current_function): Rewrite function to use it.
2822 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
2825 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
2826 and TYPE_WARN_IF_NOT_ALIGN.
2827 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
2828 (handle_warn_if_not_align): New.
2829 (place_union_field): Call handle_warn_if_not_align.
2830 (place_field): Call handle_warn_if_not_align.
2831 Copy TYPE_WARN_IF_NOT_ALIGN.
2832 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
2833 (layout_type): Likewise.
2834 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
2836 (tree_decl_common): Add warn_if_not_align.
2837 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
2838 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
2839 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
2840 (DECL_WARN_IF_NOT_ALIGN): Likewise.
2841 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
2842 * doc/extend.texi: Document warn_if_not_aligned attribute.
2843 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
2845 2017-08-17 Martin Liska <mliska@suse.cz>
2848 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
2849 (get_data_dependence): Use it as pointer type.
2850 (distribute_loop): Likewise.
2852 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2854 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
2855 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
2856 (p8_vmrgow_<mode>_direct): New define_insn.
2857 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
2858 handle endianness for vmrgew and vmrgow permute patterns.
2860 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2862 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
2863 * config/rs6000/rs6000-cpus.def: Remove comment.
2864 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
2865 (POWERPC_MASKS): Likewise.
2866 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
2867 use of TARGET_VSX_TIMODE.
2868 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
2869 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
2870 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
2871 (rs6000_option_override_internal): Remove dead code.
2872 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
2873 (rs6000_legitimize_reload_address): Likewise.
2874 (rs6000_legitimate_address_p): Likewise.
2875 (rs6000_opt_masks): Delete "vsx-timode".
2876 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
2877 from function comment.
2878 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
2879 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
2880 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
2882 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
2883 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
2884 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
2887 (define_peephole2): Likewise.
2889 2017-08-17 Martin Sebor <msebor@redhat.com>
2892 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
2893 past the end of an array.
2894 (test_pp_format): Add test cases.
2896 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
2898 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
2899 missing ECF_NOTHROW flags.
2901 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
2904 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
2905 operands residing in integer registers.
2906 (*vsx_le_perm_load_<mode>): Likewise.
2907 (*vsx_le_perm_store_<mode>): Likewise.
2908 (define_peephole2): Add peepholes to optimize the above.
2910 2017-08-17 Marek Polacek <polacek@redhat.com>
2913 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
2914 to mimic what shorten_compare did. Change the return type to bool.
2915 (fold_cond_expr_with_comparison): Update call to
2916 operand_equal_for_comparison_p.
2917 (fold_ternary_loc): Likewise.
2919 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
2921 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
2923 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
2925 2017-08-17 Richard Biener <rguenther@suse.de>
2927 * tree-ssa-structalias.c (solve_graph): When propagating
2928 to successors update the graphs succ edges and avoid duplicate work.
2930 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
2933 * config/i386/i386.c (ix86_option_override_internal): Save target
2934 specific options after ix86_stack_protector_guard_reg was changed.
2936 2017-08-17 Richard Biener <rguenther@suse.de>
2938 PR tree-optimization/81827
2939 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
2940 (new_var_info): Initialize it conservatively.
2941 (get_call_vi): Mark register vars.
2942 (new_scalar_tmp_constraint_exp): Likewise.
2943 (handle_rhs_call): Likewise.
2944 (handle_const_call): Likewise.
2945 (create_function_info_for): Likewise.
2946 (solve_constraints): Sort varinfos to separate register from
2947 non-register vars to pack points-to solution bitmaps during
2950 2017-08-17 Marek Polacek <polacek@redhat.com>
2952 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
2954 2017-08-17 Richard Biener <rguenther@suse.de>
2956 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
2957 to true when overflow is undefined and we saturated the result.
2959 2017-08-17 Alan Modra <amodra@gmail.com>
2962 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
2963 Don't use store multiple if only one reg needs saving.
2964 (interesting_frame_related_regno): New function.
2965 (rs6000_frame_related): Don't emit frame info for regs that
2967 (rs6000_emit_epilogue): Likewise.
2969 2017-08-16 Nathan Sidwell <nathan@acm.org>
2971 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
2972 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
2973 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
2974 * tree.c (free_lang_data_in_type): Use else-if chain. Always
2975 clear TYPE_LANG_1. Remove obsolete member-function stripping.
2976 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
2977 (verify_type): Adjust for TYPE_BINFO move.
2978 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
2979 process TYPE_BINFO directly.
2980 (hash_tree): Likewise.
2981 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
2983 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
2986 2017-08-16 David Malcolm <dmalcolm@redhat.com>
2988 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
2990 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
2993 * config/i386/i386.md (*anddi_1_btr): Change predicates of
2994 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
2995 Add ix86_binary_operator_ok to insn constraint.
2996 (*iordi_1_bts): Ditto.
2997 (*xordi_1_btc): Ditto.
2998 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
2999 Update corresponding peephole2 pattern.
3003 2017-08-16 Bin Cheng <bin.cheng@arm.com>
3005 PR tree-optimization/81832
3006 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
3007 copy loop header which has IFN_LOOP_DIST_ALIAS call.
3009 2017-08-16 Marek Polacek <polacek@redhat.com>
3012 * fold-const.c (fold_indirect_ref_1): Restore original behavior
3013 regarding size_zero_node.
3015 2017-08-16 Martin Liska <mliska@suse.cz>
3018 * config.gcc: Respect previously set extra_objs in case
3021 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
3023 PR tree-optimization/81835
3024 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
3025 the phi SSA_NAME. Check that the condition in a COND_EXPR does
3026 not depend on the phi.
3028 2017-08-16 Alan Modra <amodra@gmail.com>
3030 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
3033 2017-08-16 Alan Modra <amodra@gmail.com>
3035 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
3036 (save_reg_p): ..into this. Update all callers.
3037 (first_reg_to_save): Simplify.
3039 2017-08-16 Alan Modra <amodra@gmail.com>
3041 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
3044 2017-08-15 Joseph Myers <joseph@codesourcery.com>
3048 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
3049 constant count if that count is less than 32.
3051 2017-08-15 Nathan Sidwell <nathan@acm.org>
3053 * gcc.c (execute): Emit friendlier message if inferior is killed
3054 by an external cause.
3056 2017-08-15 Richard Biener <rguenther@suse.de>
3058 PR tree-optimization/81790
3059 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
3060 CONSTRUCTORs from simplifying and VN.
3062 2017-08-14 Martin Sebor <msebor@redhat.com>
3064 * builtin-attrs.def: Add comments.
3066 2017-08-14 Martin Sebor <msebor@redhat.com>
3069 * doc/extend.texi (attribute nonstring): Document new attribute.
3071 2017-08-14 Martin Sebor <msebor@redhat.com>
3074 * tree-diagnostic.c (default_tree_printer): Handle %G.
3075 * gimple-pretty-print.h (percent_G_format): Declare new function.
3076 * gimple-pretty-print.c (percent_G_format): Define.
3077 * tree-pretty-print.c (percent_K_format): Add argument.
3079 2017-08-14 Martin Sebor <msebor@redhat.com>
3081 PR translation/79998
3082 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
3083 Remove a stray space.
3085 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
3088 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
3089 (*iordi_1_bts): Ditto.
3090 (*xordi_1_btc): Ditto.
3092 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3095 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
3096 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3098 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
3099 quoted strings, and make more translator-friendly.
3100 (darwin_rs6000_override_options): Likewise.
3101 (rs6000_option_override_internal): Likewise.
3102 (rs6000_return_in_memory): Fix overlong line.
3103 (init_cmulative_args): Use quoted strings, and make more
3104 translator-friendly.
3105 (rs6000_pass_by_reference): Fix overlong line.
3106 (def_builtin): Use quoted strings.
3107 (altivec_expand_predicate_builtin): Use quoted strings, and make
3108 more translator-friendly.
3109 (htm_expand_builtin): Use quoted strings.
3110 (cpu_expand_builtin): Use quoted strings, and make more
3111 translator-friendly.
3112 (altivec_expand_builtin): Likewise.
3113 (paired_expand_predicate_builtin): Likewise.
3114 (rs6000_invalid_builtin): Likewise.
3115 (builtin_function_type): Use quoted strings.
3116 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
3117 more translator-friendly.
3118 (rs6000_trampoline_init): Likewise.
3119 (rs6000_handle_altivec_attribute): Likewise.
3120 (rs6000_inner_target_options): Use quoted strings.
3121 (rs6000_disable_incompatible_switches): Likewise.
3122 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
3123 strings, and make more translator-friendly.
3124 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
3126 2017-08-14 Bin Cheng <bin.cheng@arm.com>
3128 PR tree-optimization/81799
3129 * tree-loop-distribution.c (version_loop_by_alias_check): Force
3130 cond_expr to simple gimple operand.
3132 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
3135 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
3137 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
3141 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
3142 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
3143 TARGET_GASISR_PROLOGUES.
3144 * config/avr/avr.c (avr_option_override): Same.
3145 (avr_pass_pre_proep::execute): Same.
3147 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
3150 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
3151 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
3153 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
3155 * config/i386/i386.md (*load_tp_<mode>): Redefine as
3156 define_insn_and_split. Split to a memory load from 0 in
3157 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
3158 using PTR mode iterator.
3159 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
3160 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
3161 (*add_tp_<mode>): Redefine as define_insn_and_split.
3162 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
3163 address space. Merge with *add_tp_x32 using PTR mode iterator.
3164 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
3165 Split to an add with a memory load from 0 in
3166 DEFAULT_TLS_SEG_REG address space.
3168 2017-08-12 Andrew Pinski <apinski@cavium.com>
3170 * config/aarch64/aarch64-option-extensions.def (rdma):
3171 Fix feature string to what Linux prints out in /proc/cpuinfo.
3173 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
3176 * dwarf2out.c (modified_type_die): For C typedef types that have
3177 an ultimate origin, process the ultimate origin instead of the
3179 (gen_typedef_die): Assert that input DECLs have no ultimate
3181 (gen_type_die_with_usage): For typedef variants that have an
3182 ultimate origin, just call gen_decl_die on the original DECL.
3183 (process_scope_var): Avoid creating DIEs for local typedefs and
3184 concrete static variables.
3186 2017-08-12 Alan Modra <amodra@gmail.com>
3190 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
3191 match gnu-user.h startfile.
3192 (ENDFILE_LINUX_SPEC): Similarly.
3194 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
3197 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
3199 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
3201 2017-08-11 Tamar Christina <tamar.christina@arm.com>
3202 * config/aarch64/aarch64.md (mov<mode>): Change.
3203 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
3204 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
3205 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
3207 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
3209 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
3210 for storage order barriers.
3212 2017-08-11 Martin Liska <mliska@suse.cz>
3215 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
3216 variables of void type.
3218 2017-08-11 Martin Liska <mliska@suse.cz>
3220 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
3221 TARGET_SUPPORTS_ALIASES.
3222 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
3223 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
3224 (optimize_weakref): Likewise.
3225 * symtab.c (symtab_node::noninterposable_alias): Likewise.
3226 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
3227 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
3229 2017-08-11 Martin Liska <mliska@suse.cz>
3232 * config/i386/i386.c (make_resolver_func): Do complete
3233 refactoring of the function.
3235 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
3238 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
3239 * config/i386/i386.c (ix86_stack_protect_guard): Use
3240 ix86_stack_protect_guard_symbol_str to generate varible declaration.
3241 * doc/invoke.texi (x86 Options): Document
3242 -mstack-protector-guard-symbol= option.
3244 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
3246 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
3247 * config/i386/i386.c (ix86_split_stack_guard): New function.
3248 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
3249 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
3250 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
3251 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
3252 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
3253 (split_stack_space_check): Call ix86_split_stack_guard.
3255 2017-08-10 Martin Sebor <msebor@redhat.com>
3257 * print-tree.c (print_node): Print location using the established
3259 Replace spaces with colons.
3260 (debug_raw, debug): Ditto.
3262 2017-08-10 Martin Sebor <msebor@redhat.com>
3265 * pretty-print.c (pp_format): Correct the handling of %s precision.
3267 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
3270 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
3272 (ix86_finalize_stack_frame_flags): This. Also clear
3273 frame_pointer_needed if -fno-omit-frame-pointer is used without
3275 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
3276 with ix86_finalize_stack_frame_flags.
3277 (ix86_expand_epilogue): Likewise.
3278 (ix86_expand_split_stack_prologue): Likewise.
3279 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
3281 2017-08-10 Martin Liska <mliska@suse.cz>
3284 * c-attribs.c (handle_target_attribute):
3285 Report warning for an empty string argument of target attribute.
3287 2017-08-09 Jakub Jelinek <jakub@redhat.com>
3290 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
3292 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
3293 or DECL_NONLOCAL labels.
3294 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
3295 or DECL_NONLOCAL labels here.
3297 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
3299 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
3300 to indicate when early gimple folding has been disabled.
3301 (rs6000_gimple_fold_builtin): Add debug content.
3302 (rs6000_invalid_builtin): Fix whitespace.
3303 (rs6000_expand_builtin): Fix whitespace.
3304 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
3306 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
3309 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
3310 SAVE_MULTIPLE if not all the registers that saves, should be saved.
3312 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
3314 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
3315 (qdf24xx): Likewise.
3316 * config/aarch64/aarch64.md: Include falkor.md.
3317 * config/aarch64/falkor.md: New.
3319 2017-08-09 Marek Polacek <polacek@redhat.com>
3322 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
3323 * diagnostic.c (emit_diagnostic): Add a comment.
3324 (emit_diagnostic_valist): New function.
3326 2017-08-09 Marek Polacek <polacek@redhat.com>
3329 * input.c (make_location): New overload.
3330 * input.h (make_location): Declare.
3332 2017-08-08 Alan Modra <amodra@gmail.com>
3333 H.J. Lu <hongjiu.lu@intel.com>
3336 * gcc.c (NO_PIE_SPEC): Delete.
3337 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
3339 (LINK_PIE_SPEC): ..to here.
3340 (LINK_COMMAND_SPEC): Support -no-pie.
3341 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
3342 chain of crtbegin*.o selection, update for PIE_SPEC changes and
3344 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
3345 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
3346 (ENDFILE_CRTEND_SPEC): Similarly.
3348 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
3351 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
3352 (mstack-protector-guard-offset=): Ditto.
3353 * config/i386/i386.c (ix86_option_override): Handle
3354 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
3356 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
3357 ix86_stack_protect_guard_offset variables.
3358 (TARGET_STACK_PROTECT_GUARD): Always define.
3359 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
3360 and -mstack-protector-guard-offset= options.
3362 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3364 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
3365 boundary case for the last candidate.
3367 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3369 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
3370 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
3372 2017-08-08 Tamar Christina <tamar.christina@arm.com>
3375 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
3376 * config/aarch64/aarch64-builtins.c
3377 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
3378 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
3379 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
3381 2017-08-08 Tamar Christina <tamar.christina@arm.com>
3382 Andrew Pinski <pinskia@gmail.com>
3385 * internal-fn.def (XORSIGN): New.
3386 * optabs.def (xorsign_optab): New.
3387 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
3388 (convert_expand_mult_copysign): New.
3389 (pass_optimize_widening_mul::execute): Call
3390 convert_expand_mult_copysign.
3392 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3394 PR tree-optimization/81354
3395 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
3396 Insert on edges rather than explicitly creating landing pads.
3397 (analyze_candidates_and_replace): Commit edge inserts.
3399 2017-08-08 Richard Biener <rguenther@suse.de>
3402 * tree-ssa-loop-niter.c: Include tree-dfa.h.
3403 (expand_simple_operations): Also look through ADDR_EXPRs with
3404 MEM_REF bases treating them as POINTER_PLUS_EXPR.
3406 2017-08-08 Richard Biener <rguenther@suse.de>
3408 PR tree-optimization/81723
3409 * tree-vect-slp.c (struct bst_traits): New hash traits.
3410 (bst_fail): New global.
3411 (vect_build_slp_tree_2): New worker, split out from ...
3412 (vect_build_slp_tree): ... this now wrapping it with using
3413 bst_fail set to cache SLP tree build fails. Properly handle
3415 (vect_analyze_slp_instance): Allocate and free bst_fail.
3417 2017-08-08 Martin Liska <mliska@suse.cz>
3420 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
3421 LABEL_DECLs that can be from a different function.
3423 2017-08-08 Bin Cheng <bin.cheng@arm.com>
3425 PR tree-optimization/81744
3426 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
3427 loop's number of iterations.
3429 2017-08-08 Martin Liska <mliska@suse.cz>
3431 * asan.c: Include header files.
3432 * attribs.c (build_decl_attribute_variant): New function moved
3434 (build_type_attribute_qual_variant): Likewise.
3435 (cmp_attrib_identifiers): Likewise.
3436 (simple_cst_list_equal): Likewise.
3437 (omp_declare_simd_clauses_equal): Likewise.
3438 (attribute_value_equal): Likewise.
3439 (comp_type_attributes): Likewise.
3440 (build_type_attribute_variant): Likewise.
3441 (lookup_ident_attribute): Likewise.
3442 (remove_attribute): Likewise.
3443 (merge_attributes): Likewise.
3444 (merge_type_attributes): Likewise.
3445 (merge_decl_attributes): Likewise.
3446 (merge_dllimport_decl_attributes): Likewise.
3447 (handle_dll_attribute): Likewise.
3448 (attribute_list_equal): Likewise.
3449 (attribute_list_contained): Likewise.
3450 * attribs.h (lookup_attribute): New function moved from tree.[ch].
3451 (lookup_attribute_by_prefix): Likewise.
3452 * bb-reorder.c: Include header files.
3453 * builtins.c: Likewise.
3454 * calls.c: Likewise.
3455 * cfgexpand.c: Likewise.
3456 * cgraph.c: Likewise.
3457 * cgraphunit.c: Likewise.
3458 * convert.c: Likewise.
3459 * dwarf2out.c: Likewise.
3460 * final.c: Likewise.
3461 * fold-const.c: Likewise.
3462 * function.c: Likewise.
3463 * gimple-expr.c: Likewise.
3464 * gimple-fold.c: Likewise.
3465 * gimple-pretty-print.c: Likewise.
3466 * gimple.c: Likewise.
3467 * gimplify.c: Likewise.
3468 * hsa-common.c: Likewise.
3469 * hsa-gen.c: Likewise.
3470 * internal-fn.c: Likewise.
3471 * ipa-chkp.c: Likewise.
3472 * ipa-cp.c: Likewise.
3473 * ipa-devirt.c: Likewise.
3474 * ipa-fnsummary.c: Likewise.
3475 * ipa-inline.c: Likewise.
3476 * ipa-visibility.c: Likewise.
3478 * lto-cgraph.c: Likewise.
3479 * omp-expand.c: Likewise.
3480 * omp-general.c: Likewise.
3481 * omp-low.c: Likewise.
3482 * omp-offload.c: Likewise.
3483 * omp-simd-clone.c: Likewise.
3484 * opts-global.c: Likewise.
3485 * passes.c: Likewise.
3486 * predict.c: Likewise.
3487 * sancov.c: Likewise.
3488 * sanopt.c: Likewise.
3489 * symtab.c: Likewise.
3490 * toplev.c: Likewise.
3491 * trans-mem.c: Likewise.
3492 * tree-chkp.c: Likewise.
3493 * tree-eh.c: Likewise.
3494 * tree-into-ssa.c: Likewise.
3495 * tree-object-size.c: Likewise.
3496 * tree-parloops.c: Likewise.
3497 * tree-profile.c: Likewise.
3498 * tree-ssa-ccp.c: Likewise.
3499 * tree-ssa-live.c: Likewise.
3500 * tree-ssa-loop.c: Likewise.
3501 * tree-ssa-sccvn.c: Likewise.
3502 * tree-ssa-structalias.c: Likewise.
3503 * tree-ssa.c: Likewise.
3504 * tree-streamer-in.c: Likewise.
3505 * tree-vectorizer.c: Likewise.
3506 * tree-vrp.c: Likewise.
3508 * ubsan.c: Likewise.
3509 * varasm.c: Likewise.
3510 * varpool.c: Likewise.
3511 * tree.c: Remove functions moved to attribs.[ch].
3513 * config/aarch64/aarch64.c: Add attrs.h header file.
3514 * config/alpha/alpha.c: Likewise.
3515 * config/arc/arc.c: Likewise.
3516 * config/arm/arm.c: Likewise.
3517 * config/avr/avr.c: Likewise.
3518 * config/bfin/bfin.c: Likewise.
3519 * config/c6x/c6x.c: Likewise.
3520 * config/cr16/cr16.c: Likewise.
3521 * config/cris/cris.c: Likewise.
3522 * config/darwin.c: Likewise.
3523 * config/epiphany/epiphany.c: Likewise.
3524 * config/fr30/fr30.c: Likewise.
3525 * config/frv/frv.c: Likewise.
3526 * config/ft32/ft32.c: Likewise.
3527 * config/h8300/h8300.c: Likewise.
3528 * config/i386/winnt.c: Likewise.
3529 * config/ia64/ia64.c: Likewise.
3530 * config/iq2000/iq2000.c: Likewise.
3531 * config/lm32/lm32.c: Likewise.
3532 * config/m32c/m32c.c: Likewise.
3533 * config/m32r/m32r.c: Likewise.
3534 * config/m68k/m68k.c: Likewise.
3535 * config/mcore/mcore.c: Likewise.
3536 * config/microblaze/microblaze.c: Likewise.
3537 * config/mips/mips.c: Likewise.
3538 * config/mmix/mmix.c: Likewise.
3539 * config/mn10300/mn10300.c: Likewise.
3540 * config/moxie/moxie.c: Likewise.
3541 * config/msp430/msp430.c: Likewise.
3542 * config/nds32/nds32-isr.c: Likewise.
3543 * config/nds32/nds32.c: Likewise.
3544 * config/nios2/nios2.c: Likewise.
3545 * config/nvptx/nvptx.c: Likewise.
3546 * config/pa/pa.c: Likewise.
3547 * config/pdp11/pdp11.c: Likewise.
3548 * config/powerpcspe/powerpcspe.c: Likewise.
3549 * config/riscv/riscv.c: Likewise.
3550 * config/rl78/rl78.c: Likewise.
3551 * config/rx/rx.c: Likewise.
3552 * config/s390/s390.c: Likewise.
3553 * config/sh/sh.c: Likewise.
3554 * config/sol2.c: Likewise.
3555 * config/sparc/sparc.c: Likewise.
3556 * config/spu/spu.c: Likewise.
3557 * config/stormy16/stormy16.c: Likewise.
3558 * config/tilegx/tilegx.c: Likewise.
3559 * config/tilepro/tilepro.c: Likewise.
3560 * config/v850/v850.c: Likewise.
3561 * config/vax/vax.c: Likewise.
3562 * config/visium/visium.c: Likewise.
3563 * config/xtensa/xtensa.c: Likewise.
3565 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
3568 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
3569 constraints since the -mupper-regs-* switches have been
3571 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
3572 into a vector from a double word element that was extracted from
3573 another vector, and eliminate extra XXPERMDI instructions.
3574 (vsx_concat_<mode>_2): Likewise.
3575 (vsx_concat_<mode>_3): Likewise.
3576 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
3577 concat to allow optimizing inserts from previous extracts.
3579 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
3581 * config/i386/i386.c (ix86_stack_protect_guard): Generate
3582 memory reference to a SSP offset in TLS address space.
3583 (ix86_print_operand) <case '@'>: Remove.
3584 (ix86_print_operand_punct_valid_p): Remove '@' code.
3585 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
3587 (stack_tls_protect_set_<mode>): Remove.
3588 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
3589 (stack_tls_protect_test_<mode>): Remove.
3590 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
3592 2017-08-07 Olivier Hainque <hainque@adacore.com>
3595 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
3597 2017-08-07 Douglas Rupp <rupp@adacore.com>
3599 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
3600 variable was referenced as multidir in command.
3602 2017-08-07 Jakub Jelinek <jakub@redhat.com>
3605 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
3608 2017-08-07 Martin Liska <mliska@suse.cz>
3610 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
3611 * config/rl78/rl78.c: Add include of attribs.h.
3612 * config/sh/sh.c: Likewise.
3613 * config/v850/v850.c: Likewise.
3615 2017-08-07 Tom de Vries <tom@codesourcery.com>
3618 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
3620 2017-08-07 Martin Liska <mliska@suse.cz>
3622 * config/mips/mips.c: Include attribs.h.
3624 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
3627 * doc/invoke.texi: Document change in behvaior for -Ofast for
3630 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
3632 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
3634 (aarch64_pop_regs): Likewise.
3635 (aarch64_gen_load_pair): Likewise.
3636 (aarch64_save_callee_saves): Likewise.
3637 (aarch64_restore_callee_saves): Likewise.
3639 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3641 * config/i386/i386.c: Revert the last change.
3643 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3646 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
3648 (ix86_finalize_stack_frame_flags): This. Also clear
3649 frame_pointer_needed if -fno-omit-frame-pointer is used without
3651 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
3652 with ix86_finalize_stack_frame_flags.
3653 (ix86_expand_epilogue): Likewise.
3654 (ix86_expand_split_stack_prologue): Likewise.
3656 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
3659 * config/i386/i386.c (get_builtin_code_for_version): Set priority
3660 to P_AES for Westmere.
3662 2017-08-07 Jonathan Yong <10walls@gmail.com>
3664 * config/i386/mingw.opt (fset-stack-executable): Removed.
3665 * config/i386/cygming.opt (fset-stack-executable): Moved
3667 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
3669 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
3671 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
3673 2017-08-07 Marek Polacek <polacek@redhat.com>
3676 * fold-const.c (fold_indirect_ref_1): Check type_domain.
3678 2017-08-07 Martin Liska <mliska@suse.cz>
3680 * attribs.h (canonicalize_attr_name): New function.
3681 (cmp_attribs): Move from c-format.c and adjusted.
3682 (is_attribute_p): Moved from tree.h.
3683 * tree-inline.c: Add new includes.
3684 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
3685 (private_is_attribute_p): Remove.
3686 (private_lookup_attribute): Likewise.
3687 (private_lookup_attribute_by_prefix): Simplify.
3688 (remove_attribute): Use is_attribute_p.
3689 * tree.h: Remove removed declarations.
3691 2017-08-07 Jakub Jelinek <jakub@redhat.com>
3694 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
3695 instead of computing it in the function. Formatting fix.
3696 (expand_case): Don't rely on default_edge being the first edge,
3697 clear it if removing it, pass default_edge to
3698 emit_case_dispatch_table.
3699 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
3702 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
3704 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
3705 insn in the function, emit NOP after the insn.
3707 2017-08-06 Tom de Vries <tom@codesourcery.com>
3709 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
3712 2017-08-06 Tom de Vries <tom@codesourcery.com>
3714 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
3717 2017-08-04 Yury Gribov <tetra2005@gmail.com>
3719 PR tree-optimization/57371
3720 * match.pd: New pattern.
3722 2017-08-04 Marek Polacek <polacek@redhat.com>
3725 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
3726 perform the computation in offset_int.
3728 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3730 PR tree-optimization/81136
3731 * tree-vectorizer.h: Include tree-hash-traits.h.
3732 (vec_base_alignments): New typedef.
3733 (vec_info): Add a base_alignments field.
3734 (vect_record_base_alignments): Declare.
3735 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
3737 (DR_IS_CONDITIONAL_IN_STMT): New macro.
3738 (create_data_ref): Add an is_conditional_in_stmt argument.
3739 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
3740 the is_conditional_in_stmt field.
3741 (data_ref_loc): Add an is_conditional_in_stmt field.
3742 (get_references_in_stmt): Set the is_conditional_in_stmt field.
3743 (find_data_references_in_stmt): Update call to create_data_ref.
3744 (graphite_find_data_references_in_stmt): Likewise.
3745 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
3746 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
3747 (vect_record_base_alignment): New function.
3748 (vect_record_base_alignments): Likewise.
3749 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
3750 for nested statements even if we fail to compute a misalignment.
3751 Use pooled base alignments for unconditional references.
3752 (vect_find_same_alignment_drs): Compare base addresses instead
3754 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
3755 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
3757 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3759 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
3760 Add an explicit name for the enum. Use auto_vec for slp_instances
3762 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
3764 (_bb_vec_info): Add a constructor and destructor.
3765 (vinfo_for_stmt): Return NULL for uids of -1 as well.
3766 (destroy_loop_vec_info): Delete.
3767 (vect_destroy_datarefs): Likewise.
3768 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
3769 (vec_info::vec_info): New function.
3770 (vec_info::~vec_info): Likewise.
3771 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
3772 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
3773 destroy_loop_vec_info.
3774 * tree-vect-loop.c (new_loop_vec_info): Replace with...
3775 (_loop_vec_info::_loop_vec_info): ...this.
3776 (destroy_loop_vec_info): Replace with...
3777 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
3778 the stmt_vec_infos. Leave handling of vec_info information to its
3779 destructor. Remove explicit vector releases.
3780 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
3781 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
3782 * tree-vect-slp.c (new_bb_vec_info): Replace with...
3783 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
3784 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
3785 (destroy_bb_vec_info): Replace with...
3786 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
3787 information to its destructor.
3788 (vect_slp_analyze_bb_1): Use new and delete instead of
3789 new_bb_vec_info and destroy_bb_vec_info.
3790 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
3793 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
3795 * tree-data-ref.h (subscript): Add access_fn field.
3796 (data_dependence_relation): Add could_be_independent_p.
3797 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
3798 (same_access_functions): Move to tree-data-ref.c.
3799 * tree-data-ref.c (ref_contains_union_access_p): New function.
3800 (access_fn_component_p): Likewise.
3801 (access_fn_components_comparable_p): Likewise.
3802 (dr_analyze_indices): Add a reference to access_fn_component_p.
3803 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
3805 (constant_access_functions): Likewise.
3806 (add_other_self_distances): Likewise.
3807 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
3808 (initialize_data_dependence_relation): Use XCNEW and remove
3809 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
3810 of access functions that have the same type. Allow the
3811 subsequence to end with different bases in some circumstances.
3812 Record the chosen access functions in SUB_ACCESS_FN.
3813 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
3814 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
3815 (subscript_dependence_tester_1): Likewise dra and drb.
3816 (build_classic_dist_vector): Update calls accordingly.
3817 (subscript_dependence_tester): Likewise.
3818 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
3819 DDR_COULD_BE_INDEPENDENT_P.
3820 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
3821 comp_alias_ddrs instead of may_alias_ddrs.
3822 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
3824 (vect_analyze_data_ref_dependence): Use it if
3825 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
3826 distance vectors if that fails.
3827 (dependence_distance_ge_vf): New function.
3828 (vect_prune_runtime_alias_test_list): Use it. Don't clear
3829 LOOP_VINFO_MAY_ALIAS_DDRS.
3831 2017-08-04 Richard Biener <rguenther@suse.de>
3834 * fold-const.c (fold_binary_loc): Properly restrict
3835 minus_var0 && minus_var1 case when associating undefined overflow
3838 2017-08-04 Tom de Vries <tom@codesourcery.com>
3840 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
3842 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3844 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
3845 Don't start diagnostic messages with a capital letter.
3846 * config/rs6000/rs6000.c (rs6000_option_override_internal):
3848 (rs6000_invalid_builtin): Likewise.
3849 (rs6000_trampoline_init): Likewise.
3851 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3854 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
3855 after setting changeable df flags.
3857 2017-08-03 Richard Biener <rguenther@suse.de>
3859 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
3860 up if the use is in USE - X.
3862 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
3864 * toplev.c (dumpfile.h): New include.
3865 (internal_error_reentered): New static function. Use it...
3866 (internal_error_function): ...here to handle reentered internal_error.
3868 2017-08-03 Richard Biener <rguenther@suse.de>
3871 * fold-const.c (split_tree): Add minus_var and minus_con
3872 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
3873 here but always use minus_*.
3874 (associate_trees): Assert we never associate with MINUS_EXPR
3875 and NULL first operand. Do not recurse for PLUS_EXPR operands
3876 when associating as MINUS_EXPR either.
3877 (fold_binary_loc): Track minus_var and minus_con.
3879 2017-08-03 Tom de Vries <tom@codesourcery.com>
3882 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
3883 ACCEL_COMPILER, apply finish_options on
3884 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
3886 2017-08-03 Tom de Vries <tom@codesourcery.com>
3889 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
3890 function_entry_patch_area_size > 0.
3892 2017-08-03 Jakub Jelinek <jakub@redhat.com>
3895 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
3896 instead of 10??LU, perform unit multiplication in wide_int,
3897 don't change alloc_object_size_limit if the limit is larger
3900 PR tree-optimization/81655
3901 PR tree-optimization/81588
3902 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
3903 the case when ranges[i].low and high are 1 for unsigned type with
3907 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
3908 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
3910 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3912 * tree-vrp.h: Add include guard.
3914 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
3917 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
3918 (ud2): New insn pattern.
3919 * config/i386/i386.c (ix86_expand_epilogue):
3920 For naked functions, generate ud2 instead of trap insn.
3922 2017-08-02 Marek Polacek <polacek@redhat.com>
3925 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
3927 2017-08-02 Tom de Vries <tom@codesourcery.com>
3928 Cesar Philippidis <cesar@codesourcery.com>
3930 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
3931 Add missing edge probabilities.
3933 2017-08-02 Tamar Christina <tamar.christina@arm.com>
3935 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
3938 2017-08-02 Jakub Jelinek <jakub@redhat.com>
3941 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
3942 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
3943 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
3944 prologue_seq sequences - if any.
3946 2017-08-02 Richard Biener <rguenther@suse.de>
3948 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
3949 via vectors if supported, integer extracts via punning if supported
3950 or otherwise vector extracts.
3952 2017-08-02 Richard Biener <rguenther@suse.de>
3954 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
3956 (bitmap_insert_into_set): ... this.
3958 2017-08-02 Richard Biener <rguenther@suse.de>
3960 PR tree-optimization/81633
3962 2015-08-17 Alan Hayward <alan.hayward@arm.com>
3964 PR tree-optimization/71752
3965 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
3967 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
3969 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
3970 (machine_function::call_ms2sysv_pad_out): Remove field.
3971 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
3972 (ix86_compute_frame_layout): Likewise.
3974 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
3977 * config/i386/i386.c (ix86_set_func_type): Disallow naked
3978 attribute with interrupt attribute.
3980 2017-08-01 Andrew Pinski <apinski@cavium.com>
3982 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
3983 BIT_INSERT_EXPR's operand 1
3984 to see if the types precision matches.
3986 2017-08-01 Martin Liska <mliska@suse.cz>
3989 * builtins.c (expand_builtin_memcpy_args): Remove.
3990 (expand_builtin_memcpy): Call newly added function
3991 expand_builtin_memory_copy_args.
3992 (expand_builtin_memcpy_with_bounds): Likewise.
3993 (expand_builtin_mempcpy): Remove last argument.
3994 (expand_builtin_mempcpy_with_bounds): Likewise.
3995 (expand_builtin_memory_copy_args): New function created from
3996 expand_builtin_mempcpy_args with small modifications.
3997 (expand_builtin_mempcpy_args): Remove.
3998 (expand_builtin_stpcpy): Remove unused argument.
3999 (expand_builtin): Likewise.
4000 (expand_builtin_with_bounds): Likewise.
4002 2017-08-01 Martin Liska <mliska@suse.cz>
4005 Make mempcpy more optimal (PR middle-end/70140).
4007 2017-08-01 Jakub Jelinek <jakub@redhat.com>
4010 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
4011 __builtin_vec_cmpne verify both arguments are compatible vectors
4012 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
4013 verify arg1_type is a pointer or array type. For __builtin_vec_st,
4014 move computation of aligned to after checking the argument types.
4018 * config/rs6000/vsx.md (vextract_fp_from_shorth,
4019 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
4022 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
4023 Doug Rupp <rupp@adacore.com>
4024 Olivier Hainque <hainque@adacore.com>
4026 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
4027 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
4029 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
4030 for TARGET_OS_CPP_BUILTIN.
4031 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
4032 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
4034 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
4035 passing required abi options to the assembler for EABI configurations.
4036 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
4037 of .text.hot and .text.unlikely sections for kernel modules when
4038 using ARM style exceptions.
4039 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
4040 options. Add EXTRA_CC1_SPEC.
4041 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
4043 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
4045 (ARM_TARGET2_DWARF_FORMAT): Define.
4046 * config/arm/t-vxworks: Adjust multilib control to removal of the
4047 Diab command line options.
4049 2017-08-01 Martin Liska <mliska@suse.cz>
4051 PR gcov-profile/81561
4052 * gcov.c (unblock): Make unblocking safe as we need to preserve
4053 index correspondence of blocks and block_lists.
4055 2017-08-01 Richard Biener <rguenther@suse.de>
4057 PR tree-optimization/81181
4058 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4059 (compute_antic): ... end of iteration here.
4061 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
4063 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
4064 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
4065 (ftree-slp-vectorize): Likewise.
4066 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
4067 can no longer be set independent of flag_tree_loop_vectorize.
4068 * omp-general.c (emp_max_vf): Likewise.
4069 * opts.c (enable_fdo_optimizations): Remove references to
4070 flag_tree_vectorize, these are now implicit.
4071 (common_handle_option): Remove handling for OPT_ftree_vectorize,
4072 and leave it for the options machinery.
4074 2017-08-01 Martin Liska <mliska@suse.cz>
4077 * builtins.c (expand_builtin_memcpy_args): Remove.
4078 (expand_builtin_memcpy): Call newly added function
4079 expand_builtin_memory_copy_args.
4080 (expand_builtin_memcpy_with_bounds): Likewise.
4081 (expand_builtin_mempcpy): Remove last argument.
4082 (expand_builtin_mempcpy_with_bounds): Likewise.
4083 (expand_builtin_memory_copy_args): New function created from
4084 expand_builtin_mempcpy_args with small modifications.
4085 (expand_builtin_mempcpy_args): Remove.
4086 (expand_builtin_stpcpy): Remove unused argument.
4087 (expand_builtin): Likewise.
4088 (expand_builtin_with_bounds): Likewise.
4090 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
4093 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
4094 print "ds:" only for immediates in generic address space.
4096 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
4099 * config/i386/i386.c (ix86_funciton_naked): New prototype.
4100 (ix86_function_ok_for_sibcall): Return false for naked functions.
4102 2017-08-01 Richard Biener <rguenther@suse.de>
4104 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
4105 (compute_antic): Seed worklist with exit block predecessors.
4106 * cfganal.c (dfs_find_deadend): For a cycle return the source
4107 of the edge closing it.
4109 2017-08-01 Tamar Christina <tamar.christina@arm.com>
4111 * config/aarch64/aarch64.c
4112 (aarch64_can_const_movi_rtx_p): Move 0 check.
4114 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4116 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
4117 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
4119 * match.pd: Ditto in address comparison pattern.
4121 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4123 PR tree-optimization/81627
4124 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
4125 closed ssa form for store-store chain.
4127 2017-08-01 Bin Cheng <bin.cheng@arm.com>
4129 PR tree-optimization/81620
4130 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
4131 for store-store chain.
4133 2017-08-01 Jakub Jelinek <jakub@redhat.com>
4135 PR tree-optimization/81588
4136 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
4137 ranges[i].in_p, invert comparison code ccode. For >/>=,
4138 swap rhs1 and rhs2 and comparison code unconditionally,
4139 for </<= don't do that. Don't swap rhs1/rhs2 again if
4140 ranges[i].in_p, instead invert comparison code ccode if
4141 opcode or oe->rank is BIT_IOR_EXPR.
4144 * optabs.def (vec_extract_optab, vec_init_optab): Change from
4145 a direct optab to conversion optab.
4146 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
4147 with GET_MODE_INNER as last argument instead of optab_handler.
4148 * expmed.c (extract_bit_field_1): Likewise. Use vector from
4149 vector extraction if possible and optab is available.
4150 * expr.c (store_constructor): Use convert_optab_handler instead
4151 of optab_handler. Use vector initialization from smaller
4152 vectors if possible and optab is available.
4153 * tree-vect-stmts.c (vectorizable_load): Likewise.
4154 * doc/md.texi (vec_extract, vec_init): Document that the optabs
4156 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
4157 of vec_init from half-sized vectors with the same element mode.
4158 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
4159 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
4160 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
4161 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
4162 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
4163 after mode in gen_vec_extract* calls.
4164 (vec_extract<mode>): Renamed to ...
4165 (vec_extract<mode><ssescalarmodelower>): ... this.
4166 (vec_extract<mode><ssehalfvecmodelower>): New expander.
4167 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
4168 element mode after mode in gen_vec_init* calls.
4169 (VEC_INIT_HALF_MODE): New mode iterator.
4170 (vec_init<mode>): Renamed to ...
4171 (vec_init<mode><ssescalarmodelower>): ... this.
4172 (vec_init<mode><ssehalfvecmodelower>): New expander.
4173 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
4174 (vec_extractv2sfsf): ... this.
4175 (vec_initv2sf): Renamed to ...
4176 (vec_initv2sfsf): ... this.
4177 (vec_extractv2si): Renamed to ...
4178 (vec_extractv2sisi): ... this.
4179 (vec_initv2si): Renamed to ...
4180 (vec_initv2sisi): ... this.
4181 (vec_extractv4hi): Renamed to ...
4182 (vec_extractv4hihi): ... this.
4183 (vec_initv4hi): Renamed to ...
4184 (vec_initv4hihi): ... this.
4185 (vec_extractv8qi): Renamed to ...
4186 (vec_extractv8qiqi): ... this.
4187 (vec_initv8qi): Renamed to ...
4188 (vec_initv8qiqi): ... this.
4189 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
4190 (vec_init<mode>): Renamed to ...
4191 (vec_init<mode><VEC_base_l>): ... this.
4192 (vec_extract<mode>): Renamed to ...
4193 (vec_extract<mode><VEC_base_l>): ... this.
4194 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
4195 (vec_initv2sfsf): ... this.
4196 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
4197 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
4198 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
4199 element mode after mode in gen_vec_init* calls.
4200 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
4201 (vec_init<mode><Vel>): ... this.
4202 (vec_extract<mode>): Renamed to ...
4203 (vec_extract<mode><Vel>): ... this.
4204 * config/aarch64/iterators.md (Vel): New mode attribute.
4205 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
4206 Add element mode after mode in gen_vec_extract* calls.
4207 * config/s390/vector.md (non_vec_l): New mode attribute.
4208 (vec_extract<mode>): Renamed to ...
4209 (vec_extract<mode><non_vec_l>): ... this.
4210 (vec_init<mode>): Renamed to ...
4211 (vec_init<mode><non_vec_l>): ... this.
4212 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
4213 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
4215 * config/arm/iterators.md (V_elem_l): New mode attribute.
4216 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
4217 (vec_extract<mode><V_elem_l>): ... this.
4218 (vec_extractv2di): Renamed to ...
4219 (vec_extractv2didi): ... this.
4220 (vec_init<mode>): Renamed to ...
4221 (vec_init<mode><V_elem_l>): ... this.
4222 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
4223 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
4224 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
4225 Add element mode after gen_vec_extract* calls.
4226 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
4227 (vec_init<mode><unitmode>): ... this.
4228 (vec_extract<mode>): Renamed to ...
4229 (vec_extract<mode><unitmode>): ... this.
4230 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
4231 (vec_init<mode><unitmode>): ... this.
4232 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
4233 (vec_initv2sfsf): ... this.
4234 (vec_extractv2sf): Renamed to ...
4235 (vec_extractv2sfsf): ... this.
4236 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
4237 Add element mode after gen_vec_extract* calls.
4238 * config/mips/mips.md (unitmode): New mode iterator.
4239 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
4240 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
4241 * config/spu/spu.md (inner_l): New mode attribute.
4242 (vec_init<mode>): Renamed to ...
4243 (vec_init<mode><inner_l>): ... this.
4244 (vec_extract<mode>): Renamed to ...
4245 (vec_extract<mode><inner_l>): ... this.
4246 * config/sparc/sparc.md (veltmode): New mode iterator.
4247 (vec_init<VMALL:mode>): Renamed to ...
4248 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
4249 * config/ia64/vect.md (vec_initv2si): Renamed to ...
4250 (vec_initv2sisi): ... this.
4251 (vec_initv2sf): Renamed to ...
4252 (vec_initv2sfsf): ... this.
4253 (vec_extractv2sf): Renamed to ...
4254 (vec_extractv2sfsf): ... this.
4255 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
4256 (vec_init<mode>): Renamed to ...
4257 (vec_init<mode><VEC_base_l>): ... this.
4258 (vec_extract<mode>): Renamed to ...
4259 (vec_extract<mode><VEC_base_l>): ... this.
4260 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
4261 (vec_initv2sfsf): ... this.
4262 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
4263 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
4264 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
4265 gen_vec_init* calls.
4267 2017-08-01 Richard Biener <rguenther@suse.de>
4269 PR tree-optimization/81297
4270 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
4271 TREE_OVERFLOW from INTEGER_CSTs.
4273 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
4275 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
4277 2017-07-31 Carl Love <cel@us.ibm.com>
4279 * config/rs6000/rs6000-c: Add support for built-in functions
4280 vector signed char vec_xl_be (signed long long, signed char *);
4281 vector unsigned char vec_xl_be (signed long long, unsigned char *);
4282 vector signed int vec_xl_be (signed long long, signed int *);
4283 vector unsigned int vec_xl_be (signed long long, unsigned int *);
4284 vector signed long long vec_xl_be (signed long long, signed long long *);
4285 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
4286 vector signed short vec_xl_be (signed long long, signed short *);
4287 vector unsigned short vec_xl_be (signed long long, unsigned short *);
4288 vector double vec_xl_be (signed long long, double *);
4289 vector float vec_xl_be (signed long long, float *);
4290 * config/rs6000/altivec.h (vec_xl_be): Add #define.
4291 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
4292 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
4294 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
4295 (altivec_expand_builtin): Add switch statement to call
4296 altivec_expand_xl_be for each builtin.
4297 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
4298 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
4299 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
4300 __builtin_vsx_le_be_v16qi.
4301 * doc/extend.texi: Update the built-in documentation file for the
4302 new built-in functions.
4304 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
4307 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
4309 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
4311 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4313 * config.gcc: Add z14.
4314 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
4315 CPU model numbers for z13s and z14.
4316 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
4318 * config/s390/s390-opts.h (enum processor_type): Rename
4319 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
4320 * config/s390/s390.c (processor_table): Add field for CPU name to
4321 be passed to Binutils.
4322 (s390_asm_output_machine_for_arch): Use the new field in
4323 processor_table for Binutils.
4324 (s390_expand_builtin): Replace arch12 with z14.
4325 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
4326 (s390_get_sched_attrmask): Likewise.
4327 (s390_get_unit_mask): Likewise.
4328 * config/s390/s390.opt: Add z14 to processor_type enum.
4330 2017-07-31 Martin Jambor <mjambor@suse.cz>
4333 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
4334 regardless of optimization level.
4336 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
4337 Martin Liska <mliska@suse.cz>
4339 * predict.def: Remove old comment and adjust probability.
4340 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
4343 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
4346 * config/i386/i386.c (ix86_function_naked): New function.
4347 (ix86_can_use_return_insn_p): Return false for naked functions.
4348 (ix86_expand_prologue): Skip prologue for naked functions.
4349 (ix86_expand_epilogue): Skip epilogue for naked functions
4350 and emit trap instruction.
4351 (ix86_warn_func_return): New function.
4352 (ix86_attribute_table): Add "naked" attribute specification.
4353 (TARGET_WARN_FUNC_RETURN): Define.
4354 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
4356 2017-07-31 Martin Liska <mliska@suse.cz>
4358 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
4359 (dump_gimple_bb_header): Always dump BB info.
4360 (pp_cfg_jump): Do not append info about BB when dumping a jump.
4362 2017-07-31 Martin Liska <mliska@suse.cz>
4365 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
4366 also with current_function_decl non-null equality.
4368 2017-07-31 Jakub Jelinek <jakub@redhat.com>
4371 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
4372 change type to the element type, instead add eltype variable and
4373 use it where we are interested in the element type.
4375 PR tree-optimization/81603
4376 * ipa-polymorphic-call.c
4377 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
4378 offset arithmetic in offset_int, bail out if the resulting bit offset
4379 doesn't fit into shwi.
4381 2017-07-31 Martin Liska <mliska@suse.cz>
4383 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
4384 (gimplify_save_expr): Fix comment.
4386 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
4389 * config/i386/i386.c (ix86_function_arg): Update arguments for
4391 (ix86_compute_frame_layout): Set the initial stack offset to
4392 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
4393 INCOMING_FRAME_SP_OFFSET.
4394 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
4395 stack before exception handler returns.
4396 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
4397 the 'ERROR_CODE' for exception handler.
4399 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
4401 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
4402 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
4403 (ASM_OUTPUT_REG_POP): Ditto.
4404 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
4405 instead of asm_fprintf to output pure string.
4407 2017-07-29 Jakub Jelinek <jakub@redhat.com>
4409 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
4410 to imported_module_or_decl hook.
4411 (debug_nothing_tree_tree_tree_bool): Remove.
4412 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
4413 * debug.c (do_nothing_debug_hooks): Use
4414 debug_nothing_tree_tree_tree_bool_bool instead of
4415 debug_nothing_tree_tree_tree_bool.
4416 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
4417 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
4418 * sdbout.c (sdb_debug_hooks): Likewise.
4419 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
4420 (gen_namespace_die): Add DW_AT_export_symbols attribute if
4422 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
4423 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
4424 attribute, don't add anything.
4426 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4428 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
4429 (fold_build2_stat_loc): Likewise.
4430 (fold_build3_stat_loc): Likewise.
4431 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
4432 (fold_build1_loc): Remove macro.
4433 (fold_build2_loc): Likewise.
4434 (fold_build3_loc): Likewise.
4436 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4438 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
4439 (gimple_build_debug_bind_source_stat): Likewise.
4440 * gimple.h (gimple_build_debug_bind): Remove macro.
4441 (gimple_build_debug_bind_source): Likewise.
4443 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4445 * bitmap.c (bitmap_alloc): Adjust.
4446 (bitmap_gc_alloc): Likewise.
4447 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
4449 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4451 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
4452 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
4453 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
4454 (bitmap_gc_alloc_stat): Likewise.
4455 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
4457 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4459 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
4460 * rtl.h (shallow_copy_rtx): Remove macro.
4462 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4464 * emit-rtl.c (gen_raw_REG): Adjust.
4465 * gengenrtl.c (gendef): Likewise.
4466 * rtl.c (rtx_alloc_stat): Remove _stat from name.
4467 * rtl.h (rtx_alloc): Remove macro.
4469 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4471 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
4472 (build_tree_list_stat): Likewise.
4473 * tree.h (build_tree_list): Remove macro.
4474 (build_tree_list_vec): Likewise.
4476 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4478 * tree.c (make_vector_stat): Remove _stat from name.
4479 (build_vector_stat): Likewise.
4480 * tree.h (make_vector_stat): Remove macro.
4481 (build_vector_stat): Likewise.
4483 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4485 * tree.h (build_var_debug_value): Remove prototype.
4487 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4489 * tree.c (tree_cons_stat): Remove _stat from name.
4490 * tree.h (tree_cons): Remove macro.
4492 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4494 * tree.c (build_vl_exp_stat): Remove _stat from name.
4495 * tree.h (build_vl_exp): Remove macro.
4497 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4499 * tree.c (build_decl_stat): Remove _stat from name.
4500 * tree.h (build_decl): Remove macro.
4502 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4504 * gimple.c (gimple_build_with_ops_stat): Adjust.
4505 (gimple_alloc_stat): Remove _stat from name.
4506 * gimple.h (gimple_alloc): Remove macro.
4508 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4510 * tree.c (make_tree_vec_stat): Remove _stat from name.
4511 (grow_tree_vec_stat): Likewise.
4512 * tree.h (make_tree_vec_stat): Adjust prototype.
4513 (grow_tree_vec_stat): Likewise.
4514 (make_tree_vec): Remove macro.
4515 (grow_tree_vec): Likewise.
4517 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4519 * fold-const.c (fold_build1_stat_loc): Adjust.
4520 (fold_build2_stat_loc): Likewise.
4521 (fold_build3_stat_loc): Likewise.
4522 * tree.c (build0_stat): Remove _stat from name.
4523 (build1_stat): Likewise.
4524 (build2_stat): Likewise.
4525 (build3_stat): Likewise.
4526 (build4_stat): Likewise.
4527 (build5_stat): Likewise.
4528 * tree.h (build1_loc): Remove macro, and rename _stat function
4530 (build2_loc): Likewise.
4531 (build3_loc): Likewise.
4532 (build4_loc): Likewise.
4533 (build5_loc): Likewise.
4535 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4537 * tree.c (make_int_cst_stat): Remove _stat from name.
4538 * tree.h (make_int_cst_stat): Adjust prototype.
4539 (make_int_cst): Remove macro.
4541 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4543 * tree.c (make_tre_binfo_stat): Remove _stat from name.
4544 * tree.h (make_tree_binfo_stat): Adjust prototype.
4545 (make_tree_binfo): Remove.
4547 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4549 * tree.c (copy_node_stat): Rename to copy_node.
4550 (build_distinct_type_copy): Adjust.
4551 * tree.h (copy_node_stat): Adjust prototype.
4552 (copy_node): Remove macro.
4554 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
4556 * tree.c (make_node_stat): rename to make_node.
4557 (build_tree_list_stat): Adjust.
4558 (build0_stat): Likewise.
4559 (build2_stat): Likewise.
4560 (build3_stat): Likewise.
4561 (build4_stat): Likewise.
4562 (build5_stat): Likewise.
4563 (build_decl_stat): Likewise.
4564 * tree.h (make_node_stat): Adjust prototype.
4565 (make_node): remove macro.
4567 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
4569 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
4570 (PPC_FEATURE2_SCV): Likewise.
4571 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
4573 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4575 * config/aarch64/aarch64.c
4576 (aarch64_internal_mov_immediate): Add new special pattern.
4577 * config/aarch64/aarch64.md (*movdi_aarch64):
4578 Add reg/32bit const mov case.
4580 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4581 Richard Sandiford <richard.sandiford@linaro.org>
4583 * config/aarch64/aarch64.md (mov<mode>): Generalize.
4584 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
4585 Add integer and movi cases.
4586 (movi-split-hf-df-sf split, fp16): New.
4587 (enabled): Added TARGET_FP_F16INST.
4588 * config/aarch64/iterators.md (GPF_HF): New.
4589 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
4591 2017-07-28 Tamar Christina <tamar.christina@arm.com>
4593 * config/aarch64/aarch64.c
4594 (aarch64_simd_container_mode): Add prototype.
4595 (aarch64_expand_mov_immediate): Add HI support.
4596 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
4597 (aarch64_can_const_movi_rtx_p): New.
4598 (aarch64_preferred_reload_class):
4599 Remove restrictions of using FP registers for certain SIMD operations.
4600 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
4601 (aarch64_valid_floating_const): Add integer move validation.
4602 (aarch64_simd_imm_scalar_p): Remove.
4603 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
4604 (aarch64_legitimate_constant_p): Expand list of supported cases.
4605 * config/aarch64/aarch64-protos.h
4606 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
4607 (aarch64_reinterpret_float_as_int): New.
4608 (aarch64_simd_imm_scalar_p): Remove.
4609 * config/aarch64/constraints.md (Uvi): New.
4610 (Dd): Split into Ds and new Dd.
4611 * config/aarch64/aarch64.md (*movsi_aarch64):
4613 (*movdi_aarch64): Add SIMD mov case.
4615 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4617 * tree-predcom.c: (struct chain): Handle store-store chain in which
4618 stores for elimination only store loop invariant values.
4619 (execute_pred_commoning_chain): Ditto.
4620 (prepare_initializers_chain_store_elim): Ditto.
4621 (prepare_finalizers): Ditto.
4622 (is_inv_store_elimination_chain): New function.
4623 (initialize_root_vars_store_elim_1): New function.
4625 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4627 * tree-predcom.c: Revise general description of the pass.
4628 (enum chain_type): New enum type for store elimination.
4629 (struct chain): New field supporting store elimination.
4630 (struct component): Ditto.
4631 (dump_chain): Dump store-stores chain.
4632 (release_chain): Release resources.
4633 (split_data_refs_to_components): Compute and create component
4634 contains only stores for elimination.
4635 (get_chain_last_ref_at): New function.
4636 (make_invariant_chain): Initialization.
4637 (make_rooted_chain): Specify chain type in parameter and record it.
4638 (add_looparound_copies): Skip for store-stores chain.
4639 (determine_roots_comp): Compute type of chain and pass it to
4641 (initialize_root_vars_store_elim_2): New function.
4642 (finalize_eliminated_stores): New function.
4643 (remove_stmt): Handle store for elimination.
4644 (execute_pred_commoning_chain): Execute predictive commoning on
4646 (determine_unroll_factor): Skip unroll for store-stores chain.
4647 (prepare_initializers_chain_store_elim): New function.
4648 (prepare_initializers_chain): Hanlde store-store chain.
4649 (prepare_finalizers_chain, prepare_finalizers): New function.
4650 (tree_predictive_commoning_loop): Return integer value indicating
4651 if loop is unrolled or lcssa form is corrupted.
4652 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
4654 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4656 * tree-predcom.c (initialize_root): Delete.
4657 (execute_pred_commoning_chain): Initialize root vars and replace
4658 reference of non-combined chain directly, rather than call above
4661 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4663 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
4664 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
4666 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4668 * tree-predcom.c (struct chain): New field init_seq.
4669 (release_chain): Release init_seq.
4670 (prepare_initializers_chain): Record intialization stmts in above
4672 (insert_init_seqs): New function.
4673 (tree_predictive_commoning_loop): Call insert_init_seqs.
4675 2017-07-28 Bin Cheng <bin.cheng@arm.com>
4677 * tree-predcom.c (determine_roots_comp): Skip trivial components.
4679 2017-07-28 Richard Biener <rguenther@suse.de>
4681 * match.pd: Remove superfluous :c.
4682 * genmatch.c (simplify::id): Add member.
4683 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
4685 (current_id): New global.
4686 (dt_node::parent): Move from ...
4687 (dt_operand::parent): ... here. Add for_id member.
4688 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
4689 (decision_tree::find_node): Relax order requirement when
4690 merging DT_TRUE nodes to ones inbetween the current simplify
4691 and the one we try to merge with. Add diagnostic whenever
4692 we need to enforce pattern order by not merging.
4693 (decision_tree::insert): Set current_id.
4694 (decision_tree::print_node): Dump parent node and for_id.
4695 (parser::last_id): Add member.
4696 (parser::push_simplify): Assign unique id.
4697 (parser::parser): Initialize last_id.
4699 2017-07-28 Martin Liska <mliska@suse.cz>
4702 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
4703 gimple_build_debug_bind.
4705 2017-07-28 Richard Biener <rguenther@suse.de>
4707 PR tree-optimization/81502
4708 * match.pd: Add pattern combining BIT_INSERT_EXPR with
4710 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
4712 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
4713 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
4714 for BIT_FIELD_REF args.
4715 * fold-const.c (make_bit_field_ref): Likewise.
4716 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
4718 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4721 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
4722 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
4723 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
4724 Or it into SANITIZER_UNDEFINED.
4725 * ubsan.c: Include gimple-fold.h and varasm.h.
4726 (ubsan_expand_ptr_ifn): New function.
4727 (instrument_pointer_overflow): New function.
4728 (maybe_instrument_pointer_overflow): New function.
4729 (instrument_object_size): Formatting fix.
4730 (pass_ubsan::execute): Call instrument_pointer_overflow
4731 and maybe_instrument_pointer_overflow.
4732 * internal-fn.c (expand_UBSAN_PTR): New function.
4733 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
4734 * sanitizer.def (__ubsan_handle_pointer_overflow,
4735 __ubsan_handle_pointer_overflow_abort): New builtins.
4736 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
4737 * internal-fn.def (UBSAN_PTR): New internal function.
4738 * opts.c (sanitizer_opts): Add pointer-overflow.
4739 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
4740 * fold-const.c (build_range_check): Compute pointer range check in
4741 integral type if pointer arithmetics would be needed. Formatting
4744 2017-07-28 Martin Liska <mliska@suse.cz>
4747 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
4748 parameters that are of a variable-length.
4750 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
4752 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
4754 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
4755 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
4756 (CRT_CALL_STATIC_FUNCTION): Likewise.
4757 (ASM_DEFAULT_SPEC): New define.
4758 (ASM_SPEC32): Likewise.
4759 (ASM_SPEC64): Likewise.
4760 (ASM_SPEC_COMMON): Likewise.
4761 (ASM_SPEC): Likewise.
4762 (INVALID_64BIT): Likewise.
4763 (LINK_OS_DEFAULT_SPEC): Likewise.
4764 (LINK_OS_SPEC32): Likewise.
4765 (LINK_OS_SPEC64): Likewise.
4766 (POWERPC_LINUX): Likewise.
4767 (PTRDIFF_TYPE): Likewise.
4768 (RESTORE_FP_PREFIX): Likewise.
4769 (RESTORE_FP_SUFFIX): Likewise.
4770 (SAVE_FP_PREFIX): Likewise.
4771 (SAVE_FP_SUFFIX): Likewise.
4772 (SIZE_TYPE): Likewise.
4773 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
4774 (TARGET_64BIT): Likewise.
4775 (TARGET_64BIT): Likewise.
4776 (TARGET_AIX): Likewise.
4777 (WCHAR_TYPE_SIZE): Likewise.
4778 (WCHAR_TYPE): Undefine.
4779 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
4780 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
4781 (CPP_OS_RTEMS_SPEC): Delete.
4782 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
4783 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
4785 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
4787 2017-07-28 Jakub Jelinek <jakub@redhat.com>
4789 PR tree-optimization/81578
4790 * tree-parloops.c (build_new_reduction): Bail out if
4791 reduction_code isn't one of the standard OpenMP reductions.
4792 Move the details printing after that decision.
4794 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4796 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
4797 related to reload_in_progress.
4798 (splat_input_operand): Likewise.
4799 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
4801 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
4803 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
4804 (TARGET_INSTANTIATE_DECLS): Likewise.
4805 (legitimate_indexed_address_p): Delete reload_in_progress code.
4806 (rs6000_debug_legitimate_address_p): Likewise.
4807 (rs6000_eliminate_indexed_memrefs): Likewise.
4808 (rs6000_emit_le_vsx_store): Likewise.
4809 (rs6000_emit_move_si_sf_subreg): Likewise.
4810 (rs6000_emit_move): Likewise.
4811 (register_to_reg_type): Likewise.
4812 (rs6000_pre_atomic_barrier): Likewise.
4813 (rs6000_machopic_legitimize_pic_address): Likewise.
4814 (rs6000_allocate_stack_temp): Likewise.
4815 (rs6000_address_for_fpconvert): Likewise.
4816 (rs6000_address_for_altivec): Likewise.
4817 (rs6000_secondary_memory_needed_rtx): Delete function.
4818 (rs6000_check_sdmode): Likewise.
4819 (rs6000_alloc_sdmode_stack_slot): Likewise.
4820 (rs6000_instantiate_decls): Likewise.
4821 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
4822 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
4823 Delete reload_in_progress.
4824 (*vec_reload_and_plus_<mptrsize>): Likewise.
4825 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
4826 (vsx_div_v2di): Likewise.
4827 (vsx_udiv_v2di): Likewise.
4829 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4831 * config/rs6000/rs6000.opt (mlra): Replace with stub.
4832 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
4833 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
4834 (rs6000_debug_reg_global): Delete print of LRA status.
4835 (rs6000_option_override_internal): Delete dead LRA related code.
4836 (rs6000_lra_p): Delete function.
4837 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
4839 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
4841 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
4842 * config/riscv/rtems.h: New file.
4844 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4845 Sudakshina Das <sudi.das@arm.com>
4847 * config/aarch64/aarch64.md
4848 (define_split for and<mode>3nr_compare): Move
4849 non aarch64_logical_operand to a register.
4850 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
4851 register immediate operand to a register.
4852 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
4854 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
4857 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
4859 2017-07-27 Richard Biener <rguenther@suse.de>
4861 PR tree-optimization/81573
4862 PR tree-optimization/81494
4863 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
4864 multi defuse cycle case.
4866 2017-07-27 Richard Biener <rguenther@suse.de>
4868 PR tree-optimization/81571
4869 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
4872 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
4874 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
4875 earlier and only if MASK_FPU is set. Adjust formatting.
4877 2017-07-27 Martin Liska <mliska@suse.cz>
4879 * opt-functions.awk: Add validation of value of Init.
4880 * optc-gen.awk: Pass new argument.
4882 2017-07-27 Martin Liska <mliska@suse.cz>
4884 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
4885 Fix wrong condition.
4887 2017-07-27 Martin Liska <mliska@suse.cz>
4889 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
4890 BBs and edges seen by autoFDO.
4892 2017-07-27 Richard Biener <rguenther@suse.de>
4894 PR tree-optimization/81502
4895 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
4896 with incompatible but same sized type.
4897 (execute_update_addresses_taken): Likewise.
4899 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
4901 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
4902 flag_tree_loop_vectorize rather than flag_tree_vectorize.
4904 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4907 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
4908 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
4909 Change s_operand to memory_operand.
4911 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
4913 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
4914 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
4915 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
4916 Emit instructions rather than returning an expression. Handle TFmode
4917 and KFmode by casting to TImode.
4918 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
4919 (rs6000_emit_le_vsx_store): Likewise.
4920 * config/rs6000/vsx.md (VSX_TI): New iterator.
4921 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
4922 (*vsx_le_undo_permute_<mode>): Likewise.
4923 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
4924 emit the split sequence.
4925 (*vsx_le_perm_store_<mode>): Likewise.
4927 2017-07-27 Jakub Jelinek <jakub@redhat.com>
4929 PR tree-optimization/81555
4930 PR tree-optimization/81556
4931 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
4932 if true, force CHANGED for the recursive invocation.
4933 (reassociate_bb): Remember original length of ops array, pass
4934 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
4936 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
4937 attributes for noipa attribute. For naked attribute use
4938 lookup_attribute first before lookup_attribute_spec.
4939 * final.c (rest_of_handle_final): Disable IPA RA for functions with
4941 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
4942 for functions with noipa attribute.
4943 (cgraph_externally_visible_p): Return true for functions with noipa
4945 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
4946 for functions with noipa attribute.
4947 * doc/extend.texi: Document noipa function attribute.
4948 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
4949 also for functions with noipa attribute.
4950 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
4952 2017-07-26 Andrew Pinski <apinski@cavium.com>
4954 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
4955 vec_unalign_load_cost and vec_unalign_store_cost.
4957 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4959 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
4960 -mvsx-small-integer option.
4961 (ISA_3_0_MASKS_IEEE): Likewise.
4962 (OTHER_VSX_VECTOR_MASKS): Likewise.
4963 (POWERPC_MASKS): Likewise.
4964 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
4965 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
4966 code, only testing for DImode being allowed in non-VSX floating
4968 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
4969 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
4971 (rs6000_option_override_internal): Delete -mvsx-small-integer.
4972 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
4973 TARGET_P8_VECTOR test.
4974 (rs6000_secondary_reload_simple_move): Likewise.
4975 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
4976 since TARGET_P9_VECTOR was already tested.
4977 (rs6000_opt_masks): Remove -mvsx-small-integer.
4978 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
4979 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
4981 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
4982 test for TARGET_VEXTRACTUB was used, and that uses
4984 (p9 extract splitter): Likewise.
4985 (vsx_extract_<mode>_di_p9): Likewise.
4986 (vsx_extract_<mode>_store_p9): Likewise.
4987 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
4988 for TARGET_P9_VECTOR was used. Delete code that is now dead with
4989 the elimination of TARGET_VSX_SMALL_INTEGER.
4990 (vsx_extract_<mode>_p8): Likewise.
4991 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
4992 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
4993 (vsx_set_<mode>_p9): Likewise.
4994 (vsx_set_v4sf_p9): Likewise.
4995 (vsx_set_v4sf_p9_zero): Likewise.
4996 (vsx_insert_extract_v4sf_p9): Likewise.
4997 (vsx_insert_extract_v4sf_p9_2): Likewise.
4998 * config/rs6000/rs6000.md (sign extend splitter): Change
4999 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
5000 (floatsi<mode>2_lfiwax_mem): Likewise.
5001 (floatunssi<mode>2_lfiwzx_mem): Likewise.
5002 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
5003 since a test for TARGET_P9_VECTOR was used.
5004 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
5005 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
5006 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
5007 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
5008 TARGET_P8_VECTOR test.
5009 (fix_trunc<mode>si2_stfiwx): Likewise.
5010 (fix_trunc<mode>si2_internal): Likewise.
5011 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
5012 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
5014 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
5015 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
5016 TARGET_P8_VECTOR test.
5017 (fixuns_trunc<mode>si2_stfiwx): Likewise.
5018 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
5019 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
5021 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
5022 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
5023 since a test for TARGET_P9_VECTOR was used.
5024 (splitter for loading small constants): Likewise.
5026 2017-07-26 Andrew Pinski <apinski@cavium.com>
5028 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
5031 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
5034 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
5035 (fp_valid_at): Likewise.
5037 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
5039 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
5040 (qdf24xx_addrcost_table): Likewise.
5041 (cortexa57_tunings): Update to use generic_branch_cost.
5042 (cortexa72_tunings): Likewise.
5043 (cortexa73_tunings): Likewise.
5044 (qdf24xx_tunings): Likewise.
5046 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
5048 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
5049 (thunderx2t99_branch_cost): Likewise.
5050 (cortexa35_tunings): Update to use generic_branch_cost.
5051 (cortexa53_tunings): Likewise.
5052 (cortexa57_tunings): Likewise.
5053 (cortexa72_tunings): Likewise.
5054 (cortexa73_tunings): Likewise.
5055 (thunderx2t99_tunings): Likewise.
5057 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
5059 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
5060 (sparc_option_override): Honour MASK_FSMULD.
5061 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
5062 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
5063 * config/sparc/sparc.opt (mfsmuld): New option.
5064 * doc/invoke.texi (mfsmuld): Document option.
5066 2017-07-26 Marek Polacek <polacek@redhat.com>
5069 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
5071 2017-07-26 Richard Biener <rguenther@suse.de>
5073 * gimple-match-head.c (do_valueize): Return OP if valueize
5075 (get_def): New helper to get at the def stmt of a SSA name
5077 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
5078 do_valueize to get at the def stmt.
5079 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
5081 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
5084 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
5086 2017-07-26 Martin Liska <mliska@suse.cz>
5089 * function.c (expand_function_start): Make expansion of
5090 nonlocal_goto_save_area after parm_birth_insn.
5092 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
5094 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
5095 from all CPU target flags enable members.
5097 2017-07-26 Richard Biener <rguenther@suse.de>
5099 * genmatch.c (dt_simplify::gen): Make iterator vars const.
5100 (decision_tree::gen): Make 'type' const.
5101 (write_predicate): Likewise.
5103 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
5105 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
5106 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
5107 (rs6000_option_override_internal): Likewise.
5108 (rs6000_expand_vector_set): Likewise.
5109 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
5110 (TARGET_UPPER_REGS_SF): Likewise.
5111 (TARGET_UPPER_REGS_DI): Likewise.
5112 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
5113 (TARGET_DIRECT_MOVE_64BIT): Likewise.
5114 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
5115 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
5116 (Splitters for DI constants in Altivec registers): Likewise.
5117 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
5118 (vsx_set_v4sf_p9): Likewise.
5119 (vsx_set_v4sf_p9_zero): Likewise.
5120 (vsx_insert_extract_v4sf_p9): Likewise.
5121 (vsx_insert_extract_v4sf_p9_2): Likewise.
5123 2017-07-25 Carl Love <cel@us.ibm.com>
5125 * doc/extend.texi: Update the built-in documentation file for the
5126 existing built-in functions
5127 vector signed char vec_cnttz (vector signed char);
5128 vector unsigned char vec_cnttz (vector unsigned char);
5129 vector signed short vec_cnttz (vector signed short);
5130 vector unsigned short vec_cnttz (vector unsigned short);
5131 vector signed int vec_cnttz (vector signed int);
5132 vector unsigned int vec_cnttz (vector unsigned int);
5133 vector signed long long vec_cnttz (vector signed long long);
5134 vector unsigned long long vec_cnttz (vector unsigned long long);
5136 2017-07-25 Andrew Pinski <apinski@cavium.com>
5138 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
5139 accesses where the use is for the first operand of a BIT_INSERT.
5141 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
5144 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
5145 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
5147 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
5149 * config/i386/gstabs.h: Delete.
5150 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
5152 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
5154 * config/i386/i386.c (ix86_decompose_address): Do not check for
5155 register RTX when looking at index_reg or base_reg.
5156 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
5158 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
5160 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
5161 to update EH info here.
5163 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5165 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
5167 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5169 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
5171 2017-07-25 Torsten Duwe <duwe@suse.de>
5173 * common.opt: Introduce -fpatchable-function-entry
5174 command line option, and its variables function_entry_patch_area_size
5175 and function_entry_patch_area_start.
5176 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
5177 including a two-value parser.
5178 * target.def (print_patchable_function_entry): New target hook.
5179 * targhooks.h (default_print_patchable_function_entry): New function.
5180 * targhooks.c (default_print_patchable_function_entry): Likewise.
5181 * toplev.c (process_options): Switch off IPA-RA if
5182 patchable function entries are being generated.
5183 * varasm.c (assemble_start_function): Look at the
5184 patchable-function-entry command line switch and current
5185 function attributes and maybe generate NOP instructions by
5186 calling the print_patchable_function_entry hook.
5187 * doc/extend.texi: Document patchable_function_entry attribute.
5188 * doc/invoke.texi: Document -fpatchable_function_entry
5189 command line option.
5190 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
5192 * doc/tm.texi: Re-generate.
5194 2017-07-25 Jakub Jelinek <jakub@redhat.com>
5197 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
5198 TARGET_AVX512DQ rather than TARGET_AVX512BW.
5200 2017-07-25 Tamar Christina <tamar.christina@arm.com>
5202 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
5204 2017-07-25 Richard Biener <rguenther@suse.de>
5206 PR tree-optimization/81455
5207 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
5208 not walk in cycles when looking for guards.
5210 2017-07-25 Richard Biener <rguenther@suse.de>
5212 PR tree-optimization/81529
5213 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
5214 when optimizing backedge uses.
5216 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
5218 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
5220 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
5221 to dl_section_ref. On AIX, append an expression to subtract
5222 the size of the section length to dl_section_ref.
5224 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
5226 * configure.ac: If any of the config.* scripts fail, exit 1.
5227 * configure: Regenerate.
5229 2017-07-25 Richard Biener <rguenther@suse.de>
5232 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
5233 of immediate uses, be more verbose on errors.
5235 2017-07-25 Richard Biener <rguenther@suse.de>
5237 PR tree-optimization/81510
5238 * tree-vect-loop.c (vect_is_simple_reduction): When the
5239 reduction stmt is not inside the loop bail out.
5241 2017-07-25 Richard Biener <rguenther@suse.de>
5243 PR tree-optimization/81303
5244 * tree-vect-loop-manip.c (vect_loop_versioning): Build
5245 profitability check against LOOP_VINFO_NITERSM1.
5247 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
5249 * domwalk.c (cmp_bb_postorder): Simplify.
5250 (sort_bbs_postorder): New function. Use it...
5251 (dom_walker::walk): ...here to optimize common cases.
5253 2017-07-25 Martin Liska <mliska@suse.cz>
5256 * ipa-visibility.c (function_and_variable_visibility): Make the
5257 redirection just on target that supports aliasing.
5258 Fix GNU coding style.
5260 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
5263 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
5265 * config/arm/rtems.h: Likewise.
5266 * config/bfin/rtems.h: Likewise.
5267 * config/i386/rtemself.h: Likewise.
5268 * config/lm32/rtems.h: Likewise.
5269 * config/m32c/rtems.h: Likewise.
5270 * config/m68k/rtemself.h: Likewise.
5271 * config/microblaze/rtems.h: Likewise.
5272 * config/mips/rtems.h: Likewise.
5273 * config/moxie/rtems.h: Likewise.
5274 * config/nios2/rtems.h: Likewise.
5275 * config/powerpcspe/rtems.h: Likewise.
5276 * config/rs6000/rtems.h: Likewise.
5277 * config/rtems.h: Likewise.
5278 * config/sh/rtems.h: Likewise.
5279 * config/sh/rtemself.h: Likewise.
5280 * config/sparc/rtemself.h: Likewise.
5282 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
5285 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
5286 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
5287 * tree-ssa-structalias.c (alias_get_name): Same.
5289 2017-07-25 Bin Cheng <bin.cheng@arm.com>
5292 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
5293 instructions if no du chain is found.
5295 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
5297 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
5299 2017-07-25 Richard Biener <rguenther@suse.de>
5302 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
5305 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
5307 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
5309 (ISA_2_7_MASKS_SERVER): Likewise.
5310 (ISA_3_0_MASKS_IEEE): Likewise.
5311 (OTHER_P8_VECTOR_MASKS): Likewise.
5312 (OTHER_VSX_VECTOR_MASKS): Likewise.
5313 (POWERPC_MASKS): Likewise.
5314 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
5315 duplicate list of options.
5316 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
5317 explicit -mupper-regs options.
5318 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
5319 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
5320 alias for -mupper-regs-df.
5321 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
5322 (rs6000_init_hard_regno_mode_ok): Likewise.
5323 (rs6000_option_override_internal): Likewise.
5324 (rs6000_opt_masks): Likewise.
5325 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
5326 options in terms of whether -mvsx or -mpower8-vector was used.
5327 (TARGET_UPPER_REGS_DI): Likewise.
5328 (TARGET_UPPER_REGS_SF): Likewise.
5329 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
5330 -mupper-regs-* options.
5332 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
5334 * passes.c (emergency_dump_function): Print some empty lines and a
5335 header before the RTL dump.
5337 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
5339 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
5341 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
5344 * config/aarch64/aarch64.c (aarch64_classify_symbol):
5345 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
5347 2017-07-24 Carl Love <cel@us.ibm.com>
5349 * config/rs6000/rs6000-c.c: Add support for built-in functions
5350 vector float vec_extract_fp32_from_shorth (vector unsigned short);
5351 vector float vec_extract_fp32_from_shortl (vector unsigned short);
5352 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
5353 vec_extract_fp_from_shortl): Add defines for the two builtins.
5354 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
5355 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
5357 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
5358 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
5359 * doc/extend.texi: Update the built-in documentation file for the
5360 new built-in function.
5362 2017-07-24 Jakub Jelinek <jakub@redhat.com>
5365 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
5367 * doc/generic.texi: Likewise.
5368 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
5369 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
5371 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
5373 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
5374 (aarch64_mls_elt_merge<mode>): Likewise.
5376 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
5378 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
5379 having __cxa_atexit.
5381 2017-07-23 Michael Collison <michael.collison@arm.com>
5383 * config/arm/arm.c (arm_option_override): Deprecate
5384 use of -mstructure-size-boundary.
5385 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
5386 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
5388 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5391 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
5392 Reduce cost estimate for direct moves.
5394 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
5397 * config/i386/i386.c (ix86_option_override_internal): Disable
5398 BMI, BMI2 and TBM instructions for -m16.
5400 2017-07-21 Carl Love <cel@us.ibm.com>
5402 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5403 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5404 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5405 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5406 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5407 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5408 VMULOSW): New enum "unspec" values.
5409 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5410 altivec_vmulosw): New patterns.
5411 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5412 VMULOSW): Add definitions.
5414 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
5416 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
5417 (qdf24xx): Likewise.
5418 * config/aarch64/aarch64-options-extensions.def (rdma); New.
5419 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
5420 (AARCH64_FL_V8_1): Renumber.
5421 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
5422 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
5423 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
5424 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
5425 rdma to feature modifiers list.
5427 2017-07-21 Yury Gribov <tetra2005@gmail.com>
5430 * ipa-visibility (function_and_variable_visibility): Convert
5431 recursive PLT call to direct call if appropriate.
5433 2017-07-21 Andrew Pinski <apinski@cavium.com>
5435 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
5436 operand 1 to see if the types precision matches.
5437 * fold-const.c (operand_equal_p): Likewise.
5439 2017-07-21 Richard Biener <rguenther@suse.de>
5441 PR tree-optimization/81303
5442 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
5443 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
5444 (vect_peeling_hash_get_lowest_cost): Adjust.
5445 (vect_enhance_data_refs_alignment): Likewise. Use
5446 vect_get_peeling_costs_all_drs to compute the penalty for no
5447 peeling to match up costs.
5449 2017-07-21 Richard Biener <rguenther@suse.de>
5451 PR tree-optimization/81500
5452 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
5453 we didn't identify a reduction path.
5455 2017-07-21 Tom de Vries <tom@codesourcery.com>
5456 Cesar Philippidis <cesar@codesourcery.com>
5458 PR gcov-profile/81442
5459 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
5462 2017-07-21 Tom de Vries <tom@codesourcery.com>
5465 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
5467 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
5468 nvptx_override_options_after_change.
5470 2017-07-21 Ulrich Drepper <drepper@redhat.com>
5472 * dwarf2out.c (output_file_names): Avoid double testing for
5475 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
5477 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
5479 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5481 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
5483 (try_crossjump_to_edge): Do not punt on partitioned functions.
5485 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
5487 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
5488 Put all BBs reachable only via paths crossing cold region to cold
5490 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
5492 2016-07-21 Richard Biener <rguenther@suse.de>
5494 PR tree-optimization/81303
5495 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
5496 into account prologue and epilogue iterations when raising
5497 min_profitable_iters to sth at least covering one vector iteration.
5499 2017-07-21 Tamar Christina <tamar.christina@arm.com>
5501 * config/arm/arm.c (arm_test_cpu_arch_dat):
5504 2017-07-20 Nathan Sidwell <nathan@acm.org>
5506 Remove TYPE_METHODS.
5507 * tree.h (TYPE_METHODS): Delete.
5508 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
5509 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
5510 (dbxout_type_methods): Scan TYPE_FIELDS.
5511 (dbxout_type): Don't check TYPE_METHODS here.
5512 * function.c (use_register_for_decl): Always ignore register for
5513 class types when not optimizing.
5514 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
5515 * tree.c (free_lang_data_in_type): Stitch out member functions and
5516 templates from TYPE_FIELDS.
5517 (build_distinct_type_copy, verify_type_variant,
5518 verify_type): Member fns are on TYPE_FIELDS.
5519 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
5520 * tree-pretty-print.c (dump_generic_node): Likewise.
5522 2017-07-20 Jakub Jelinek <jakub@redhat.com>
5525 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
5526 V2TImode and V4TImode.
5527 (ix86_expand_vector_extract): Likewise.
5528 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
5529 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
5530 (ssescalarmode): Handle V4TImode and V2TImode.
5531 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
5532 (*vec_extractv2ti, *vec_extractv4ti): New insns.
5533 (VEXTRACTI128_MODE): New mode iterator.
5534 (splitter for *vec_extractv?ti first element): New.
5535 (VEC_INIT_MODE): New mode iterator.
5536 (vec_init<mode>): Consolidate 3 expanders into one using
5537 VEC_INIT_MODE mode iterator.
5539 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
5541 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
5542 non_spilled_static_chain_regno_p.
5544 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
5546 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
5548 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
5550 * bb-reorder.c (connect_traces): Allow copying of blocks within
5553 2017-07-20 Richard Biener <rguenther@suse.de>
5555 * gimple.h (gimple_phi_result): Add gphi * overload.
5556 (gimple_phi_result_ptr): Likewise.
5557 (gimple_phi_arg): Likewise. Adjust index assert to only
5558 allow actual argument accesses rather than all slots available
5560 (gimple_phi_arg_def): Add gphi * overload.
5561 * tree-phinodes.c (make_phi_node): Initialize only actual
5563 (resize_phi_node): Clear memory not covered by old node,
5564 do not initialize excess argument slots.
5565 (reserve_phi_args_for_new_edge): Initialize new argument slot
5568 2017-07-20 Bin Cheng <bin.cheng@arm.com>
5570 PR tree-optimization/81388
5572 2016-07-21 Bin Cheng <bin.cheng@arm.com>
5574 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
5575 by removing computation of may_be_zero.
5577 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5578 Tom de Vries <tom@codesourcery.com>
5581 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
5582 when gimple level profile disagrees with what RTL expander did.
5584 2017-07-20 Richard Biener <rguenther@suse.de>
5586 PR tree-optimization/61171
5587 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
5588 (vect_analyze_stmt): Add slp instance parameter.
5589 (vectorizable_reduction): Likewise.
5590 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
5591 (vect_is_simple_reduction): Deal with chains not detected
5592 as SLP reduction chain, specifically not properly associated
5593 chains containing a mix of plus/minus.
5594 (get_reduction_op): Remove.
5595 (get_initial_defs_for_reduction): Simplify, pass in whether
5596 this is a reduction chain, pass in the SLP node for the PHIs.
5597 (vect_create_epilog_for_reduction): Get the SLP instance as
5599 (vectorizable_reduction): Get the SLP instance as arg.
5600 During analysis remember the SLP node with the PHIs in the
5601 instance. Simplify getting at the vectorized reduction PHIs.
5602 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
5603 through SLP instance.
5604 (vect_slp_analyze_operations): Likewise.
5605 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
5606 (vect_transform_stmt): Likewise.
5608 2017-07-20 Tom de Vries <tom@codesourcery.com>
5610 PR tree-optimization/81489
5611 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
5612 read of phi arg location to before loop that modifies phi.
5614 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
5616 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
5619 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5622 * except.c (execute): Fix ordering issue.
5624 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
5626 PR rtl-optimization/81423
5627 * combine.c (make_compound_operation_int): Don't try to optimize
5628 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
5630 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
5632 PR rtl-optimization/81423
5633 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
5634 with a constant that is -1 in the truncated to mode.
5636 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5638 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
5639 (determine_unlikely_bbs): ... here.
5640 * predict.h (propagate_unlikely_bbs_forward): Declare.
5641 * cfgexpand.c (pass_expand::execute): Use it.
5642 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
5644 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
5645 propagate_unlikely_bbs_forward.
5647 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
5650 * except.c (maybe_add_nop_after_section_switch): New function.
5653 2017-07-19 Tom de Vries <tom@codesourcery.com>
5655 * gimple.h (gimple_phi_set_arg): Make assert more strict.
5657 2017-07-19 Tom de Vries <tom@codesourcery.com>
5659 * gimple.h (gimple_phi_arg): Make assert more strict.
5661 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
5663 * config.gcc (powerpc*-*-*): Add mmintrin.h.
5664 * config/rs6000/mmintrin.h: New file.
5665 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
5667 2017-07-19 Jakub Jelinek <jakub@redhat.com>
5669 PR tree-optimization/81346
5670 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
5672 2017-07-19 Tom de Vries <tom@codesourcery.com>
5674 * config/nvptx/nvptx.md (VECIM): Add V2DI.
5676 2017-07-19 Tom de Vries <tom@codesourcery.com>
5678 * config/nvptx/nvptx-modes.def: Add V2DImode.
5679 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
5680 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
5681 (nvptx_output_mov_insn): Handle lack of mov.b128.
5682 (nvptx_print_operand): Handle 'H' and 'L' codes.
5683 (nvptx_vector_mode_supported): Allow V2DImode.
5684 (nvptx_preferred_simd_mode): New function.
5685 (nvptx_data_alignment): New function.
5686 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
5687 nvptx_preferred_simd_mode.
5688 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
5690 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
5692 2017-07-19 Tom de Vries <tom@codesourcery.com>
5694 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
5695 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
5696 (nvptx_vector_mode_supported): New function. Allow V2SImode.
5697 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
5698 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
5699 (mov<VECIM>_insn): New define_insn.
5700 (define_expand "mov<VECIM>): New define_expand.
5702 2017-07-19 Tom de Vries <tom@codesourcery.com>
5704 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
5706 2017-07-19 Jakub Jelinek <jakub@redhat.com>
5708 PR tree-optimization/81346
5709 * fold-const.h (fold_div_compare, range_check_type): Declare.
5710 * fold-const.c (range_check_type): New function.
5711 (build_range_check): Use range_check_type.
5712 (fold_div_compare): No longer static, rewritten into
5713 a match.pd helper function.
5714 (fold_comparison): Don't call fold_div_compare here.
5715 * match.pd (X / C1 op C2): New optimization using fold_div_compare
5718 2017-07-19 Nathan Sidwell <nathan@acm.org>
5720 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
5721 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
5722 * tree.c (find_decls_types_r, verify_type): Use
5723 TYPE_{MIN,MAX}_VALUE_RAW.
5724 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
5725 (hash_tree): Likewise.
5726 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
5728 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
5731 2017-07-18 Tom de Vries <tom@codesourcery.com>
5734 * omp-expand.c (expand_omp_for_static_chunk): Handle
5735 equal-argument loop exit phi.
5737 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
5740 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
5741 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
5742 operand 2 predicate.
5743 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
5744 operand 2 predicate.
5745 (ror,rol -> rorx splitters): Use const_int_operand as
5746 operand 2 predicate.
5748 2017-06-18 Richard Biener <rguenther@suse.de>
5750 PR tree-optimization/81410
5751 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
5752 the gap in the ! slp_perm SLP case after each group.
5754 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5757 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
5760 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
5763 * predict.c (set_even_probabilities): Cleanup; do not affect
5764 probabilities that are already known.
5765 (combine_predictions_for_bb): Call even when count is set.
5767 2017-07-18 Nathan Sidwell <nathan@acm.org>
5769 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
5772 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5775 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
5776 optimization for loop niter analysis.
5778 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
5781 * config/avr/avr.c (avr_optimize_casesi): Don't use
5782 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
5784 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
5786 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
5787 body_cost_vec from _vect_peel_extended_info.
5788 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
5789 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
5792 2017-07-18 Bin Cheng <bin.cheng@arm.com>
5794 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
5796 2017-07-18 Richard Biener <rguenther@suse.de>
5798 PR tree-optimization/80620
5799 PR tree-optimization/81403
5800 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
5801 info when re-using a VN table entry.
5803 2017-07-18 Richard Biener <rguenther@suse.de>
5805 PR tree-optimization/81418
5806 * tree-vect-loop.c (vectorizable_reduction): Properly compute
5807 vectype_in. Verify that with lane-reducing reduction operations
5808 we have a single def-use cycle.
5810 2017-07-17 Carl Love <cel@us.ibm.com>
5812 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
5814 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
5815 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
5816 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
5817 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
5818 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
5819 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
5820 VMULOSW): New enum "unspec" values.
5821 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
5822 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
5823 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
5824 altivec_vmulosw): New patterns.
5825 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
5826 VMULOSW): Add definitions.
5827 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
5829 * config/alpha/alpha.c: Include predict.h.
5831 2017-07-17 Yury Gribov <tetra2005@gmail.com>
5833 * tree-vrp.c (compare_assert_loc): Fix comparison function
5834 to return predictable results.
5836 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5838 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
5841 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
5842 * doc/invoke.texi (mexpand-adddi): Update text.
5844 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5846 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
5847 that also clobbers the CC register. The old expand code is moved
5849 (*arc_clzsi2): ... here.
5850 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
5851 the CC register. The old expand code is moved to ...
5852 (arc_ctzsi2): ... here.
5854 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5856 * config/arc/arc.opt (mindexed-loads): Use initial value
5857 TARGET_INDEXED_LOADS_DEFAULT.
5858 (mauto-modify-reg): Use initial value
5859 TARGET_AUTO_MODIFY_REG_DEFAULT.
5860 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5861 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5862 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
5863 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
5865 2017-07-17 Martin Liska <mliska@suse.cz>
5868 * opts.c (finish_options): Do not allow -fgnu-tm
5869 w/ -fsanitize={kernel-,}address. Say sorry.
5871 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5874 * tree-loop-distribution.c (classify_partition): Only assert on
5875 numer of iterations.
5876 (merge_dep_scc_partitions): Delete prameter. Update function call.
5877 (distribute_loop): Remove code handling loop with unknown niters.
5878 (pass_loop_distribution::execute): Skip loop with unknown niters.
5880 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5883 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
5884 function sort_partitions_by_post_order.
5886 2017-07-17 Bin Cheng <bin.cheng@arm.com>
5888 PR tree-optimization/81374
5889 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
5890 the max index of basic blocks, rather than number of basic blocks.
5892 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5894 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
5896 (arc_legitimate_pic_operand_p): Likewise.
5897 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
5899 (arc_needs_pcl_p): Likewise.
5900 (arc_legitimate_pc_offset_p): Likewise.
5901 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
5902 function is also used in constrains.md.
5903 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
5904 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
5905 PLUS. Only return true/false in known cases, otherwise assert.
5906 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
5907 is already called in arc_legitimate_constant_p.
5908 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
5910 (LEGITIMATE_PIC_OPERAND_P): Use
5911 arc_raw_symbolic_reference_mentioned_p function.
5912 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
5917 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5918 Andrew Burgess <andrew.burgess@embecosm.com>
5920 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
5921 (arc_return_address_register): New function.
5922 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
5923 (arc_handle_fndecl_attribute): Add naked attribute.
5924 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
5925 (TARGET_WARN_FUNC_RETURN): Likewise.
5926 (arc_allocate_stack_slots_for_args): New function.
5927 (arc_warn_func_return): Likewise.
5928 (machine_function): Change type fn_type.
5929 (arc_compute_function_type): Consider new naked function type,
5930 change function return type.
5931 (arc_must_save_register): Adapt to handle new
5932 arc_compute_function_type's return type.
5933 (arc_expand_prologue): Likewise.
5934 (arc_expand_epilogue): Likewise.
5935 (arc_return_address_regs): Delete.
5936 (arc_return_address_register): New function.
5937 (arc_epilogue_uses): Use above function.
5938 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
5939 (arc_function_type): Change encoding, add naked type.
5940 (ARC_INTERRUPT_P): Change to handle the new encoding.
5941 (ARC_FAST_INTERRUPT_P): Likewise.
5942 (ARC_NORMAL_P): Define.
5943 (ARC_NAKED_P): Likewise.
5944 (arc_compute_function_type): Delete prototype.
5945 * config/arc/arc.md (in_ret_delay_slot): Use
5946 arc_return_address_register function.
5947 (simple_return): Likewise.
5948 (p_return_i): Likewise.
5950 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5952 PR tree-optimization/81428
5953 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
5954 can't be built for those types.
5956 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
5958 Remove stuff dead since r239246.
5960 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
5961 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
5962 (avr_inform_devices): Remove dead stuff.
5964 2017-07-17 Tamar Christina <tamar.christina@arm.com>
5966 * config/arm/arm_neon.h: Fix softp typo.
5968 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5970 PR tree-optimization/81365
5971 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
5972 aggregate moves onto bb predecessor edges, make sure there are no
5973 loads that could alias the lhs in between the start of bb and the
5976 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
5979 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
5980 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
5981 [LSHIFTRT, outer_code = TRUNCATE]: Same.
5983 2017-07-17 Jakub Jelinek <jakub@redhat.com>
5985 PR tree-optimization/81396
5986 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
5987 (init_symbolic_number): Initialize it to 1.
5988 (perform_symbolic_merge): Add n_ops from both operands into the new
5990 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
5991 without base_addr as useless if they need more than one operation.
5992 (bswap_replace): Handle !bswap case for NULL base_addr.
5994 2017-07-17 Tom de Vries <tom@codesourcery.com>
5997 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
6000 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
6002 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
6003 conditional builtin define __FIX_LEON3FT_B2BST.
6005 2017-07-17 Daniel Cederman <cederman@gaisler.com>
6007 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
6008 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
6011 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
6013 PR rtl-optimization/81424
6014 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
6015 to remove potential trapping from operands if -fnon-call-exceptions.
6017 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6019 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
6020 profile_proability for scalling.
6021 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
6023 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6025 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
6027 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6029 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
6030 fixpoint arithmetics.
6032 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6034 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
6035 fixpoint arithmetics.
6037 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6039 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
6040 fixpoint arithmetics.
6042 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
6044 * profile-count.h (profile_probability::from_reg_br_prob_note,
6045 profile_probability::to_reg_br_prob_note): New functions.
6046 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
6047 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
6048 * predict.c (probability_reliable_p): Update.
6049 (edge_probability_reliable_p): Update.
6050 (br_prob_note_reliable_p): Update.
6051 (invert_br_probabilities): Update.
6052 (add_reg_br_prob_note): New function.
6053 (combine_predictions_for_insn): Update.
6054 * asan.c (asan_clear_shadow): Update.
6055 * cfgbuild.c (compute_outgoing_frequencies): Update.
6056 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
6057 (update_br_prob_note): Update.
6058 (rtl_verify_edges): Update.
6059 (purge_dead_edges): Update.
6060 (fixup_reorder_chain): Update.
6061 * emit-rtl.c (try_split): Update.
6062 * ifcvt.c (cond_exec_process_insns): Update.
6063 (cond_exec_process_if_block): Update.
6064 (dead_or_predicable): Update.
6065 * internal-fn.c (expand_addsub_overflow): Update.
6066 (expand_neg_overflow): Update.
6067 (expand_mul_overflow): Update.
6068 * loop-doloop.c (doloop_modify): Update.
6069 * loop-unroll.c (compare_and_jump_seq): Update.
6070 * optabs.c (emit_cmp_and_jump_insn_1): Update.
6071 * predict.h: Update.
6072 * reorg.c (mostly_true_jump): Update.
6074 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
6075 * config/alpha/alpha.c (emit_unlikely_jump): Update.
6076 * config/arc/arc.c: (emit_unlikely_jump): Update.
6077 * config/arm/arm.c: (emit_unlikely_jump): Update.
6078 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
6079 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
6080 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
6081 (ix86_print_operand): Update.
6082 (ix86_split_fp_branch): Update.
6083 (predict_jump): Update.
6084 * config/ia64/ia64.c (ia64_print_operand): Update.
6085 * config/mmix/mmix.c (mmix_print_operand): Update.
6086 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
6087 (rs6000_expand_split_stack_prologue): Update.
6088 * config/rs6000/rs6000.c: Update.
6089 * config/s390/s390.c (s390_expand_vec_strlen): Update.
6090 (s390_expand_vec_movstr): Update.
6091 (s390_expand_cs_tdsi): Update.
6092 (s390_expand_split_stack_prologue): Update.
6093 * config/sh/sh.c (sh_print_operand): Update.
6094 (expand_cbranchsi4): Update.
6095 (expand_cbranchdi4): Update.
6096 * config/sparc/sparc.c (output_v9branch): Update.
6097 * config/spu/spu.c (get_branch_target): Update.
6098 (ea_load_store_inline): Update.
6099 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
6100 * config/tilepro/tilepro.c: Update.
6102 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
6104 * gimplify.c (mostly_copy_tree_r): Revert latest change.
6105 (gimplify_save_expr): Likewise.
6107 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6109 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
6111 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6113 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
6115 * timevar.def (TV_IPA_FNSUMMARY): Define.
6117 2017-07-16 Daniel Cederman <cederman@gaisler.com>
6119 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
6120 to back store errata sensitive sequence from being generated.
6121 (sqrtdf2_fix): Likewise.
6123 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6125 * tree-ssa-threadupdate.c (compute_path_counts,
6126 update_joiner_offpath_counts): Use profile_probability.
6128 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
6131 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6133 * config/arm/arm-c.c (arm_cpu_builtins): Define
6134 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
6136 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
6138 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6139 array entries to represent __ieee128 versions of the
6140 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
6141 scalar_extract_sig, and scalar_insert_exp built-in functions.
6142 (altivec_resolve_overloaded_builtin): Add special case handling
6143 for the __builtin_scalar_insert_exp function, as represented by
6144 the P9V_BUILTIN_VEC_VSIEDP constant.
6145 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
6146 exponent support for __ieee128 argument.
6147 (VSESQP): Add scalar extract signature support for __ieee128
6149 (VSTDCNQP): Add scalar test negative support for __ieee128
6151 (VSIEQP): Add scalar insert exponent support for __int128 argument
6152 with __ieee128 result.
6153 (VSIEQPF): Add scalar insert exponent support for __ieee128
6154 argument with __ieee128 result.
6155 (VSTDCQP): Add scalar test data class support for __ieee128
6157 (VSTDCNQP): Add overload support for scalar test negative with
6159 (VSTDCQP): Add overload support for scalar test data class
6161 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
6163 (UNSPEC_VSX_SIEXPQP): New constant.
6164 (xsxexpqp): New insn for VSX scalar extract exponent quad
6166 (xsxsigqp): New insn for VSX scalar extract significand quad
6168 (xsiexpqpf): New insn for VSX scalar insert exponent quad
6169 precision with floating point argument.
6170 (xststdcqp): New expand for VSX scalar test data class quad
6172 (xststdcnegqp): New expand for VSX scalar test negative quad
6174 (xststdcqp): New insn to match expansions for VSX scalar test data
6175 class quad precision and VSX scalar test negative quad precision.
6176 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
6177 special case operand checking to enforce that second operand of
6178 VSX scalar test data class with quad precision argument is a 7-bit
6180 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
6181 prototypes and descriptions of __ieee128 versions of
6182 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
6183 scalar_test_data_class, and scalar_test_neg built-in functions.
6185 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6187 PR tree-optimization/81162
6188 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
6189 replace a negate with an add.
6191 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
6193 * doc/invoke.texi (arm/-mcpu): Document +crypto.
6195 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6197 * config/arm/arm-c.c (arm_cpu_builtins): Define
6198 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
6200 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6202 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
6203 (armv8-r): Set ARM Cortex-R52 as default CPU.
6204 * config/arm/arm-tables.opt: Regenerate.
6205 * config/arm/arm-tune.md: Regenerate.
6206 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
6208 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
6209 extension for -mcpu=cortex-r52.
6211 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
6213 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
6214 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
6215 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
6216 (fp-armv8): Define it as FP_ARMv8 only.
6217 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
6218 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
6220 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
6221 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
6222 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
6223 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
6224 than TARGET_FPU_ARMV8.
6225 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
6226 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
6227 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
6229 * config/arm/neon.md (neon_vrint): Likewise.
6230 (neon_vcvt): Likewise.
6231 (neon_<fmaxmin_op><mode>): Likewise.
6232 (<fmaxmin><mode>3): Likewise.
6233 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
6234 * config/arm/predicates.md (arm_cond_move_operator): Check against
6235 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
6237 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
6239 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
6242 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6244 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
6245 loop in comment with memset.
6247 2017-07-14 Martin Liska <mliska@suse.cz>
6249 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
6250 * dwarf2out.c (is_java): Remove the function.
6251 (output_pubname): Remove usage of the function.
6252 (lower_bound_default): Remove usage of DW_LANG_Java.
6253 (gen_compile_unit_die): Likewise.
6254 * gcc.c: Remove compiler defaults for .java and .zip files.
6255 * gimple-expr.c (remove_suffix): Change as there's no longer
6256 extension than 4-letter one.
6257 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
6258 (gimplify_save_expr): Likewise.
6259 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
6260 as it's possible even for other languages than Java.
6261 * langhooks.h (struct lang_hooks): Remove Java from a comment.
6262 * lto-opts.c (lto_write_options): Remove reference to Java.
6263 * opts.c (strip_off_ending): Update file extension handling.
6264 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
6265 * tree-eh.c (lower_resx): Likewise.
6266 * tree.c (free_lang_data_in_type): Remove dead code.
6267 (find_decls_types_r): Likewise.
6268 (build_common_builtin_nodes): Remove Java from a comment.
6269 (verify_type): Remove dead code.
6270 * varasm.c (assemble_external): Remove Java from a comment.
6272 2017-07-14 Martin Liska <mliska@suse.cz>
6274 * opts.c (finish_options): Add quotes.
6275 (common_handle_option): Likewise.
6277 2017-07-14 Martin Liska <mliska@suse.cz>
6279 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
6280 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
6282 * dwarf2out.c (lower_bound_default): Do not handle
6284 (gen_compile_unit_die): Likewise.
6285 * gcc.c: Remove default extension binding for GNU Pascal.
6286 * stmt.c: Remove Pascal language from a comment.
6287 * xcoffout.c: Likewise.
6289 2017-07-13 David Malcolm <dmalcolm@redhat.com>
6292 * diagnostic-show-locus.c (fixit_cmp): New function.
6293 (layout::layout): Sort m_fixit_hints.
6294 (column_range::column_range): Assert that the values are valid.
6295 (struct char_span): New struct.
6296 (correction::overwrite): New method.
6297 (struct source_line): New struct.
6298 (line_corrections::add_hint): Add assertions. Reimplement memcpy
6299 calls in terms of classes source_line and char_span, and
6300 correction::overwrite.
6301 (selftest::test_overlapped_fixit_printing_2): New function.
6302 (selftest::diagnostic_show_locus_c_tests): Call it.
6304 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
6306 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
6307 early if there is no lhs.
6309 2017-07-13 Martin Liska <mliska@suse.cz>
6311 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
6312 (gen_reference_type_die): Likewise.
6313 * stor-layout.c: Remove Pascal-related comment.
6315 2017-07-13 Martin Liska <mliska@suse.cz>
6317 * opts.c (finish_options): Add quotes to error messages.
6318 (parse_sanitizer_options): Likewise.
6320 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
6322 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
6324 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
6326 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
6328 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
6330 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
6332 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
6334 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
6337 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
6338 provides the hardware capability bits, define the macro
6339 __BUILTIN_CPU_SUPPORTS__.
6340 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
6341 if GLIBC does not provide the hardware capability bits. Add a
6342 gcc_unreachable call if the built-in cpu function is neither
6343 __builtin_cpu_is nor __builtin_cpu_supports.
6344 (rs6000_get_function_versions_dispatcher): Change the warning
6345 that an old GLIBC is used which does not export the capability
6346 bits to be an error.
6347 * doc/extend.texi (target_clones attribute): Document the
6348 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
6349 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
6350 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
6351 the macros defined by GCC if the newer GLIBC is available.
6353 2017-07-12 Jeff Law <law@redhat.com>
6355 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
6356 remaining includes slightly.
6357 * config/riscv/riscv-builtins.c: Include profile-count.h.
6359 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
6362 * config/avr/avr.c (avr_set_current_function): In diagnostic
6363 messages: Quote keywords and (parts of) identifiers.
6364 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
6367 2017-07-12 Carl Love <cel@us.ibm.com>
6369 * config/rs6000/rs6000-c.c: Add support for built-in functions
6370 vector bool char vec_revb (vector bool char);
6371 vector bool short vec_revb (vector short char);
6372 vector bool int vec_revb (vector bool int);
6373 vector bool long long vec_revb (vector bool long long);
6374 * doc/extend.texi: Update the built-in documentation file for the
6375 new built-in functions.
6377 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6379 * config/s390/s390.md: Remove movcc splitter.
6381 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6383 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
6384 load/store on condition.
6386 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
6389 * config/avr/avr.c (avr_encode_section_info)
6390 [progmem && !TREE_READONLY]: Error if progmem object needs
6393 2017-07-11 Michael Collison <michael.collison@arm.com>
6395 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
6398 2017-07-11 Carl Love <cel@us.ibm.com>
6400 * config/rs6000/rs6000-c.c: Add support for builtins
6401 vector unsigned int vec_parity_lsbb (vector signed int);
6402 vector unsigned int vec_parity_lsbb (vector unsigned int);
6403 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
6404 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
6405 vector unsigned long long vec_parity_lsbb (vector signed long long);
6406 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
6407 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
6408 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
6409 * doc/extend.texi: Update the built-in documentation file for the
6410 new built-in functions.
6412 2017-07-11 David Malcolm <dmalcolm@redhat.com>
6414 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
6415 (layout::m_primary_loc): New field.
6416 (layout::layout): Initialize new field. Move location filtering
6417 logic from here to...
6418 (layout::maybe_add_location_range): ...this new method. Add
6419 support for filtering to just the lines already specified by other
6421 (layout::will_show_line_p): New method.
6422 (gcc_rich_location::add_location_if_nearby): New method.
6423 (selftest::test_add_location_if_nearby): New test function.
6424 (selftest::diagnostic_show_locus_c_tests): Call it.
6425 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
6428 2017-07-11 Tom de Vries <tom@codesourcery.com>
6430 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
6431 (bb_first_real_insn): New function.
6432 (nvptx_single): Add extra initialization of broadcasted condition
6435 2017-07-11 Nathan Sidwell <nathan@acm.org>
6437 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
6439 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
6441 * doc/extend.texi (AVR Function Attributes): Remove weblink to
6442 Binutils doc as TEXI will mess them up.
6443 * doc/invoke.texi (AVR Options): Same here.
6445 2017-07-11 Daniel Cederman <cederman@gaisler.com>
6447 * config/sparc/sparc.opt (mfix-ut700): New option.
6448 (mfix-gr712rc): Likewise.
6449 (sparc_fix_b2bst): New variable.
6450 * doc/invoke.texi (SPARC options): Document them.
6451 (ARM options): Fix warnings.
6452 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
6453 instructions to prevent sequences that can trigger the store-store
6454 errata for certain LEON3FT processors.
6455 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
6456 (sparc_option_override): Set sparc_fix_b2bst appropriately.
6457 * config/sparc/sparc.md (fix_b2bst): New attribute.
6458 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
6460 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
6463 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
6465 (*rsqrtsf2_sse): Ditto.
6467 (div<mode>3): Macroize insn from divdf3 and divsf3
6468 using MODEF mode iterator.
6470 2017-07-10 Martin Sebor <msebor@redhat.com>
6472 PR tree-optimization/80397
6473 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
6474 instead of testing for equality to INTEGER_TYPE.
6476 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
6478 * config.gcc: Remove uclibc from arc target spec.
6480 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
6482 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
6484 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
6487 * lto-wrapper.c (remove_option): New function.
6488 (merge_and_complain): Merge PIC/PIE options more realistically.
6490 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
6492 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
6496 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
6497 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
6498 * config.in: Regenerate.
6499 * configure: Regenerate.
6500 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
6501 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
6502 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
6503 (TARGET_GASISR_PROLOGUES): ...target mask.
6504 * common/config/avr/avr-common.c
6505 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
6506 Set -mgas-isr-prologues.
6507 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
6508 INSERT_PASS_BEFORE for it.
6509 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
6510 * config/avr/avr.c (avr_option_override)
6511 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
6512 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
6513 (avr_attribute_table) <no_gccisr>: Add new function attribute.
6514 (avr_set_current_function) <is_no_gccisr>: Init machine field.
6515 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
6517 (make_avr_pass_pre_proep): New function.
6518 (emit_push_sfr) <treg>: Add argument to function and use it
6520 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
6521 and set machine->gasisr.yes.
6522 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
6523 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
6524 __gcc_isr.n_pushed to .L__stack_usage.
6525 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
6526 (avr_asm_final_postscan_insn): ...this new static function.
6527 * config/avr/avr.h (machine_function)
6528 <is_no_gccisr, use_L__stack_usage>: New fields.
6529 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
6530 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
6531 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
6532 (gasisr, *gasisr): New expander and insn.
6533 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
6534 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
6535 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
6537 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
6539 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
6542 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
6544 Move jump-tables out of .text again.
6547 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
6548 (ASM_OUTPUT_ADDR_VEC): New function.
6549 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
6550 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
6551 INSN_ADDRESSes as asm comment.
6552 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
6553 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
6554 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
6555 * config/avr/avr.md (*tablejump): Adjust comment.
6556 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
6557 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
6559 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
6560 (avr_output_addr_vec): New proto.
6561 (avr_log_t) <insn_addresses>: New field.
6563 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
6566 * config/i386/i386.c (ix86_function_arg_advance): Set
6567 outgoing_args_on_stack to true if there are outgoing arguments
6569 (ix86_function_arg): Likewise.
6570 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
6571 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
6572 * config/i386/i386.h (machine_function): Add
6573 outgoing_args_on_stack.
6575 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
6577 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
6579 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
6581 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
6583 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
6584 (REAL_H): Remove $(MACHMODE_H).
6585 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
6587 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
6588 $(MACHMODE_H) and double-int.h.
6589 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
6591 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
6594 2017-07-07 Andrew Pinski <apinski@cavium.com>
6596 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
6597 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
6599 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
6601 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
6602 Add warning if GCC was not configured to link against a GLIBC that
6603 exports the hardware capability bits.
6604 (make_resolver_func): Make resolver function private and not a
6605 COMDAT function. Create the name with clone_function_name instead
6606 of make_unique_name.
6609 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
6610 correct operand in doing the split.
6612 2017-07-07 Carl Love <cel@us.ibm.com>
6614 * config/rs6000/rs6000-c: Add support for built-in function
6615 vector unsigned short vec_pack_to_short_fp32 (vector float,
6617 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
6618 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
6619 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
6620 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
6621 (convert_4f32_8i16): Add define_expand.
6622 * doc/extend.texi: Update the built-in documentation file for the
6623 new built-in function.
6625 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6627 * config/sparc/m8.md: New file.
6628 * config/sparc/sparc.md: Include m8.md.
6630 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6632 * config/sparc/sparc.opt: New option -mvis4b.
6633 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
6634 (sparc_option_override): Handle VIS4B.
6635 (enum sparc_builtins): Define
6636 SPARC_BUILTIN_DICTUNPACK{8,16,32},
6637 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
6638 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
6639 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
6640 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
6641 (check_constant_argument): New function.
6642 (sparc_vis_init_builtins): Define builtins
6643 __builtin_vis_dictunpack{8,16,32},
6644 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
6645 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
6646 __builtin_vis_fpcmpde{8,16,32}shl and
6647 __builtin_vis_fpcmpur{8,16,32}shl.
6648 (sparc_expand_builtin): Check that the constant operands to
6649 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
6650 constant and in range.
6651 * config/sparc/sparc-c.c (sparc_target_macros): Handle
6653 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
6654 (SPARC_IMM5_P): Likewise.
6655 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
6656 (enabled): Handle vis4b.
6657 (UNSPEC_DICTUNPACK): New unspec.
6658 (UNSPEC_FPCMPSHL): Likewise.
6659 (UNSPEC_FPUCMPSHL): Likewise.
6660 (UNSPEC_FPCMPDESHL): Likewise.
6661 (UNSPEC_FPCMPURSHL): Likewise.
6662 (cpu_feature): New CPU feature `vis4b'.
6663 (dictunpack{8,16,32}): New insns.
6664 (FPCSMODE): New mode iterator.
6665 (fpcscond): New code iterator.
6666 (fpcsucond): Likewise.
6667 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
6668 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
6669 (fpcmpde{8,16,32}{si,di}shl): Likewise.
6670 (fpcmpur{8,16,32}{si,di}shl): Likewise.
6671 * config/sparc/constraints.md: Define constraints `q' for unsigned
6672 2-bit integer constants and `t' for unsigned 5-bit integer
6674 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
6676 (imm5_operand_dictunpack16): Likewise.
6677 (imm5_operand_dictunpack32): Likewise.
6678 (imm2_operand): Likewise.
6679 * doc/invoke.texi (SPARC Options): Document -mvis4b.
6680 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
6681 ditunpack* and fpcmp*shl builtins.
6683 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6685 * config.gcc: Handle m8 in --with-{cpu,tune} options.
6686 * config.in: Add HAVE_AS_SPARC6 define.
6687 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
6689 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
6691 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
6692 (CPP_CPU_SPEC): Handle m8.
6693 (ASM_CPU_SPEC): Likewise.
6694 * config/sparc/sparc-opts.h (enum processor_type): Add
6696 * config/sparc/sparc.c (m8_costs): New struct.
6697 (sparc_option_override): Handle TARGET_CPU_m8.
6698 (sparc32_initialize_trampoline): Likewise.
6699 (sparc64_initialize_trampoline): Likewise.
6700 (sparc_issue_rate): Likewise.
6701 (sparc_register_move_cost): Likewise.
6702 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
6703 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
6704 (ASM_CPU64_DEFAULT_SPEC): Likewise.
6705 (CPP_CPU_SPEC): Handle M8.
6706 (ASM_CPU_SPEC): Likewise.
6707 (AS_M8_FLAG): Define.
6708 * config/sparc/sparc.md: Add m8 to the cpu attribute.
6709 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
6710 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
6712 * configure: Regenerate.
6713 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
6716 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6718 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
6720 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
6721 ("*movdi_insn_sp32"): Do not set v3pipe.
6722 ("*movsi_insn"): Likewise.
6723 ("*movdi_insn_sp64"): Likewise.
6724 ("*movsf_insn"): Likewise.
6725 ("*movdf_insn_sp32"): Likewise.
6726 ("*movdf_insn_sp64"): Likewise.
6727 ("*zero_extendsidi2_insn_sp64"): Likewise.
6728 ("*sign_extendsidi2_insn"): Likewise.
6729 ("*mov<VM32:mode>_insn"): Likewise.
6730 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6731 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6732 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6733 ("<vlop:code><VL:mode>3"): Likewise.
6734 ("*not_<vlop:code><VL:mode>3"): Likewise.
6735 ("*nand<VL:mode>_vis"): Likewise.
6736 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
6737 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
6738 ("one_cmpl<VL:mode>2"): Likewise.
6739 ("faligndata<VM64:mode>_vis"): Likewise.
6740 ("alignaddrsi_vis"): Likewise.
6741 ("alignaddrdi_vis"): Likweise.
6742 ("alignaddrlsi_vis"): Likewise.
6743 ("alignaddrldi_vis"): Likewise.
6744 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6745 ("bmaskdi_vis"): Likewise.
6746 ("bmasksi_vis"): Likewise.
6747 ("bshuffle<VM64:mode>_vis"): Likewise.
6748 ("cmask8<P:mode>_vis"): Likewise.
6749 ("cmask16<P:mode>_vis"): Likewise.
6750 ("cmask32<P:mode>_vis"): Likewise.
6751 ("pdistn<P:mode>_vis"): Likewise.
6752 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6754 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6756 * config/sparc/sparc.md ("subtype"): New insn attribute.
6757 ("*wrgsr_sp64"): Set insn subtype.
6758 ("*rdgsr_sp64"): Likewise.
6759 ("alignaddrsi_vis"): Likewise.
6760 ("alignaddrdi_vis"): Likewise.
6761 ("alignaddrlsi_vis"): Likewise.
6762 ("alignaddrldi_vis"): Likewise.
6763 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
6764 ("fexpand_vis"): Likewise.
6765 ("fpmerge_vis"): Likewise.
6766 ("faligndata<VM64:mode>_vis"): Likewise.
6767 ("bshuffle<VM64:mode>_vis"): Likewise.
6768 ("cmask8<P:mode>_vis"): Likewise.
6769 ("cmask16<P:mode>_vis"): Likewise.
6770 ("cmask32<P:mode>_vis"): Likewise.
6771 ("fchksm16_vis"): Likewise.
6772 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
6773 ("fmean16_vis"): Likewise.
6774 ("fp<plusminus_insn>64_vis"): Likewise.
6775 ("<plusminus_insn>v8qi3"): Likewise.
6776 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
6777 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
6778 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
6779 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
6780 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
6781 ("*movqi_insn"): Likewise.
6782 ("*movhi_insn"): Likewise.
6783 ("*movsi_insn"): Likewise.
6784 ("movsi_pic_gotdata_op"): Likewise.
6785 ("*movdi_insn_sp32"): Likewise.
6786 ("*movdi_insn_sp64"): Likewise.
6787 ("movdi_pic_gotdata_op"): Likewise.
6788 ("*movsf_insn"): Likewise.
6789 ("*movdf_insn_sp32"): Likewise.
6790 ("*movdf_insn_sp64"): Likewise.
6791 ("*zero_extendhisi2_insn"): Likewise.
6792 ("*zero_extendqihi2_insn"): Likewise.
6793 ("*zero_extendqisi2_insn"): Likewise.
6794 ("*zero_extendqidi2_insn"): Likewise.
6795 ("*zero_extendhidi2_insn"): Likewise.
6796 ("*zero_extendsidi2_insn_sp64"): Likewise.
6797 ("ldfsr"): Likewise.
6798 ("prefetch_64"): Likewise.
6799 ("prefetch_32"): Likewise.
6800 ("tie_ld32"): Likewise.
6801 ("tie_ld64"): Likewise.
6802 ("*tldo_ldub_sp32"): Likewise.
6803 ("*tldo_ldub1_sp32"): Likewise.
6804 ("*tldo_ldub2_sp32"): Likewise.
6805 ("*tldo_ldub_sp64"): Likewise.
6806 ("*tldo_ldub1_sp64"): Likewise.
6807 ("*tldo_ldub2_sp64"): Likewise.
6808 ("*tldo_ldub3_sp64"): Likewise.
6809 ("*tldo_lduh_sp32"): Likewise.
6810 ("*tldo_lduh1_sp32"): Likewise.
6811 ("*tldo_lduh_sp64"): Likewise.
6812 ("*tldo_lduh1_sp64"): Likewise.
6813 ("*tldo_lduh2_sp64"): Likewise.
6814 ("*tldo_lduw_sp32"): Likewise.
6815 ("*tldo_lduw_sp64"): Likewise.
6816 ("*tldo_lduw1_sp64"): Likewise.
6817 ("*tldo_ldx_sp64"): Likewise.
6818 ("*mov<VM32:mode>_insn"): Likewise.
6819 ("*mov<VM64:mode>_insn_sp64"): Likewise.
6820 ("*mov<VM64:mode>_insn_sp32"): Likewise.
6822 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6824 * config/sparc/sparc.md ("type"): New insn type viscmp.
6825 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
6827 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
6828 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
6829 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
6830 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
6832 ("n7_vis_logical_11cycle"): Likewise.
6833 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
6834 * config/sparc/niagara2.md ("niag3_vis": Likewise.
6835 * config/sparc/niagara.md ("niag_vis"): Likewise.
6836 * config/sparc/ultra3.md ("us3_fga"): Likewise.
6837 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
6839 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
6841 * config/sparc/sparc.md: New instruction type `bmask'.
6842 (bmaskdi_vis): Use the `bmask' type.
6843 (bmasksi_vis): Likewise.
6844 * config/sparc/ultra3.md (us3_array): Likewise.
6845 * config/sparc/niagara7.md (n7_array): Likewise.
6846 * config/sparc/niagara4.md (n4_array): Likewise.
6847 * config/sparc/niagara2.md (niag2_vis): Likewise.
6848 (niag3_vis): Likewise.
6849 * config/sparc/niagara.md (niag_vis): Likewise.
6851 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6853 * ipa-comdats.c: Remove optimize check from gate.
6854 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
6855 for functions not optimized.
6856 (ipa_fn_summary_read): Skip optimize check.
6857 (ipa_fn_summary_write): Likewise.
6858 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
6860 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
6862 (can_inline_edge_p): Check flag_pcc_struct_return for match.
6863 (check_callers): Give up on caller which is not optimized.
6864 (inline_small_functions): Likewise.
6865 (ipa_inline): Do not give up when not optimizing.
6866 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
6867 away unoptimizes cdtors.
6868 (whole_program_function_and_variable_visibility): Do
6869 ipa_discover_readonly_nonaddressable_vars in LTO mode.
6870 * ipa.c (process_references): Do not check optimize.
6871 (symbol_table::remove_unreachable_nodes): Update optimize check.
6872 (set_writeonly_bit): Update optimize check.
6873 (pass_ipa_cdtor_merge::gate): Do not check optimize.
6874 (pass_ipa_single_use::gate): Remove.
6876 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
6878 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
6879 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
6880 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
6881 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
6882 permute_load, permute_store, adjust_extract, adjust_splat,
6883 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
6884 replace_swap_with_copy, dump_swap_insn_table,
6885 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
6886 recombine_lvx_pattern, recombine_stvx_pattern,
6887 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
6888 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
6889 to file rs6000-p8swap.c.
6890 * config/rs6000/rs6000-p8swap.c: New file.
6891 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
6892 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
6893 and rs6000*-*-* targets.
6895 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6897 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
6899 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6901 * lto-wrapper.c (merge_and_complain): Do not merge
6902 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
6903 fsigned_zeros, ftrapping_math, fwrapv.
6904 (append_compiler_options): Do not track these options.
6905 (append_linker_options): Likewie
6907 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6909 * cgraphunit.c (cgraph_node::finalize_function): When
6910 !flag_toplevel_reorde set no_reorder flag.
6911 (varpool_node::finalize_decl): Likewise.
6912 (symbol_table::compile): Drop no toplevel reorder path.
6914 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6916 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
6917 edges; zero probability is not better than uninitialized.
6919 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6921 * asan.h (asan_sanitize_allocas_p): Declare.
6922 * asan.c (asan_sanitize_allocas_p): New function.
6923 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
6924 (handle_builtin_alloca): Likewise.
6925 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
6926 if !asan_sanitize_allocas_p.
6927 * params.def (asan-instrument-allocas): Add new option.
6928 * params.h (ASAN_PROTECT_ALLOCAS): Define.
6929 * opts.c (common_handle_option): Disable allocas sanitization for
6932 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
6934 * asan.c: Include gimple-fold.h.
6935 (get_last_alloca_addr): New function.
6936 (handle_builtin_stackrestore): Likewise.
6937 (handle_builtin_alloca): Likewise.
6938 (asan_emit_allocas_unpoison): Likewise.
6939 (get_mem_refs_of_builtin_call): Add new parameter, remove const
6940 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
6941 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
6942 (instrument_builtin_call): Pass gimple iterator to
6943 get_mem_refs_of_builtin_call.
6944 (last_alloca_addr): New global.
6945 * asan.h (asan_emit_allocas_unpoison): Declare.
6946 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
6947 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
6948 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
6949 if function calls alloca.
6950 * gimple-fold.c (replace_call_with_value): Remove static keyword.
6951 * gimple-fold.h (replace_call_with_value): Declare.
6952 * internal-fn.c: Include asan.h.
6953 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
6954 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
6956 2017-07-06 David Malcolm <dmalcolm@redhat.com>
6958 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
6959 (C_SELFTEST_FLAGS): New.
6960 (CPP_SELFTEST_FLAGS): New.
6961 (SELFTEST_DEPS): New, from deps of s-selftest.
6962 (C_SELFTEST_DEPS): New, from deps of s-selftest.
6963 (CPP_SELFTEST_DEPS): New.
6964 (selftest): Add dependency on s-selftest-c++.
6965 (s-selftest): Rename to...
6966 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
6967 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
6968 than SELFTEST_FLAGS.
6969 (selftest-gdb): Rename to...
6970 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
6972 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
6973 (selftest-valgrind): Rename to...
6974 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
6976 (selftest-valgrind): Reintroduce as an alias for
6977 selftest-c-valgrind.
6978 (s-selftest-c++): New.
6979 (selftest-c++-gdb): New.
6980 (selftest-c++-valgrind): New.
6982 2017-07-06 Olivier Hainque <hainque@adacore.com>
6984 * gcc.c (process_command): When deciding if undefined variables
6985 should be ignored when processing specs, accept "gcc -v" as well.
6987 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
6989 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
6990 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
6992 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
6994 * config/arm/arm-cpus.in (armv8-r): Add new entry.
6995 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
6996 * config/arm/arm-tables.opt: Regenerate.
6997 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
6999 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
7001 2017-07-06 Carl Love <cel@us.ibm.com>
7003 * ChangeLog: Clean up from mid air collision
7005 2017-07-06 Carl Love <cel@us.ibm.com>
7007 * config/rs6000/rs6000-c.c: Add support for built-in functions
7008 vector signed int vec_subc (vector signed int, vector signed int);
7009 vector signed __int128 vec_subc (vector signed __int128,
7010 vector signed __int128);
7011 vector unsigned __int128 vec_subc (vector unsigned __int128,
7012 vector unsigned __int128);
7013 vector signed int vec_sube (vector signed int, vector signed int,
7015 vector unsigned int vec_sube (vector unsigned int,
7016 vector unsigned int,
7017 vector unsigned int);
7018 vector signed __int128 vec_sube (vector signed __int128,
7019 vector signed __int128,
7020 vector signed__int128);
7021 vector unsigned __int128 vec_sube (vector unsigned __int128,
7022 vector unsigned __int128,
7023 vector unsigned __int128);
7024 vector signed int vec_subec (vector signed int, vector signed int,
7026 vector unsigned int vec_subec (vector unsigned int,
7027 vector unsigned int,
7028 vector unsigned int);
7029 vector signed __int128 vec_subec (vector signed __int128,
7030 vector signed __int128,
7031 vector signed__int128);
7032 vector unsigned __int128 vec_subec (vector unsigned __int128,
7033 vector unsigned __int128,
7034 vector unsigned __int128);
7035 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
7036 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
7037 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
7038 BU_ALTIVEC_OVERLOAD_X definitions.
7039 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
7040 * doc/extend.texi: Update the built-in documentation file for the new
7043 2017-07-06 David Malcolm <dmalcolm@redhat.com>
7046 * diagnostic-show-locus.c (layout::layout): Use start and finish
7047 spelling location for the start and finish of each range.
7048 * genmatch.c (linemap_client_expand_location_to_spelling_point):
7049 Add unused aspect param.
7050 * input.c (expand_location_1): Add "aspect" param, and use it
7051 to access the correct part of the location.
7052 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
7054 (expand_location_to_spelling_point): Likewise.
7055 (linemap_client_expand_location_to_spelling_point): Add "aspect"
7056 param, and pass it to expand_location_1.
7058 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
7060 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
7061 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
7062 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
7063 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
7064 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
7065 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
7066 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
7067 _mm_maskz_getmant_ss): New intrinsics.
7068 (__builtin_ia32_getexpss128_mask): Changed to ...
7069 __builtin_ia32_getexpss128_round ... this.
7070 (__builtin_ia32_getexpsd128_mask): Changed to ...
7071 __builtin_ia32_getexpsd128_round ... this.
7072 * config/i386/i386-builtin-types.def
7073 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
7074 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
7075 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
7076 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
7077 __builtin_ia32_getmantss_mask_round): New builtins.
7078 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
7079 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
7080 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
7081 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
7082 * config/i386/sse.md
7083 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
7084 avx512f_sgetexp<mode><mask_scalar_name>
7085 <round_saeonly_scalar_name> ... this.
7086 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
7087 %0, %1, %2<round_saeonly_op3>}): Changed to ...
7088 vgetexp<ssescalarmodesuffix>
7089 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7090 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
7091 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
7092 avx512f_vgetmant<mode><mask_scalar_name>
7093 <round_saeonly_scalar_name> ... this.
7094 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
7095 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
7096 vgetmant<ssescalarmodesuffix>
7097 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
7098 %0<mask_scalar_operand4>, %1, %2
7099 <round_saeonly_scalar_mask_op4>, %3} ... this.
7100 * config/i386/subst.md (mask_scalar_operand4,
7101 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
7102 round_saeonly_scalar_nimm_predicate): New subst attributes.
7104 2017-07-06 Julia Koval <julia.koval@intel.com>
7106 * config/i386/i386.c (ix86_erase_embedded_rounding):
7107 Remove code for old rounding pattern.
7109 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
7111 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
7113 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
7115 * doc/sourcebuild.texi (Test Directives, Variants of
7116 dg-require-support): Add documentation for dg-require-stack-check.
7118 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
7120 * config/i386/subst.md (mask_scalar, round_scalar,
7121 round_saeonly_scalar): New meta-templates.
7122 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
7123 round_scalar_mask_operand3, round_scalar_mask_op3,
7124 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
7125 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
7126 round_saeonly_scalar_constraint,
7127 round_saeonly_scalar_prefix): New subst attribute.
7128 * config/i386/sse.md
7129 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
7130 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
7131 <round_scalar_name> ... this.
7132 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
7133 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
7134 <round_scalar_name> ... this.
7135 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
7136 <sse>_vm<code><mode>3<mask_scalar_name>
7137 <round_saeonly_scalar_name> ... this.
7138 (v<plusminus_mnemonic><ssescalarmodesuffix>
7139 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
7140 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
7141 v<plusminus_mnemonic><ssescalarmodesuffix>
7142 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7143 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
7144 (v<multdiv_mnemonic><ssescalarmodesuffix>
7145 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
7146 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
7147 v<multdiv_mnemonic><ssescalarmodesuffix>
7148 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7149 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
7150 (v<maxmin_float><ssescalarmodesuffix>
7151 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
7152 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
7153 v<maxmin_float><ssescalarmodesuffix>
7154 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
7155 %0<mask_scalar_operand3>, %1, %<iptr>2
7156 <round_saeonly_scalar_mask_op3>} ... this.
7158 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
7160 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
7161 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
7163 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
7164 Alan Hayward <alan.hayward@arm.com>
7165 David Sherwood <david.sherwood@arm.com>
7167 * combine.c (simplify_if_then_else): Remove "enum" before
7169 * compare-elim.c (can_eliminate_compare): Likewise.
7170 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
7172 (aarch64_lookup_simd_builtin_type): Likewise.
7173 (aarch64_simd_builtin_type): Likewise.
7174 (aarch64_init_simd_builtin_types): Likewise.
7175 (aarch64_simd_expand_args): Likewise.
7176 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
7178 (aarch64_reverse_mask): Likewise.
7179 (aarch64_simd_emit_reg_reg_move): Likewise.
7180 (aarch64_gen_adjusted_ldpstp): Likewise.
7181 (aarch64_ccmp_mode_to_code): Likewise.
7182 (aarch64_operands_ok_for_ldpstp): Likewise.
7183 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7184 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
7186 (aarch64_min_divisions_for_recip_mul): Likewise.
7187 (aarch64_reassociation_width): Likewise.
7188 (aarch64_get_condition_code_1): Likewise.
7189 (aarch64_simd_emit_reg_reg_move): Likewise.
7190 (aarch64_simd_attr_length_rglist): Likewise.
7191 (aarch64_reverse_mask): Likewise.
7192 (aarch64_operands_ok_for_ldpstp): Likewise.
7193 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7194 (aarch64_gen_adjusted_ldpstp): Likewise.
7195 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
7197 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
7198 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
7199 (arm_lookup_simd_builtin_type): Likewise.
7200 (arm_simd_builtin_type): Likewise.
7201 (arm_init_simd_builtin_types): Likewise.
7202 (arm_expand_builtin_args): Likewise.
7203 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
7204 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
7205 (ft32_setup_incoming_varargs): Likewise.
7206 (ft32_function_arg): Likewise.
7207 (ft32_function_arg_advance): Likewise.
7208 (ft32_pass_by_reference): Likewise.
7209 (ft32_arg_partial_bytes): Likewise.
7210 (ft32_valid_pointer_mode): Likewise.
7211 (ft32_addr_space_pointer_mode): Likewise.
7212 (ft32_addr_space_legitimate_address_p): Likewise.
7213 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
7215 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
7216 (ix86_emit_outlined_ms2sysv_restore): Likewise.
7217 (iamcu_alignment): Likewise.
7218 (canonicalize_vector_int_perm): Likewise.
7219 (ix86_noce_conversion_profitable_p): Likewise.
7220 (ix86_mpx_bound_mode): Likewise.
7221 (ix86_operands_ok_for_move_multiple): Likewise.
7222 * config/microblaze/microblaze-protos.h
7223 (microblaze_expand_conditional_branch_reg): Likewise.
7224 * config/microblaze/microblaze.c
7225 (microblaze_expand_conditional_branch_reg): Likewise.
7226 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
7228 (rs6000_reassociation_width): Likewise.
7229 (rs6000_invalid_binary_op): Likewise.
7230 (fusion_p9_p): Likewise.
7231 (emit_fusion_p9_load): Likewise.
7232 (emit_fusion_p9_store): Likewise.
7233 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
7235 (riscv_hard_regno_mode_ok_p): Likewise.
7236 (riscv_address_insns): Likewise.
7237 (riscv_split_symbol): Likewise.
7238 (riscv_legitimize_move): Likewise.
7239 (riscv_function_value): Likewise.
7240 (riscv_hard_regno_nregs): Likewise.
7241 (riscv_expand_builtin): Likewise.
7242 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
7243 (riscv_build_integer): Likewise.
7244 (riscv_split_integer): Likewise.
7245 (riscv_legitimate_constant_p): Likewise.
7246 (riscv_cannot_force_const_mem): Likewise.
7247 (riscv_regno_mode_ok_for_base_p): Likewise.
7248 (riscv_valid_base_register_p): Likewise.
7249 (riscv_valid_offset_p): Likewise.
7250 (riscv_valid_lo_sum_p): Likewise.
7251 (riscv_classify_address): Likewise.
7252 (riscv_legitimate_address_p): Likewise.
7253 (riscv_address_insns): Likewise.
7254 (riscv_load_store_insns): Likewise.
7255 (riscv_force_binary): Likewise.
7256 (riscv_split_symbol): Likewise.
7257 (riscv_force_address): Likewise.
7258 (riscv_legitimize_address): Likewise.
7259 (riscv_move_integer): Likewise.
7260 (riscv_legitimize_const_move): Likewise.
7261 (riscv_legitimize_move): Likewise.
7262 (riscv_address_cost): Likewise.
7263 (riscv_subword): Likewise.
7264 (riscv_output_move): Likewise.
7265 (riscv_canonicalize_int_order_test): Likewise.
7266 (riscv_emit_int_order_test): Likewise.
7267 (riscv_function_arg_boundary): Likewise.
7268 (riscv_pass_mode_in_fpr_p): Likewise.
7269 (riscv_pass_fpr_single): Likewise.
7270 (riscv_pass_fpr_pair): Likewise.
7271 (riscv_get_arg_info): Likewise.
7272 (riscv_function_arg): Likewise.
7273 (riscv_function_arg_advance): Likewise.
7274 (riscv_arg_partial_bytes): Likewise.
7275 (riscv_function_value): Likewise.
7276 (riscv_pass_by_reference): Likewise.
7277 (riscv_setup_incoming_varargs): Likewise.
7278 (riscv_print_operand): Likewise.
7279 (riscv_elf_select_rtx_section): Likewise.
7280 (riscv_save_restore_reg): Likewise.
7281 (riscv_for_each_saved_reg): Likewise.
7282 (riscv_register_move_cost): Likewise.
7283 (riscv_hard_regno_mode_ok_p): Likewise.
7284 (riscv_hard_regno_nregs): Likewise.
7285 (riscv_class_max_nregs): Likewise.
7286 (riscv_memory_move_cost): Likewise.
7287 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
7288 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
7289 (rl78_addr_space_address_mode): Likewise.
7290 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7292 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
7293 (rs6000_reassociation_width): Likewise.
7294 (rs6000_invalid_binary_op): Likewise.
7295 (fusion_p9_p): Likewise.
7296 (emit_fusion_p9_load): Likewise.
7297 (emit_fusion_p9_store): Likewise.
7298 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
7299 (ok_for_simple_move_operands): Likewise.
7300 (ok_for_simple_move_strict_operands): Likewise.
7301 (ok_for_simple_arith_logic_operands): Likewise.
7302 (visium_legitimize_reload_address): Likewise.
7303 (visium_select_cc_mode): Likewise.
7304 (output_cbranch): Likewise.
7305 (visium_split_double_move): Likewise.
7306 (visium_expand_copysign): Likewise.
7307 (visium_expand_int_cstore): Likewise.
7308 (visium_expand_fp_cstore): Likewise.
7309 * config/visium/visium.c (visium_pass_by_reference): Likewise.
7310 (visium_function_arg): Likewise.
7311 (visium_function_arg_advance): Likewise.
7312 (visium_libcall_value): Likewise.
7313 (visium_setup_incoming_varargs): Likewise.
7314 (visium_legitimate_constant_p): Likewise.
7315 (visium_legitimate_address_p): Likewise.
7316 (visium_legitimize_address): Likewise.
7317 (visium_secondary_reload): Likewise.
7318 (visium_register_move_cost): Likewise.
7319 (visium_memory_move_cost): Likewise.
7320 (prepare_move_operands): Likewise.
7321 (ok_for_simple_move_operands): Likewise.
7322 (ok_for_simple_move_strict_operands): Likewise.
7323 (ok_for_simple_arith_logic_operands): Likewise.
7324 (visium_function_value_1): Likewise.
7325 (rtx_ok_for_offset_p): Likewise.
7326 (visium_legitimize_reload_address): Likewise.
7327 (visium_split_double_move): Likewise.
7328 (visium_expand_copysign): Likewise.
7329 (visium_expand_int_cstore): Likewise.
7330 (visium_expand_fp_cstore): Likewise.
7331 (visium_split_cstore): Likewise.
7332 (visium_select_cc_mode): Likewise.
7333 (visium_split_cbranch): Likewise.
7334 (output_cbranch): Likewise.
7335 (visium_print_operand_address): Likewise.
7336 * expmed.c (flip_storage_order): Likewise.
7337 * expmed.h (emit_cstore): Likewise.
7338 (flip_storage_order): Likewise.
7339 * genrecog.c (validate_pattern): Likewise.
7340 * hsa-gen.c (gen_hsa_addr): Likewise.
7341 * internal-fn.c (expand_arith_overflow): Likewise.
7342 * ira-color.c (allocno_copy_cost_saving): Likewise.
7343 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7344 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
7345 (process_invariant_for_inheritance): Likewise.
7346 * lra-eliminations.c (move_plus_up): Likewise.
7347 * omp-low.c (lower_oacc_reductions): Likewise.
7348 * simplify-rtx.c (simplify_subreg): Likewise.
7349 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
7350 (TARGET_CHKP_BOUND_MODE): Likewise..
7351 * targhooks.c (default_chkp_bound_mode): Likewise.
7352 (default_setup_incoming_vararg_bounds): Likewise.
7353 * targhooks.h (default_chkp_bound_mode): Likewise.
7354 (default_setup_incoming_vararg_bounds): Likewise.
7355 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
7356 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
7357 (have_whole_vector_shift): Likewise.
7358 * tree-vect-stmts.c (vectorizable_load): Likewise.
7359 * doc/tm.texi: Regenerate.
7361 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
7363 Graceful degrade if Binutils PR21472 is not available.
7366 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
7367 .rodata in flash test fails.
7368 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
7369 * confgure: Regenerate.
7370 * config.in: Regenerate.
7371 * config/avr/avr.c (avr_asm_named_section)
7372 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
7373 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
7374 (avr_asm_init_sections): Same.
7376 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
7378 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
7379 (fma<VH:mode>4_intrinsic): Likewise.
7380 (*fmsub<VCVTF:mode>4): Likewise.
7381 (*fmsub<VH:mode>4_intrinsic): Likewise.
7383 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
7386 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
7387 Don't depend on "optimize > 0".
7388 (out_movhi_r_mr, out_movqi_mr_r): Same.
7389 (out_movhi_mr_r, out_movqi_r_mr): Same.
7390 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
7391 io_address_operand on "optimize > 0".
7393 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7395 * tree-loop-distribution.c: Add general explanantion on the pass.
7396 (generate_loops_for_partition): Mark distributed loop.
7397 (pg_add_dependence_edges): New parameter. Handle alias data
7398 dependence specially and record it in the parameter if asked.
7399 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
7400 (init_partition_graph_vertices, add_partition_graph_edge): New.
7401 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
7402 (free_partition_graph_vdata, build_partition_graph): New.
7403 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
7404 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
7405 (data_ref_segment_size, latch_dominated_by_data_ref): New.
7406 (compute_alias_check_pairs, version_loop_by_alias_check): New.
7407 (version_for_distribution_p, finalize_partitions): New.
7408 (distribute_loop): Handle alias data dependence specially. Factor
7409 out loop fusion code as functions and call these functions.
7411 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7413 * tree-loop-distribution.c (classify_partition): New parameter and
7414 better handle reduction statement.
7415 (rdg_build_partitions): Revise comment.
7416 (distribute_loop): Compute statements in all partitions and pass it
7417 to classify_partition.
7419 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7421 * tree-loop-distribution.c (enum partition_type): New.
7422 (struct partition): New field type.
7423 (partition_merge_into): Add parameter. Update partition type.
7424 (data_dep_in_cycle_p, update_type_for_merge): New functions.
7425 (build_rdg_partition_for_vertex): Compute partition type.
7426 (rdg_build_partitions): Dump partition type.
7427 (distribute_loop): Update calls to partition_merge_into.
7429 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7431 * tree-loop-distribution.c (struct ddr_hasher): New.
7432 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
7434 (classify_partition): Call get_data_dependence.
7435 (pg_add_dependence_edges): Ditto.
7436 (distribute_loop): Release data dependence hash table.
7438 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7440 * tree-loop-distribution.c (ref_base_address): Delete.
7441 (similar_memory_accesses): Rename ...
7442 (share_memory_accesses): ... to this. Check if partitions access
7443 the same memory reference.
7444 (distribute_loop): Call share_memory_accesses.
7446 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7448 * tree-loop-distribution.c (struct partition): New field recording
7450 (partition_alloc, partition_free): Init and release data refs.
7451 (partition_merge_into): Merge data refs.
7452 (build_rdg_partition_for_vertex): Collect data refs for partition.
7453 (pg_add_dependence_edges): Change parameters from vector to bitmap.
7455 (distribute_loop): Remve data refs from vertice data of partition
7458 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7460 * tree-loop-distribution.c (params.h): Include header file.
7461 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
7462 (datarefs_vec): New global var.
7463 (create_rdg_vertices): Use datarefs_vec directly.
7464 (free_rdg): Don't free data references.
7465 (build_rdg): Update use. Don't free data references.
7466 (distribute_loop): Compute global variable for data references.
7467 Bail out if there are too many data references.
7469 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7471 * tree-loop-distribution.c (loop_nest): New global var.
7472 (build_rdg): Use loop directly, rather than loop nest.
7473 (pg_add_dependence_edges): Remove loop nest parameter. Use global
7475 (distribute_loop): Compute global variable loop nest. Update use.
7477 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7479 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
7480 (partition_merge_into): New parameter. Dump reason for fusion.
7481 (distribute_loop): Update use of partition_merge_into.
7483 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7485 * tree-loop-distribution.c (bb_top_order_index): New.
7486 (bb_top_order_index_size, bb_top_order_cmp): New.
7487 (stmts_from_loop): Use topological order.
7488 (pass_loop_distribution::execute): Compute and release topological
7489 order for basic blocks.
7491 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7493 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
7496 2017-07-05 Bin Cheng <bin.cheng@arm.com>
7498 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
7499 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
7500 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
7501 * internal-fn.def (LOOP_DIST_ALIAS): New.
7502 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
7503 (fold_loop_internal_call): ... this.
7504 (vect_loop_dist_alias_call): New function.
7505 (set_uid_loop_bbs): Call fold_loop_internal_call.
7506 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
7509 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7512 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
7513 Require dead FLAGS_REG at the beginning of a peephole.
7515 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
7518 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
7519 arguments in the call to __builtin_ia32_sbb_u32.
7520 (_subborrow_u64): Swap _X and _Y arguments in the call to
7521 __builtin_ia32_sbb_u64.
7523 2017-07-04 Jakub Jelinek <jakub@redhat.com>
7526 * tree-vrp.c (compare_assert_loc): Turn into a function template
7527 with stable template parameter. Only test if a->e is NULL,
7528 !a->e == !b->e has been verified already. Use e == NULL or
7529 e != NULL instead of e or ! e tests. If stable is true, don't use
7530 iterative_hash_expr, on the other side allow a or b or both NULL
7531 and sort the NULLs last.
7532 (process_assert_insertions): Sort using compare_assert_loc<false>
7533 instead of compare_assert_loc, later sort using
7534 compare_assert_loc<true> before calling process_assert_insertions_for
7535 in a loop. Use break instead of continue once seen NULL pointer.
7537 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
7539 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
7540 Cortex-R7 and Cortex-R8 processors.
7542 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7544 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
7545 uninitialized while src is not.
7547 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
7549 * common/config/arm/arm-common.c: Adjust include path for
7551 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
7552 (arm-cpu.h): Create in build directory. Adjust dependency rules.
7553 (arm-cpu-data.h): Likewise.
7554 (arm-cpu-cdata.h): Likewise.
7555 * config/arm/arm-cpu.h: Delete.
7556 * config/arm/arm-cpu-cdata.h: Delete.
7557 * config/arm/arm-cpu-data.h: Delete.
7559 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
7561 * config/arm/arm-cpus.in (cortex-a55): New.
7562 (cortex-a75): Likewise.
7563 (cortex-a75.cortex-a55): Likewise.
7564 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
7566 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
7567 * config/arm/arm-cpu-cdata.h: Regenerate.
7568 * config/arm/arm-cpu-data.h: Regenerate.
7569 * config/arm/arm-cpu.h: Regenerate.
7570 * config/arm/arm-tables.opt: Regenerate.
7571 * config/arm/arm-tune.md: Regenerate.
7573 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7575 * haifa-sched.c (sched_create_recovery_edges): Update profile.
7577 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
7579 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
7582 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
7584 PR tree-optimization/81292
7585 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
7586 full_string_p, also call adjust_related_strinfos if the adjustment
7587 is simple, otherwise invalidate related strinfos.
7589 2017-07-04 Martin Liska <mliska@suse.cz>
7592 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
7593 newly created variable as DECL_IGNORED_P.
7595 2017-07-04 Martin Liska <mliska@suse.cz>
7598 * ipa-inline.c (inline_small_functions):
7599 Use xstrdup_for_dump.
7601 2017-07-04 Tom de Vries <tom@codesourcery.com>
7603 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
7605 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
7608 * config/darwin.c (darwin_function_switched_text_sections):
7611 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
7613 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
7615 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
7617 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
7619 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7621 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
7622 min_profitable_iters, and th as inclusive lower bounds.
7623 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
7624 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
7625 for min_profitable_iters and min_profitable_estimate.
7626 (vect_transform_loop): Treat th as an inclusive lower bound.
7627 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
7629 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
7632 * config/darwin.c (darwin_function_switched_text_sections):
7633 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
7634 in two pieces, and suppress the use of buf.
7636 2017-07-03 Nathan Sidwell <nathan@acm.org>
7638 * hash-table.h (hash_table_mod1): Fix indentation.
7640 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7643 * predict.c (force_edge_cold): Be more careful about propagation
7645 * profile-count.h (profile_probability::guessed,
7646 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
7648 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
7650 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
7652 * doc/invoke.texi (rcpc architecture extension): Document it.
7654 2017-07-03 Richard Biener <rguenther@suse.de>
7656 PR tree-optimization/60510
7657 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
7658 the scalar reduction PHI and use it.
7659 (vectorizable_reduction): Properly guard the single_defuse_cycle
7660 path for non-SLP reduction chains where we cannot use it.
7661 Rework reduc_def/index and vector type deduction. Rework
7662 vector operand gathering during reduction op code-gen.
7663 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
7664 chains dissolve the chain and leave it to non-SLP reduction
7667 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7669 * tree-data-ref.h (dr_alignment): Declare.
7670 * tree-data-ref.c (dr_alignment): New function.
7671 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
7672 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
7674 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
7676 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7678 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
7679 and base_misalignment fields.
7680 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
7681 * tree-data-ref.c: Include builtins.h.
7682 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
7683 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
7684 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
7685 * tree-vect-data-refs.c: Include tree-cfg.h.
7686 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
7687 fields instead of calculating an alignment here.
7688 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
7689 innermost_loop_behavior fields.
7691 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7693 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
7695 (DR_STEP_ALIGNMENT): New macro.
7696 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
7697 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
7698 (create_data_ref): Print it.
7699 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
7700 to tell whether the step preserves vector (mis)alignment.
7701 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7702 Move the check for an integer step and generalise to all INTEGER_CST.
7703 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
7704 Print the outer step alignment.
7706 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7708 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
7709 with offset_alignment.
7710 (DR_ALIGNED_TO): Delete.
7711 (DR_OFFSET_ALIGNMENT): New macro.
7712 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
7713 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
7714 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
7715 (create_data_ref): Likewise.
7716 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
7717 (vect_analyze_data_refs): Likewise.
7718 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
7719 creating dummy innermost behavior.
7721 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7723 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
7724 with a "innermost_loop_behavior *" and refeence tree.
7725 * tree-data-ref.c (dr_analyze_innermost): Likewise.
7726 (create_data_ref): Update call accordingly.
7727 * tree-predcom.c (find_looparound_phi): Likewise.
7729 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7731 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
7732 fields with dr_wrt_vec_loop.
7733 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
7734 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
7735 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
7736 (vect_dr_behavior): New function.
7737 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7738 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
7739 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
7740 track whether the step preserves the misalignment.
7741 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
7742 Use vect_dr_behavior.
7743 (vect_setup_realignment): Update call accordingly.
7744 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
7745 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
7746 call to vect_create_addr_base_for_vector_ref.
7747 (vect_create_cond_for_align_checks): Likewise.
7748 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
7749 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
7750 (vect_recog_mask_conversion_pattern): Likewise.
7751 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
7752 (new_stmt_vec_info): Remove redundant zeroing.
7754 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
7756 * common/config/arm/arm-common.c (arm_be8_option): New function.
7757 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
7758 (ISA_ARMv6): Add isa_bit_be8.
7759 * config/arm/arm.h (arm_be8_option): Add prototype.
7760 (BE8_SPEC_FUNCTION): New define.
7761 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
7762 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
7763 (mlittle-endian): Similarly.
7764 (mbe8, mbe32): New options.
7765 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
7766 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
7768 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7770 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
7772 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7774 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
7775 (cleanup_tree_cfg_bb): Use it.
7776 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
7778 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
7780 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7783 * loop-doloop.c (add_test): Update profile.
7785 2017-07-03 Martin Liska <mliska@suse.cz>
7788 * sanopt.c (rewrite_usage_of_param): New function.
7789 (sanitize_rewrite_addressable_params): Likewise.
7790 (pass_sanopt::execute): Call rewrite_usage_of_param.
7792 2017-07-03 Richard Biener <rguenther@suse.de>
7794 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
7795 back to using VIEW_CONVERT_EXPR.
7797 2017-07-03 Martin Liska <mliska@suse.cz>
7800 * doc/extend.texi: Document when a resolver function is
7801 generated for target_clones.
7803 2017-07-03 Martin Liska <mliska@suse.cz>
7805 * asan.c (asan_emit_stack_protection): Unpoison just red zones
7806 and shadow memory of auto variables which are subject of
7807 use-after-scope sanitization.
7808 (asan_expand_mark_ifn): Add do set only when is_poison.
7810 2016-07-03 Richard Biener <rguenther@suse.de>
7812 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
7814 (vect_force_simple_reduction): Record reduction def -> phi mapping.
7815 (vectorizable_reduction): Perform reduction PHI creation when
7816 visiting a reduction PHI and adjust and simplify code generation
7817 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
7818 (vect_transform_loop): Visit reduction PHIs.
7819 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
7820 defs into the SLP tree.
7821 (vect_build_slp_tree): Reduction defs terminate the recursion.
7822 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
7824 (vect_get_vec_defs_for_stmt_copy): Export.
7825 (vect_get_vec_defs): Likewise.
7826 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
7828 (vect_get_vec_defs_for_stmt_copy): Declare.
7829 (vect_get_vec_defs): Likewise.
7831 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
7833 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
7834 parameter with a "loop" parameter and use it instead of the
7835 loop containing DR_STMT. Don't check simple_iv when doing
7836 BB analysis. Describe the two analysis modes in the comment.
7838 2017-07-03 Tom de Vries <tom@codesourcery.com>
7840 PR tree-optimization/69468
7841 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
7842 (find_same_succ_bb): Handle ignore_edge_flags.
7844 2017-07-03 Tom de Vries <tom@codesourcery.com>
7846 PR tree-optimization/81192
7847 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
7849 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
7851 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
7853 2017-07-03 Tom de Vries <tom@codesourcery.com>
7855 PR tree-optimization/81192
7856 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
7857 BB_SAME_SUCC (bb) == NULL.
7859 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7861 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
7864 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7866 * dumpfile.c: Include profile-count.h
7867 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
7869 (insert_cond_bb): Update profile.
7870 * tree-cfg.h (insert_cond_bb): Update prototype.
7871 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
7872 * tree-dump.c: Do not include tree-cfg.
7874 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7876 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
7878 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7880 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
7883 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
7885 * tree-complex.c (expand_complex_div_wide): update profile.
7887 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7888 Alan Hayward <alan.hayward@arm.com>
7889 David Sherwood <david.sherwood@arm.com>
7891 * Makefile.in (MACHMODE_H): Remove insn-modes.h
7892 (CORETYPES_H): New define.
7893 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
7894 (insn-modes-inline.h, s-modes-inline-h): New rules.
7895 (generated_files): Add insn-modes-inline.h.
7896 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
7897 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
7898 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
7899 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
7900 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
7901 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
7902 (build/gencodes.o, build/genconditions.o): Likewise.
7903 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
7904 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
7905 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
7906 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
7907 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
7908 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
7909 * coretypes.h: Include everything up to real.h for generators.
7910 Include insn-modes.h first. Include wide-int-print.h after
7911 wide-int.h. Include insn-modes-inline.h and then machmode.h.
7912 * machmode.h: Don't include insn-modes.h here.
7913 * function-tests.c: Remove includes of signop.h, machmode.h,
7914 double-int.h and wide-int.h.
7916 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
7918 * optc-save-gen.awk: Likewise.
7919 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
7920 * godump.c: Remove include of wide-int-print.h.
7921 * pretty-print.h: Likewise.
7922 * wide-int-print.cc: Likewise.
7923 * wide-int.cc: Likewise.
7924 * hash-map-tests.c: Remove include of signop.h.
7925 * hash-set-tests.c: Likewise.
7926 * rtl-tests.c: Likewise.
7927 * mkconfig.sh: Remove include of machmode.h.
7928 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
7930 (emit_insn_modes_inline_h): ...this new function. Emit the code
7931 into an insn-modes-inline.h header file, adding appropriate
7932 include guards and end comments.
7933 (emit_insn_modes_c_header): Remove include of machmode.h.
7934 (emit_min_insn_modes_c_header): Include coretypes.h rather than
7936 (main): Handle -i flag and call emit_insn_modes_inline_h when
7939 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7941 * tree-ssa-strlen.c (strinfo): Rename the length field to
7942 nonzero_chars. Add a full_string_p field.
7943 (compare_nonzero_chars, zero_length_string_p): New functions.
7944 (get_addr_stridx): Add an offset_out parameter.
7945 Use compare_nonzero_chars.
7946 (get_stridx): Update accordingly. Use compare_nonzero_chars.
7947 (new_strinfo): Update after above changes to strinfo.
7948 (set_endptr_and_length): Set full_string_p.
7949 (get_string_length): Update after above changes to strinfo.
7950 (unshare_strinfo): Update call to new_strinfo.
7951 (maybe_invalidate): Likewise.
7952 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
7953 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
7954 as a uhwi instead of an shwi. Update after above changes to
7955 strinfo and new_strinfo.
7956 (zero_length_string): Assert that chainsi contains full strings.
7957 Use zero_length_string_p. Update call to new_strinfo.
7958 (adjust_related_strinfos): Update after above changes to strinfo.
7959 Copy full_string_p from origsi.
7960 (adjust_last_stmt): Use zero_length_string_p.
7961 (handle_builtin_strlen): Update after above changes to strinfo and
7962 new_strinfo. Install the lhs as the string length if the previous
7963 entry didn't describe a full string.
7964 (handle_builtin_strchr): Update after above changes to strinfo
7966 (handle_builtin_strcpy): Likewise.
7967 (handle_builtin_strcat): Likewise.
7968 (handle_builtin_malloc): Likewise.
7969 (handle_pointer_plus): Likewise.
7970 (handle_builtin_memcpy): Likewise. Track nonzero characters
7971 that aren't necessarily followed by a nul terminator.
7972 (handle_char_store): Likewise.
7974 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7976 PR tree-optimization/80769
7977 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
7978 for malloc and calloc. Document the new invariant that all related
7979 strinfos have delayed lengths or none do.
7980 (verify_related_strinfos): Move earlier in file.
7981 (set_endptr_and_length): New function, split out from...
7982 (get_string_length): ...here. Also set the lengths of related
7984 (zero_length_string): Assert that chainsi has known (rather than
7986 (adjust_related_strinfos): Likewise.
7988 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
7990 PR tree-optimization/81136
7991 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
7992 assert that two references with the same misalignment have the same
7993 compile-time misalignment if those compile-time misalignments
7996 2017-07-01 Andi Kleen <ak@linux.intel.com>
7998 * print-tree.c (print_node): Print all attributes.
8000 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8002 * cfg.c (scale_bbs_frequencies): New function.
8003 * cfg.h (scale_bbs_frequencies): Declare it.
8004 * cfgloopanal.c (single_likely_exit): Cleanup.
8005 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
8007 (scale_loop_profile): Likewise.
8008 (loop_version): Likewise.
8009 (create_empty_loop_on_edge): Update.
8010 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
8011 scale_loop_frequencies, scale_loop_profile, loopify,
8012 loop_version): Update prototypes.
8013 * modulo-sched.c (sms_schedule): Update.
8014 * predict.c (unlikely_executed_edge_p): Also check probability.
8015 (probably_never_executed_edge_p): Fix typo.
8016 * tree-if-conv.c (version_loop_for_if_conversion): Update.
8017 * tree-parloops.c (gen_parallel_loop): Update.
8018 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
8019 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8020 * tree-ssa-loop-split.c (split_loop): Update.
8021 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
8022 * tree-vect-loop-manip.c (vect_do_peeling): Update.
8023 (vect_loop_versioning): Update.
8024 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8026 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8028 * trans-mem.c (split_bb_make_tm_edge): Update profile.
8030 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8032 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
8033 to keep profile consistent.
8035 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8037 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
8038 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
8039 * profile-count.h (max_safe_multiplier): Make unsigned.
8040 (profile_count::guessed_zero): New.
8042 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8044 * bb-reorder.c (fix_up_crossing_landing_pad,
8045 fix_crossing_conditional_branches): Use make_single_succ_edge
8046 to keep profile consistent.
8048 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
8050 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
8053 2017-07-01 Jakub Jelinek <jakub@redhat.com>
8056 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
8057 the right scopes, make sure cond_jump isn't preserved between multiple
8058 iterations. Search for fallthru edge whenever there are 3+ edges and
8059 use find_fallthru_edge for it.
8061 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8063 Patch by Alexander Monakov <amonakov@ispras.ru>
8064 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
8065 probabilities consistently.
8067 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8069 * pa.c (pa_expand_compare_and_swap_loop): Update call of
8070 emit_cmp_and_jump_insns.
8072 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8075 * tree-inline.c (expand_call_inline): Combine profile statuses.
8077 2017-06-30 Andrew Pinski <apinski@cavium.com>
8079 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
8080 fold_stmt returned true.
8082 2017-06-30 Nathan Sidwell <nathan@acm.org>
8084 * ggc.h (empty_string): Delete.
8085 * cfgexpand.c (expand_asm_stmt): Use plain "".
8086 * optabs.c (expand_asm_memory_barrier): Likewise.
8087 * stringpool.c (empty_string): Delete.
8088 (digit_vector, digit_string): Delete.
8089 (ggc_alloc_string): Use plain "", don't optimize single digit
8090 strings. Use ggc_alloc_atomic.
8092 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
8094 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
8095 comparison set and one other set, use the cost of the non-comparison
8098 2017-06-30 Nathan Sidwell <nathan@acm.org>
8100 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
8103 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
8105 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
8106 loops. Remove now unneeded calls to gimple_switch_set_label() that
8107 just set removed labels to NULL_TREE.
8109 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
8111 * tree-ssanames.c (set_range_info_raw): Abstract from ...
8112 (set_range_info): ...here. Only call set_range_info_raw if domain
8114 (set_nonzero_bits): Call set_range_info_raw.
8115 * tree-ssanames.h (set_range_info_raw): New.
8117 2017-06-30 Jakub Jelinek <jakub@redhat.com>
8120 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
8121 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
8122 of nonimmediate_operand and <store_mask_constraint> instead of m
8123 for the input operand. For V8FI iterator, always split if input
8124 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
8125 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
8126 <store_mask_predicate> instead of register_operand and
8127 <store_mask_constraint> instead of v for the input operand. Make
8128 sure both operands aren't MEMs for if not <mask_applied>.
8130 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
8132 * lto-wrapper.c (copy_file) Close both file descriptors before
8135 2017-06-30 Martin Liska <mliska@suse.cz>
8138 * multiple_target.c (create_dispatcher_calls): Make ifunc
8139 also for function that don't have calls or are not referenced.
8141 2017-06-30 Richard Biener <rguenther@suse.de>
8143 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
8144 analyze the first scalar stmt. Move vector type computation
8145 for the BB case here from ...
8146 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
8147 live operation processing in the SLP case properly.
8149 2017-06-30 Richard Biener <rguenther@suse.de>
8151 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
8153 2017-06-30 Martin Liska <mliska@suse.cz>
8156 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
8157 before BUILT_IN_UNWIND_RESUME when ASAN is used.
8159 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
8161 * doc/invoke.texi (AArch64): Add missing options and remove redundant
8164 2017-06-30 Richard Biener <rguenther@suse.de>
8166 PR tree-optimization/81249
8167 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
8168 condition reduction result to original scalar type.
8170 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8172 * profile-count.h (enum profile_quality): Fix typos and whitespace
8175 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8177 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
8178 type for branch probabilities.
8180 2017-06-29 Julian Brown <julian@codesourcery.com>
8181 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8183 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
8184 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
8185 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
8186 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
8188 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
8190 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
8191 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
8192 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
8193 CC usage from generic code to here.
8194 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
8195 CC usage into the target macros.
8197 2017-06-29 Maya Rashish <coypu@sdf.org>
8199 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
8202 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8204 * arm/arm-builtins.c: Include profile-count.h
8205 * except.c (sjlj_emit_function_enter): Use
8206 profile_probability::unlikely.
8208 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8210 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
8211 and tocrel_offset be pointer args rather than implicitly using
8213 (legitimate_constant_pool_address_p, rs6000_emit_move,
8214 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
8215 tocrel_offset and use in toc_relative_expr_p call.
8216 (print_operand, print_operand_address): Use static tocrel_base_oac
8217 and tocrel_offset_oac.
8218 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
8221 2017-06-29 Maya Rashish <coypu@sdf.org>
8223 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
8225 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
8227 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
8228 objects, take into account only the alignment of 'op0' and 'mode1' if
8231 2017-06-29 Steve Ellcey <sellcey@cavium.com>
8233 * ccmp.c (ccmp_tree_comparison_p): New function.
8234 (ccmp_candidate_p): Update to use above function.
8235 (get_compare_parts): New function.
8236 (expand_ccmp_next): Update to use new functions.
8237 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
8239 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
8240 take mode as argument.
8241 * ccmp.h (expand_ccmp_expr): Add mode as argument.
8242 * expr.c (expand_expr_real_1): Pass mode as argument.
8244 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
8246 * combine.c (combine_instructions): Print insns to dump_file, together
8249 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
8251 * asan.c (asan_emit_stack_protection): Update.
8252 (create_cond_insert_point): Update.
8253 * auto-profile.c (afdo_propagate_circuit): Update.
8254 * basic-block.h (struct edge_def): Turn probability to
8255 profile_probability.
8256 (EDGE_FREQUENCY): Update.
8257 * bb-reorder.c (find_traces_1_round): Update.
8258 (better_edge_p): Update.
8259 (sanitize_hot_paths): Update.
8260 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
8261 (make_single_succ_edge): Update.
8262 (check_bb_profile): Update.
8263 (dump_edge_info): Update.
8264 (update_bb_profile_for_threading): Update.
8265 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
8266 probabilitycount to 0.
8267 * cfgbuild.c (compute_outgoing_frequencies): Update.
8268 * cfgcleanup.c (try_forward_edges): Update.
8269 (outgoing_edges_match): Update.
8270 (try_crossjump_to_edge): Update.
8271 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
8272 (expand_gimple_tailcall): Update.
8273 (construct_init_block): Use make_single_succ_edge.
8274 (construct_exit_block): Use make_single_succ_edge.
8275 * cfghooks.c (verify_flow_info): Update.
8276 (redirect_edge_succ_nodup): Update.
8277 (split_edge): Update.
8278 (account_profile_record): Update.
8279 * cfgloopanal.c (single_likely_exit): Update.
8280 * cfgloopmanip.c (scale_loop_profile): Update.
8281 (set_zero_probability): Remove.
8282 (duplicate_loop_to_header_edge): Update.
8283 * cfgloopmanip.h (loop_version): Update prototype.
8284 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
8285 (force_nonfallthru_and_redirect): Update.
8286 (update_br_prob_note): Update.
8287 (rtl_verify_edges): Update.
8288 (purge_dead_edges): Update.
8289 (rtl_lv_add_condition_to_bb): Update.
8290 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
8291 * cgraphunit.c (init_lowered_empty_function): Update.
8292 (cgraph_node::expand_thunk): Update.
8293 * cilk-common.c: Include profile-count.h
8294 * dojump.c (inv): Remove.
8295 (jumpifnot): Update.
8296 (jumpifnot_1): Update.
8297 (do_jump_1): Update.
8299 (do_jump_by_parts_greater_rtx): Update.
8300 (do_compare_rtx_and_jump): Update.
8301 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
8302 do_jump_1. do_compare_rtx_and_jump): Update prototype.
8303 * dwarf2cfi.c: Include profile-count.h
8304 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
8305 (sjlj_emit_dispatch_table): Likewise.
8306 * explow.c: Include profile-count.h
8307 * expmed.c (emit_store_flag_force): Update.
8308 (do_cmp_and_jump): Update.
8309 * expr.c (compare_by_pieces_d::generate): Update.
8310 (compare_by_pieces_d::finish_mode): Update.
8311 (emit_block_move_via_loop): Update.
8312 (store_expr_with_bounds): Update.
8313 (store_constructor): Update.
8314 (expand_expr_real_2): Update.
8315 (expand_expr_real_1): Update.
8316 * expr.h (try_casesi, try_tablejump): Update prototypes.
8317 * gimple-pretty-print.c (dump_probability): Update.
8318 (dump_profile): New.
8319 (dump_gimple_label): Update.
8320 (dump_gimple_bb_header): Update.
8321 * graph.c (draw_cfg_node_succ_edges): Update.
8322 * hsa-gen.c (convert_switch_statements): Update.
8323 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
8324 (find_if_case_1): Update.
8325 (find_if_case_2): Update.
8326 * internal-fn.c (expand_arith_overflow_result_store): Update.
8327 (expand_addsub_overflow): Update.
8328 (expand_neg_overflow): Update.
8329 (expand_mul_overflow): Update.
8330 (expand_vector_ubsan_overflow): Update.
8331 * ipa-cp.c (good_cloning_opportunity_p): Update.
8332 * ipa-split.c (split_function): Use make_single_succ_edge.
8333 * ipa-utils.c (ipa_merge_profiles): Update.
8334 * loop-doloop.c (add_test): Update.
8335 (doloop_modify): Update.
8336 * loop-unroll.c (compare_and_jump_seq): Update.
8337 (unroll_loop_runtime_iterations): Update.
8338 * lra-constraints.c (lra_inheritance): Update.
8339 * lto-streamer-in.c (input_cfg): Update.
8340 * lto-streamer-out.c (output_cfg): Update.
8341 * mcf.c (adjust_cfg_counts): Update.
8342 * modulo-sched.c (sms_schedule): Update.
8343 * omp-expand.c (expand_omp_for_init_counts): Update.
8344 (extract_omp_for_update_vars): Update.
8345 (expand_omp_ordered_sink): Update.
8346 (expand_omp_for_ordered_loops): Update.
8347 (expand_omp_for_generic): Update.
8348 (expand_omp_for_static_nochunk): Update.
8349 (expand_omp_for_static_chunk): Update.
8350 (expand_cilk_for): Update.
8351 (expand_omp_simd): Update.
8352 (expand_omp_taskloop_for_outer): Update.
8353 (expand_omp_taskloop_for_inner): Update.
8354 * omp-simd-clone.c (simd_clone_adjust): Update.
8355 * optabs.c (expand_doubleword_shift): Update.
8356 (expand_abs): Update.
8357 (emit_cmp_and_jump_insn_1): Update.
8358 (expand_compare_and_swap_loop): Update.
8359 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
8360 * predict.c (predictable_edge_p): Update.
8361 (edge_probability_reliable_p): Update.
8362 (set_even_probabilities): Update.
8363 (combine_predictions_for_insn): Update.
8364 (combine_predictions_for_bb): Update.
8365 (propagate_freq): Update.
8366 (estimate_bb_frequencies): Update.
8367 (force_edge_cold): Update.
8368 * profile-count.c (profile_count::dump): Add missing space into dump.
8369 (profile_count::debug): Add newline.
8370 (profile_count::differs_from_p): Explicitly convert to unsigned.
8371 (profile_count::stream_in): Update.
8372 (profile_probability::dump): New member function.
8373 (profile_probability::debug): New member function.
8374 (profile_probability::differs_from_p): New member function.
8375 (profile_probability::differs_lot_from_p): New member function.
8376 (profile_probability::stream_in): New member function.
8377 (profile_probability::stream_out): New member function.
8378 * profile-count.h (profile_count_quality): Rename to ...
8379 (profile_quality): ... this one.
8380 (profile_probability): New.
8381 (profile_count): Update.
8382 * profile.c (compute_branch_probabilities): Update.
8383 * recog.c (peep2_attempt): Update.
8384 * sched-ebb.c (schedule_ebbs): Update.
8385 * sched-rgn.c (find_single_block_region): Update.
8386 (compute_dom_prob_ps): Update.
8387 (schedule_region): Update.
8388 * sel-sched-ir.c (compute_succs_info): Update.
8389 * stmt.c (struct case_node): Update.
8390 (do_jump_if_equal): Update.
8391 (get_outgoing_edge_probs): Update.
8392 (conditional_probability): Update.
8393 (emit_case_dispatch_table): Update.
8394 (expand_case): Update.
8395 (expand_sjlj_dispatch_table): Update.
8396 (emit_case_nodes): Update.
8397 * targhooks.c: Update.
8398 * tracer.c (better_p): Update.
8399 (find_best_successor): Update.
8400 * trans-mem.c (expand_transaction): Update.
8401 * tree-call-cdce.c: Update.
8402 * tree-cfg.c (gimple_split_edge): Upate.
8403 (move_sese_region_to_fn): Upate.
8404 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
8405 * tree-eh.c (lower_resx): Upate.
8406 (cleanup_empty_eh_move_lp): Upate.
8407 * tree-if-conv.c (version_loop_for_if_conversion): Update.
8408 * tree-inline.c (copy_edges_for_bb): Update.
8409 (copy_cfg_body): Update.
8410 * tree-parloops.c (gen_parallel_loop): Update.
8411 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
8412 (gimple_gen_time_profiler): Update.
8413 * tree-ssa-dce.c (remove_dead_stmt): Update.
8414 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
8415 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
8416 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
8417 (unloop_loops): Update.
8418 (try_peel_loop): Update.
8419 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
8420 * tree-ssa-loop-split.c (connect_loops): Update.
8421 (split_loop): Update.
8422 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
8423 (hoist_guard): Update.
8424 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
8425 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
8426 (value_replacement): Update.
8427 * tree-ssa-reassoc.c (branch_fixup): Update.
8428 * tree-ssa-tail-merge.c (replace_block_by): Update.
8429 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
8430 (create_edge_and_update_destination_phis): Update.
8431 (compute_path_counts): Update.
8432 (recompute_probabilities): Update.
8433 (update_joiner_offpath_counts): Update.
8434 (freqs_to_counts_path): Update.
8435 (duplicate_thread_path): Update.
8436 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
8437 (struct switch_conv_info): Update.
8438 (gen_inbound_check): Update.
8439 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
8440 (vect_do_peeling): Update.
8441 (vect_loop_versioning): Update.
8442 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
8443 (optimize_mask_stores): Update.
8444 * ubsan.c (ubsan_expand_null_ifn): Update.
8445 * value-prof.c (gimple_divmod_fixed_value): Update.
8446 (gimple_divmod_fixed_value_transform): Update.
8447 (gimple_mod_pow2): Update.
8448 (gimple_mod_pow2_value_transform): Update.
8449 (gimple_mod_subtract): Update.
8450 (gimple_mod_subtract_transform): Update.
8451 (gimple_ic): Update.
8452 (gimple_stringop_fixed_value): Update.
8453 (gimple_stringops_transform): Update.
8454 * value-prof.h: Update.
8456 2017-06-29 Carl Love <cel@us.ibm.com>
8458 * config/rs6000/rs6000-c.c: Add support for built-in functions
8459 vector signed int vec_signed (vector float);
8460 vector signed long long vec_signed (vector double);
8461 vector signed int vec_signed2 (vector double, vector double);
8462 vector signed int vec_signede (vector double);
8463 vector signed int vec_signedo (vector double);
8464 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
8465 instruction generator.
8466 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8467 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
8468 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
8470 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
8471 vunsignede_v2df): Add define_expands.
8472 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
8473 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
8474 VEC_UNSIGNEDO): Add definitions.
8475 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
8476 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
8477 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
8478 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
8479 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
8480 * config/rs6000/altivec.h (vec_signed, vec_signed2,
8481 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
8482 vec_unsignede, vec_unsignedo): Add builtin defines.
8483 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
8485 * doc/extend.texi: Update the built-in documentation file for the
8486 new built-in functions.
8488 2017-06-29 Richard Biener <rguenther@suse.de>
8490 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
8491 reduction chains to LOOP_VINFO_REDUCTIONS.
8492 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
8493 SLP reductions after processing reduction chains.
8495 2017-06-29 Nathan Sidwell <nathan@acm.org>
8497 * builtins.c (fold_builtin_FUNCTION): Use
8498 lang_hooks.decl_printable_name.
8500 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
8503 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
8504 with only one label.
8505 * stmt.c (expand_case): Assert NCASES is greater than one.
8507 2017-06-29 Richard Biener <rguenther@suse.de>
8509 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
8511 (group_case_labels): Likewise.
8512 (find_taken_edge): Push sanity checking on val to workers...
8513 (find_taken_edge_cond_expr): ... here
8514 (find_taken_edge_switch_expr): ... and here, handle cases
8515 with just a default label.
8516 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
8517 (group_case_labels): Likewise.
8518 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
8519 group_case_labels does anything cleanup the CFG again.
8521 2017-06-29 Bin Cheng <bin.cheng@arm.com>
8523 PR tree-optimization/81196
8524 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
8525 exit condition comparing two IVs.
8527 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
8529 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
8530 profile to the dummy entry at the end of the list of architectures.
8531 * config/arm/arm-cpu-cdata.h: Regenerated.
8533 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8534 Michael Collison <michael.collison@arm.com>
8537 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
8539 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
8540 (*aarch64_reg_<mode>3_minus_mask): New pattern.
8541 (*aarch64_<optab>_reg_di3_mask2): New pattern.
8542 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
8543 of shift when the shift amount is masked with constant equal to
8544 the size of the mode.
8545 * config/aarch64/predicates.md (subreg_lowpart_operator): New
8548 2017-06-29 Martin Liska <mliska@suse.cz>
8550 * config/i386/i386.opt: Change range from [1,5] to [0,5].
8552 2017-06-29 Yury Gribov <tetra2005@gmail.com>
8555 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
8558 (edge_growth_cache_entry::edge_growth_cache_entry): New
8560 (reset_edge_growth_cache): Update to use constructor.
8562 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
8564 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
8565 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
8566 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
8568 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
8570 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
8571 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
8573 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
8575 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
8576 (*-linux-uclibc*): Add t-uclibc tmake_file.
8577 * config/t-musl: New.
8578 * config/t-uclibc: New.
8580 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
8582 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
8584 (gen_comm_data): Emit architectural setting of arch_prof.
8585 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
8587 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
8588 (armv8-m.base, armv8-m.main): Likewise.
8589 * arm-protos.h (arm_build_target): Add profile field.
8590 (arch_option): Likewise.
8591 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
8593 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
8594 arm_active_target.profile.
8596 2017-06-28 Richard Biener <rguenther@suse.de>
8599 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
8600 TYPE_OVERFLOW_WRAPS.
8601 * match.pd (negate_expr_p): Likewise.
8602 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
8603 fold_build2, not fold_binary.
8605 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
8607 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
8608 Convert memory address to Pmode.
8609 (aarch64_print_operand): Assert MEM operands are always Pmode.
8611 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
8614 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
8615 Remove redundant if.
8616 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
8617 * config/arm/aarch-common-protos.h
8618 (aarch_forward_to_shift_is_not_shifted_re): Remove.
8619 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
8621 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
8624 * multiple_target.c (create_dispatcher_calls): Set the default
8625 clone to be static, not public.
8627 2017-06-28 Richard Biener <rguenther@suse.de>
8629 * tree-vect-loop.c (vectorizable_reduction): Move special
8630 cond reduction IV var creation ...
8631 (vect_create_epilog_for_reduction): ... here. Remove induction_index
8632 parameter. Use STMT_VINFO_VECTYPE.
8633 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
8636 2017-06-28 Martin Liska <mliska@suse.cz>
8639 * ipa-visibility.c (non_local_p): Handle visibility.
8641 2017-06-28 Martin Liska <mliska@suse.cz>
8644 * common.opt: Add IntegerRange to various options.
8645 * opt-functions.awk (integer_range_info): New function.
8646 * optc-gen.awk: Add integer_range_info to cl_options struct.
8647 * opts-common.c (decode_cmdline_option): Handle
8648 CL_ERR_INT_RANGE_ARG.
8649 (cmdline_handle_error): Likewise.
8650 * opts.c (print_filtered_help): Show valid interval in
8651 when --help is provided.
8652 * opts.h (struct cl_option): Add range_min and range_max fields.
8653 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
8655 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
8657 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
8658 (x * C EQ/NE y * C): New transformation.
8660 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
8662 * genmultilib (combination_space): Accept '+' in option names.
8664 2017-06-28 Martin Liska <mliska@suse.cz>
8667 * asan.c (instrument_derefs): Bail out inner references
8668 that are hard register variables.
8670 2017-06-28 Jakub Jelinek <jakub@redhat.com>
8673 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
8674 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
8676 2017-06-28 Richard Biener <rguenther@suse.de>
8678 * tree-vectorizer.h (vect_get_vec_defs): Remove.
8679 (vect_get_slp_defs): Adjust.
8680 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
8682 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
8684 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
8685 get_initial_defs_for_reduction instead of vect_get_vec_defs.
8686 (vectorizable_reduction): Adjust.
8687 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
8689 (vect_get_slp_defs): Likewise.
8690 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
8691 (vectorizable_bswap): Adjust.
8692 (vectorizable_call): Likewise.
8693 (vectorizable_conversion): Likewise.
8694 (vectorizable_assignment): Likewise.
8695 (vectorizable_shift): Likewise.
8696 (vectorizable_operation): Likewise.
8697 (vectorizable_store): Likewise.
8698 (vectorizable_condition): Likewise.
8699 (vectorizable_comparison): Likewise.
8701 2017-06-28 Michael Collison <michael.collison@arm.com>
8704 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
8706 (arm_gen_movmemqi): Removed unused variable 'i'.
8707 Convert 'for' loop into 'while' loop.
8708 (arm_expand_prologue): Remove last unnecessary set of insn.
8709 (thumb_pop): Remove unused variable 'pushed_words'.
8710 (thumb_exit): Remove last unnecessary set of regs_to_pop.
8712 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8714 * config/s390/predicates.md: Use s390_rel_address_ok_p.
8715 * config/s390/s390-protos.h: Add prototype of
8716 s390_rel_address_ok_p.
8717 * config/s390/s390.c (s390_got_symbol): New function.
8718 (s390_rel_address_ok_p): New function.
8719 (legitimize_pic_address): Use s390_rel_address_ok_p.
8720 (s390_load_got): Use s390_got_symbol.
8721 (s390_option_override): Issue error if
8722 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
8723 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
8725 * config/s390/s390.opt: New option mpic-data-is-text-relative.
8727 2017-06-27 Andrew Pinski <apinski@cavium.com>
8729 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
8730 (X * copysign (1.0, X)): New pattern.
8731 (X * copysign (1.0, -X)): New pattern.
8732 (copysign (-1.0, CST)): New pattern.
8734 2017-06-27 Joseph Myers <joseph@codesourcery.com>
8736 * genmultilib (combination_space): Remove variable.
8737 Validate reuse rules against regular expression for any sequence
8738 of multilib options in any order.
8740 2017-06-27 Michael Collison <michael.collison@arm.com>
8742 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
8743 call aarch64_split_simd_combine.
8744 * (aarch64_combine_internal<mode>): Delete pattern.
8745 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
8746 Allow register and subreg operands.
8748 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8750 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
8751 specific need, just fallback on defaults.
8752 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
8754 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8755 Olivier Hainque <hainque@adacore.com>
8757 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
8759 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
8760 targets. Pick a default if no particular attempt applied.
8761 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
8764 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8766 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
8767 (x86_64-wrs-vxworks7): Likewise.
8769 2017-06-27 Marek Polacek <polacek@redhat.com>
8772 * ubsan.c (instrument_null): Check get_base_address's result for null.
8774 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8776 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
8778 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
8780 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
8781 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
8782 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
8783 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
8785 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
8786 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
8787 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
8788 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
8789 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
8790 BUILT_IN_FEUPDATEENV): New builtins.
8791 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
8792 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
8793 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
8794 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
8796 (builtin_structptr_types): Adjust size.
8797 * tree.c (builtin_structptr_types): Add four entries.
8799 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8800 Olivier Hainque <hainque@adacore.com>
8802 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
8803 (TLS_SYM): New local macro, forcing reference to __tls__ on
8804 link command lines for VxWorks 7 RTPs, triggering initialization
8806 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
8807 OS features TLS support, true for RTPs on VxWorks 7.
8808 * config/vxworks.c (vxworks_override_options): Setup emutls
8811 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8813 * predict.c (test_prediction_value_range): Use -1U instead of -1
8814 to avoid narrowing conversion warning.
8815 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
8816 to avoid narrowing conversion warning.
8817 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
8819 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
8821 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
8823 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
8824 64bit configurations.
8825 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
8826 (SIZE_TYPE): Likewise.
8827 * config/vxworks.c (vxworks_emutls_var_fields): Use
8828 long_unsigned_type_node instead of unsigned_type_node as the offset
8829 field type, which is "pointer" mode in emutls.c.
8831 2017-06-27 Jakub Jelinek <jakub@redhat.com>
8834 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
8837 * gimple-fold.c (replace_call_with_call_and_fold): Handle
8838 gimple_vuse copying separately from gimple_vdef copying.
8840 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8842 * value-prof.c (free_hist): Remove call to memset and the enclosing if
8845 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8846 Olivier Hainque <hainque@adacore.com>
8848 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
8849 for all vxworks7 targets.
8850 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
8851 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
8852 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
8853 variations for VX6/VX7 and 32/64bits later on in ...
8854 (VXWORKS_LIB_SPEC): Leverage new macros.
8855 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
8856 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
8858 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
8860 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
8861 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
8863 2017-06-26 Carl Love <cel@us.ibm.com>
8865 * config/rs6000/rs6000-c.c: Add support for built-in functions
8866 vector bool char vec_reve (vector bool char);
8867 vector signed char vec_reve (vector signed char);
8868 vector unsigned char vec_reve (vector unsigned char);
8869 vector bool int vec_reve (vector bool int);
8870 vector signed int vec_reve (vector signed int);
8871 vector unsigned int vec_reve (vector unsigned int);
8872 vector bool long long vec_reve (vector bool long long);
8873 vector signed long long vec_reve (vector signed long long);
8874 vector unsigned long long vec_reve (vector unsigned long long);
8875 vector bool short vec_reve (vector bool short);
8876 vector signed short vec_reve (vector signed short);
8877 vector double vec_reve (vector double);
8878 vector float vec_reve (vector float);
8879 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
8880 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
8881 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
8882 (altivec_vreve): New pattern.
8883 * config/rs6000/altivec.h (vec_reve): New define.
8884 * doc/extend.texi (vec_rev): Update the built-in documentation file
8885 for the new built-in functions.
8887 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8889 PR tree-optimization/71815
8890 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
8892 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
8894 (slsr_process_phi): Likewise.
8895 (replace_uncond_cands_and_profitable_phis): Don't replace a
8896 multiply candidate with a stride of 1 (copy or cast).
8897 (phi_incr_cost): Call uses_consumed_by_stmt rather than
8899 (lowest_cost_path): Likewise.
8900 (total_savings): Likewise.
8902 2017-06-26 Richard Biener <rguenther@suse.de>
8905 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
8906 Use def_builtin_pure for all gather builtins.
8908 2017-06-26 Richard Biener <rguenther@suse.de>
8910 PR tree-optimization/81203
8911 * tree-tailcall.c (find_tail_calls): Do not move stmts into
8914 2017-06-26 Marek Polacek <polacek@redhat.com>
8917 * doc/invoke.texi: Document -Wmultistatement-macros.
8919 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
8921 * doc/sourcebuild.texi (ARM-specific attributes): Document new
8922 arm_neon_ok_no_float_abi effective target.
8924 2017-06-26 Richard Biener <rguenther@suse.de>
8926 PR tree-optimization/80928
8927 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
8928 (copy_bbs): Set BB_DUPLICATED flag early.
8929 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
8931 (execute_on_shrinking_pred): Likewise.
8932 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
8933 BB_DUPLICATED blocks.
8934 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
8935 iterate over all PHIs considering removal of *gsi.
8937 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
8939 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
8942 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8944 * config/rs6000/rs6000-string.c: (expand_block_clear,
8945 do_load_for_compare, select_block_compare_mode,
8946 compute_current_alignment, expand_block_compare,
8947 expand_strncmp_align_check, expand_strn_compare,
8948 expand_block_move, rs6000_output_load_multiple)
8949 Move functions related to string/block move/compare
8951 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
8952 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
8953 for this function which is now used in two files.
8954 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
8955 * config.gcc: Add rs6000-string.o to extra_objs for
8956 targets powerpc*-*-* and rs6000*-*-*.
8958 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
8961 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
8962 32-bit, since indexed is not valid for DImode.
8963 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
8964 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
8965 (define_peephole2 for Altivec d-form load): Add 32-bit support.
8966 (define_peephole2 for Altivec d-form store): Likewise.
8969 * multiple_target.c (create_dispatcher_calls): Only create the
8970 dispatcher call if the function is the default clone of a
8973 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
8976 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
8977 a call, force the call to not be a tail call.
8979 2017-06-23 Jeff Law <law@redhat.com>
8981 * doc/contrib.texi: Add entry for Steven Pemberton's work on
8984 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
8986 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
8987 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
8988 handling for early expansion of vector shifts (sl,sr,sra,rl).
8989 (builtin_function_type): Add vector shift right instructions
8990 to the unsigned argument list.
8992 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
8994 rtl-optimizatoin/79286
8995 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
8996 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
8997 trap. PIC register plus a const unspec without offset can never trap.
8999 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
9001 * tree.h (builtin_structptr_type): New type.
9002 (builtin_structptr_types): Declare new array.
9003 * tree.c (builtin_structptr_types): New array.
9004 (free_lang_data, build_common_tree_nodes): Use it.
9006 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
9009 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
9012 2017-06-22 Matt Turner <mattst88@gmail.com>
9014 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
9015 Lake models to skylake case. Assume skylake for unknown
9016 models with clflushopt.
9018 2017-06-22 Jeff Law <law@redhat.com>
9020 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
9021 frame sizes that do not satisfy aarch64_uimm12_shift.
9023 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
9025 * profile-count.h (apply_probability,
9026 apply_scale, probability_in): Fix checks for zero.
9028 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9030 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
9031 * doc/cppdiropts.texi (-I @var{dir}): Document it.
9033 2016-06-22 Richard Biener <rguenther@suse.de>
9035 * tree-vect-loop.c (vect_model_reduction_cost): Handle
9036 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
9037 REDUC_MAX_EXPR support.
9038 (vectorizable_reduction): Likewise.
9039 (vect_create_epilog_for_reduction): Likewise.
9041 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
9043 * match.pd (A / (1 << B) -> A >> B): New.
9044 * generic-match-head.c: Include optabs-tree.h.
9045 * gimple-match-head.c: Likewise.
9046 * optabs-tree.h (target_supports_op_p): New.
9047 * optabs-tree.c (target_supports_op_p): New.
9049 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9051 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
9052 $gcc_cv_ld --help output.
9053 (gcc_cv_ld_demangle): Likewise.
9054 (gcc_cv_ld_eh_frame_hdr): Likewise.
9055 (gcc_cv_ld_pie): Likewise.
9056 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
9057 (gcc_cv_ld_buildid): Likewise.
9058 (gcc_cv_ld_sysroot): Likewise.
9059 (ld_bndplt_support): Likewise.
9060 (ld_pushpopstate_support): Likewise.
9061 * configure: Regenerate.
9062 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
9064 2017-06-21 Jakub Jelinek <jakub@redhat.com>
9067 * config/i386/sse.md (round<mode>2): Renumber match_dup and
9068 operands indexes to avoid gap between operands and match_dups.
9070 2017-06-21 Andrew Pinski <apinski@cavium.com>
9072 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
9073 Increment Arith_shift and Arith_shift_reg by 1.
9074 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
9076 * config/aarch64/aarch64.c (thunderx_tunings): Enable
9077 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
9078 (aarch64_strip_extend): Add new argument and test for it.
9079 (aarch64_cheap_mult_shift_p): New function.
9080 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
9081 add a cost if it is true.
9082 Update calls to aarch64_strip_extend.
9083 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
9085 2017-06-21 Andrew Pinski <apinski@cavium.com>
9087 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
9089 (thunderxt88): Likewise.
9090 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
9091 (thunderx_prefetch_tune): New variable.
9092 (thunderx2t99_prefetch_tune): Update for the correct values.
9093 (thunderxt88_tunings): New variable.
9094 (thunderx_tunings): Use thunderx_prefetch_tune instead of
9095 generic_prefetch_tune.
9096 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
9098 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9100 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
9101 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
9102 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
9103 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
9104 (aarch64_atomic_cas<mode>, GPI): Likewise.
9106 2017-06-21 Martin Liska <mliska@suse.cz>
9108 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
9109 statements on cold and hot labels.
9110 * predict.c (tree_estimate_probability_bb): Remove the
9111 prediction from this place.
9113 2017-06-21 Martin Liska <mliska@suse.cz>
9115 PR tree-optimization/79489
9116 * gimplify.c (maybe_add_early_return_predict_stmt): New
9118 (gimplify_return_expr): Call the function.
9119 * predict.c (tree_estimate_probability_bb): Remove handling
9121 * predict.def: Update comment about early return predictor.
9122 * gimple-predict.h (is_gimple_predict): New function.
9123 * predict.def: Change default value of early return to 66.
9124 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
9126 * passes.def: Put pass_strip_predict_hints to the beginning of
9129 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
9131 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
9132 FUNCTION_DECL declarations.
9133 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
9135 (dwaf2out_decl): Likewise.
9136 * godump.c (go_early_global_decl): Skip call to the real debug hook
9137 for FUNCTION_DECL declarations.
9138 * passes.c (rest_of_decl_compilation): Skip call to the
9139 early_global_decl debug hook for FUNCTION_DECL declarations, unless
9140 -fdump-go-spec is passed.
9142 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
9144 * config/i386/i386.c (struct builtin_isa): New field pure_p.
9145 Reorder for compactness.
9146 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
9147 (def_builtin_pure, def_builtin_pure2): New functions.
9148 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
9150 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
9152 * match.pd (nop_convert): New predicate.
9153 ((A +- CST1) +- CST2): Allow some NOP conversions.
9155 2017-06-21 Jakub Jelinek <jakub@redhat.com>
9158 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
9159 with ctors/dtors if GOVD_SHARED is set.
9161 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9163 * config/aarch64/aarch64.md (movti_aarch64):
9164 Emit mov rather than orr.
9165 (movtf_aarch64): Likewise.
9166 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
9167 Emit mov rather than orr.
9169 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9171 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
9172 Swap alternatives, make integer dup more expensive.
9174 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
9176 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
9177 Return true for non-tls symbols.
9179 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
9181 * config/aarch64/aarch64-cores.def (cortex-a55): New.
9182 (cortex-a75): Likewise.
9183 (cortex-a75.cortex-a55): Likewise.
9184 * config/aarch64/aarch64-tune.md: Regenerate.
9185 * doc/invoke.texi (-mtune): Document new values for -mtune.
9187 2017-06-21 Tom de Vries <tom@codesourcery.com>
9189 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
9191 (Effective-Target Keywords, Other attributes): Suggest using
9192 dg-add-options stack_size feature to get stack limit in stack_size
9193 effective target documentation.
9195 2017-06-21 Julian Brown <julian@codesourcery.com>
9196 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9198 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
9199 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
9200 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
9202 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
9203 attribute type list for neon_multiply.
9204 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
9205 attribute type list for neon_multiply.
9206 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
9207 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
9208 attribute type list for neon_multiply.
9209 * config/arm/types.md (crypto_pmull): Add.
9210 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
9211 attribute type list.
9213 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
9215 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
9218 2017-06-20 Jakub Jelinek <jakub@redhat.com>
9220 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
9221 to make sure not to dereference a NULL cost_classes_ptr pointer.
9223 2017-06-20 Carl Love <cel@us.ibm.com>
9225 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9226 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9227 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9228 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9229 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9230 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
9231 VMULOSW): New enum "unspec" values.
9232 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
9233 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
9234 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
9235 altivec_vmulosw): New patterns.
9236 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9237 VMULOSW): Add definitions.
9239 2017-06-20 Julia Koval <julia.koval@intel.com>
9241 * config/i386/i386.c: Fix rounding expand for new pattern.
9242 * config/i386/subst.md: Fix pattern (parallel -> unspec).
9244 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9246 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
9247 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
9249 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9251 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
9254 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
9256 * config/aarch64/aarch64-cores.def: Rearrange to sort by
9257 architecture, then by implementer ID.
9258 * config/aarch64/aarch64-tune.md: Regenerate.
9260 2017-06-20 Richard Biener <rguenther@suse.de>
9263 * fold-const.c (split_tree): Fold to type before negating.
9265 2017-06-20 David Malcolm <dmalcolm@redhat.com>
9267 * diagnostic-show-locus.c
9268 (selftest::test_fixit_deletion_affecting_newline): New function.
9269 (selftest::diagnostic_show_locus_c_tests): Call it.
9271 2017-06-20 Andreas Schwab <schwab@suse.de>
9274 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
9277 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
9279 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
9280 __ARM_FEATURE_COPROC according to support.
9282 2017-06-20 Jakub Jelinek <jakub@redhat.com>
9284 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
9285 Rewritten to avoid overflow for > 32-bit pointers.
9288 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
9289 by removing enum keyword.
9290 (ubsan_type_descriptor): Likewise. Formatting fix.
9293 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
9294 splitter): Require TARGET_SSE2 in the condition.
9296 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
9299 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
9300 for doing vector set of SFmode on ISA 3.0.
9301 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
9302 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
9304 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
9305 SFmode value into a V4SF variable that was extracted from another
9306 V4SF variable without converting the element to double precision
9307 and back to single precision vector format.
9308 (vsx_insert_extract_v4sf_p9_2): Likewise.
9310 2017-06-19 Jakub Jelinek <jakub@redhat.com>
9312 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
9313 in UWHI to avoid undefined overflow.
9316 * ubsan.h (enum ubsan_encode_value_phase): New.
9317 (ubsan_encode_value): Change second argument to
9318 enum ubsan_encode_value_phase with default value of
9319 UBSAN_ENCODE_VALUE_GENERIC.
9320 * ubsan.c (ubsan_encode_value): Change second argument to
9321 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
9322 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
9323 create_tmp_var_raw instead of create_tmp_var and use a
9325 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
9326 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
9327 ubsan_encode_value callers.
9330 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
9331 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
9332 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
9334 2017-06-19 Richard Biener <rguenther@suse.de>
9337 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
9338 estimates if we changed anything.
9340 2017-06-19 Richard Biener <rguenther@suse.de>
9342 PR tree-optimization/80887
9343 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
9344 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
9345 simplified lookups, then reset mprts_hook.
9346 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
9348 (try_to_simplify): Likewise.
9350 2017-06-19 Martin Liska <mliska@suse.cz>
9353 * gimplify.c (gimplify_switch_expr):
9354 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
9356 2017-06-19 Martin Liska <mliska@suse.cz>
9358 * doc/install.texi: Document that PGO runs in 4 stages.
9360 2017-06-19 Martin Liska <mliska@suse.cz>
9363 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
9364 to dispatcher function name.
9365 * multiple_target.c (replace_function_decl): New function.
9366 (create_dispatcher_calls): Redirect both edges and references.
9368 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
9370 * profile-count.c (profile_count::dump): Dump quality.
9371 (profile_count::differs_from_p): Update for unsigned val.
9372 * profile-count.h (profile_count_quality): New enum.
9373 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
9375 2017-06-19 Richard Biener <rguenther@suse.de>
9377 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
9378 struct function as arg.
9379 (estimate_numbers_of_iterations): Export overload with loop arg.
9380 (free_numbers_of_iterations_estimates_loop): Use an overload of
9381 free_numbers_of_iterations_estimates instead.
9382 * tree-cfg.c (remove_bb): Adjust.
9383 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
9384 * tree-parloops.c (gen_parallel_loop): Likewise.
9385 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
9387 (tree_unroll_loops_completely): Likewise.
9388 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
9389 Use an overload instead and export.
9390 (estimated_loop_iterations): Adjust.
9391 (max_loop_iterations): Likewise.
9392 (likely_max_loop_iterations): Likewise.
9393 (estimate_numbers_of_iterations): Take struct function as arg
9395 (loop_exits_before_overflow): Adjust.
9396 (free_numbers_of_iterations_estimates_loop): Use an overload.
9397 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
9398 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
9400 2017-06-19 Richard Biener <rguenther@suse.de>
9403 * ipa-prop.c (find_constructor_constant_at_offset): Handle
9404 RANGE_EXPR conservatively.
9406 2017-06-16 Carl Love <cel@us.ibm.com>
9408 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9409 definitions for vec_float, vec_float2, vec_floato,
9410 vec_floate built-ins.
9411 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
9412 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
9414 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
9415 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
9416 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
9417 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
9418 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
9419 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
9420 vec_floato): Add builtin defines.
9421 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
9422 Update the built-in documentation file for the new built-in
9425 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9427 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
9428 (mthumb): Mark as the negative of -marm.
9430 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9432 * doc/invoke.texi (ARM Options, -mcpu): Document supported
9434 (ARM Options, -mtune): Document that this accepts the same
9435 extension options as -mcpu.
9436 (ARM Options, -mfpu): Document addition of -mfpu=auto.
9438 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9440 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
9441 permitted extensions.
9443 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9445 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
9446 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
9447 (armv8-m.main): Add option +nodsp.
9448 * config/arm/arm-cpu-cdata.h: Regenerated.
9450 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9452 * config/arm/t-fuchsia: New file.
9453 * config.gcc (arm*-*-fuchsia*): Use it.
9455 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9457 * config/arm/t-symbian: Rewrite for new option infrastructure.
9459 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9461 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
9462 (MULTILIB_REQUIRED): Likewise.
9464 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9466 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
9467 (MULTILIB_RESUE): Likewise.
9468 (MULTILIB_MATCHES): Likewise.
9469 (MULTLIB_REQUIRED): Likewise.
9471 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9473 * config/arm/t-rtems: Rewrite for new option framework.
9475 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9477 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
9478 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
9479 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
9480 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
9481 * config/arm/t-multilib: ... here.
9482 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
9483 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
9484 armv7-a and armv8*-a when A-profile libraries have not been built.
9485 * config/arm/t-rmprofile: Rewrite.
9487 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9489 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
9490 with a backslash. Remove the backslash after substituting unescaped
9492 * doc/fragments.texi (MULTILIB_REUSE): Document it.
9494 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9496 * config.gcc: (arm*-*-*): When building a-profile libraries, force
9497 the driver to pass through the default setting of -mfloat-abi.
9498 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
9500 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
9501 (all_feat_combs): New rule.
9502 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
9504 * config/arm/t-aprofile: Rewrite.
9506 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9508 * config/arm/arm.h (FPUTYPE_AUTO): Define.
9509 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
9510 fpu is not specified by the user/command-line.
9511 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
9512 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
9513 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
9514 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
9515 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
9516 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
9518 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9520 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
9521 * config/arm/t-arm-elf: Rewritten.
9523 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9525 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
9526 have some floating-point instructions.
9527 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
9528 (TARGET_MAYBE_HARD_FLOAT): New macro.
9529 * config/arm/arm-builtins.c (arm_init_builtins): Use
9530 TARGET_MAYBE_HARD_FLOAT.
9531 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
9533 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9535 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
9536 (configargs.h): Include it.
9537 (arm_print_hint_for_fpu_option): New function.
9538 (arm_parse_fpu_option): New function.
9539 (candidate_extension): New class.
9540 (arm_canon_for_multilib): New function.
9541 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
9542 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
9543 (ARCH_CANONICAL_SPECS): New macro.
9544 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
9546 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9548 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
9549 are set after handling multilib fragments. Set target_cpu_default2
9552 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9554 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
9556 (arm*-*-*): Set target_cpu_default2 to a quoted string.
9557 * config/arm/parsecpu.awk (check_cpu): Validate any extension
9559 (check_arch): Likewise.
9560 * config/arm/arm.c (arm_configure_build_target): Handle
9561 TARGET_CPU_DEFAULT being a string constant. Scan any feature
9562 options in the default.
9564 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9566 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
9567 when an option is an alias of another.
9568 * config/arm/parsecpu.awk (optalias): New parser token.
9569 (gen_comm_data): Mark non-alias options as such. Emit entries
9570 for extension aliases.
9571 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
9572 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
9573 (armv6kz, armv6zk, armv6t2): Likewise.
9574 (armv7): Make vfpv3-d16 an alias.
9575 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
9577 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
9578 Sort in canonical order.
9579 (armv8-a): Sort in canonical order.
9580 (armv8.1-a, armv8.2-a): Likewise.
9581 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
9583 (cortex-a9): Sort in canonical order.
9584 * config/arm/arm.c (selftests.h): Include it.
9585 (arm_test_cpu_arch_data): New function.
9586 (arm_run_self_tests): New function.
9587 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
9588 (targetm): Move declaration to the end of the file.
9589 * arm-cpu-cdata.h: Regenerated.
9591 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9593 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
9594 call to target_mode_check describing the type of option passed.
9595 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
9596 (arm_target_thumb_only): Use arm_parse_arch_option_name or
9597 arm_parse_cpu_option_name to match parameters against list of
9599 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
9600 arm_arch_core_flags data structure.
9601 * config/arm/arm-cpu_cdata.h: Regenerated.
9603 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9605 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
9607 (arm_print_hint_for_cpu_option): Likewise.
9608 (arm_print_hint_for_arch_option): Likewise.
9609 (arm_parse_cpu_option_name): Likewise.
9610 (arm_parse_arch_option_name): Likewise.
9611 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
9612 of entries in the all_fpus list.
9613 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
9614 (arm_parse_cpu_option_name): Declare.
9615 (arm_parse_arch_option_name): Declare.
9616 (arm_parse_option_features): Declare.
9617 (arm_intialize_isa): Declare.
9618 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
9620 (gen_comm_data): ... here. Make definitions non-static.
9621 * config/arm/arm-cpu-data.h: Regenerated.
9622 * config/arm/arm-cpu-cdata.h: Regenerated.
9624 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9626 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
9627 (cpu_arch_extension): New structure.
9628 (cpu_arch_option, arch_option, cpu_option): New structures.
9629 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
9631 (gen_data): Generate new format data tables.
9632 * config/arm/arm.c (cpu_tune): New structure.
9633 (cpu_option, processors): Delete.
9634 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
9635 (arm_print_hint_for_cpu_option): ... this and ...
9636 (arm_print_hint_for_arch_option): ... this.
9637 (arm_parse_arch_cpu_name): Delete. Replace with ...
9638 (arm_parse_cpu_option_name): ... this and ...
9639 (arm_parse_arch_option_name): ... this.
9640 (arm_unrecognized_feature): Change type of target parameter to
9642 (arm_parse_arch_cpu_features): Delete. Replace with ...
9643 (arm_parse_option_features): ... this.
9644 (arm_configure_build_target): Rework to use new configuration data
9646 (arm_print_tune_info): Rework for new configuration data tables.
9647 * config/arm/arm-cpu-data.h: Regenerated.
9648 * config/arm/arm-cpu.h: Regenerated.
9650 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9652 * Makefile.in (OBJS): Move sbitmap.o from here ...
9653 (OBJS-libcommon): ... to here.
9655 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9657 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
9658 (ISA_ALL_CRYPTO): New macro.
9659 (ISA_ALL_SIMD): New macro
9660 (ISA_ALL_FP): New macro.
9661 * config/arm/arm.c (fpu_bitlist): Update initializer.
9662 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
9664 (arm9e): Add fpu. Add option for nofp
9665 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
9666 (arm926ej-s, arm1026ej-s): Likewise.
9667 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
9668 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
9669 neon-fp16, neon-vfpv4, nofp and nosimd.
9670 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
9671 (cortex-a8): Add fpu. Add option for nofp.
9672 (cortex-a9): Add fpu. Add options for nosimd and nofp.
9673 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
9674 (cortex-r4f): Add fpu.
9675 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
9676 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
9678 (cortex-r8): Likewise.
9679 (cortex-m4): Add fpu. Add option for nofp.
9680 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
9681 (cortex-a17.cortex-a7): Likewise.
9682 (cortex-a32): Add fpu. Add options for crypto and nofp.
9683 (cortex-a35, cortex-a53): Likewise.
9684 (cortex-a57): Add fpu. Add option for crypto.
9685 (cortex-a72, cortex-a73): Likewise.
9686 (exynos-m1): Likewise.
9687 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
9688 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
9689 (cortex-m33): Add fpu. Add option for nofp.
9690 * config/arm/arm-cpu-cdata.h: Regenerated
9691 * config/arm/arm-cpu-data.h: Regenerated.
9693 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9695 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
9696 (armv5te, armv5tej): Likewise.
9697 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
9698 (armv7): Add options fp and vfpv3-d16.
9699 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
9700 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
9702 (armv7ve): Likewise.
9703 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
9704 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
9705 (armv8-a): Add nocrypto option.
9706 (armv8.1-a, armv8.2-a): Likewise.
9707 (armv8-m.main): add options fp, fp.dp and nofp.
9709 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9711 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
9713 (armv8-a+crc): Delete.
9714 (armv8.1-a): Add options simd, crypto and nofp.
9715 (armv8.2-a): Add options fp16, simd, crypto and nofp.
9716 (armv8.2-a+fp16): Delete.
9717 (armv8-m.main): Add option dsp.
9718 (armv8-m.main+dsp): Delete.
9719 (cortex-a8): Add fpu. Add nofp option.
9720 (cortex-a9): Add fpu. Add nofp and nosimd options.
9721 * config/arm/parsecpu.awk (gen_data): Generate option tables and
9722 link to main cpu and architecture data structures.
9723 (gen_comm_data): Only put isa attributes from the main architecture
9725 (option): New statement for architecture and CPU entries.
9726 * arm.c (struct cpu_option): New structure.
9727 (struct processors): Add entry for options.
9728 (arm_unrecognized_feature): New function.
9729 (arm_parse_arch_cpu_name): Ignore any characters after the first
9731 (arm_parse_arch_cpu_feature): New function.
9732 (arm_configure_build_target): Separate out any CPU and architecture
9733 features and parse separately. Don't error out if -mfpu=auto is
9734 used with only an architecture string.
9735 (arm_print_asm_arch_directives): New function.
9736 (arm_file_start): Call it.
9737 * config/arm/arm-cpu-cdata.h: Regenerated.
9738 * config/arm/arm-cpu-data.h: Likewise.
9739 * config/arm/arm-tables.opt: Likewise.
9741 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9743 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
9744 assembler when it is not -mfpu=auto.
9746 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9748 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
9749 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
9750 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
9751 (ASM_CPU_SPEC): Rewrite.
9752 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
9753 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
9754 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
9755 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
9756 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
9757 copied string is NUL-terminated. Also strip any characters prefixed
9759 (arm_rewrite_selected_arch): New function.
9760 (arm_rewrite_march): New function.
9762 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
9764 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
9765 (x_arm_cpu_string, x_arm_tune_string): Likewise.
9766 (march, mcpu, mtune): Convert to string-based options.
9767 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
9768 (arm_parse_arch_cpu_name): New function.
9769 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
9770 identify selected architecture or CPU.
9771 (arm_option_save): New function.
9772 (TARGET_OPTION_SAVE): Redefine.
9773 (arm_option_restore): Restore string options.
9774 (arm_option_print): Print string options.
9776 2017-06-16 Martin Sebor <msebor@redhat.com>
9778 PR tree-optimization/80933
9779 PR tree-optimization/80934
9780 * builtins.c (fold_builtin_3): Do not handle bcmp here.
9781 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
9782 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
9783 (gimple_fold_builtin): Call them.
9785 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9787 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
9788 as unlikely; update profile.
9790 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9792 * predict.c (force_edge_cold): Handle declaring edges impossible
9795 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9797 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
9799 (try_unroll_loop_completely): Fix reporting.
9801 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9803 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
9805 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
9808 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
9809 if given a non-constant argument for an intrinsic which requires a
9812 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9814 * profile.c (compare_freqs): New function.
9815 (branch_prob): Sort edge list.
9816 (find_spanning_tree): Assume that the list is priority sorted.
9818 2017-06-16 Richard Biener <rguenther@suse.de>
9820 PR tree-optimization/81090
9821 * passes.def (pass_record_bounds): Remove.
9822 * tree-pass.h (make_pass_record_bounds): Likewise.
9823 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
9824 make_pass_record_bounds): Likewise.
9825 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
9826 not free niter estimates at the beginning but at the end.
9827 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
9829 2017-06-16 Richard Biener <rguenther@suse.de>
9831 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
9832 initializer to workaround ICE in host GCC 4.8.
9834 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
9836 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
9838 (clone_inlined_nodes): Update.
9840 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9842 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
9843 prefetch settings, and enable prefetching by default at -O3.
9845 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9847 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9848 Set flag_prefetch_loop_arrays according to tuning data.
9850 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
9852 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
9854 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
9855 [Unrelated to main purpose of the patch] Place the pointer field last
9856 to enable type checking errors when tune structure are wrongly merged.
9857 * config/aarch64/aarch64.c (generic_prefetch_tune,)
9858 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
9859 (thunderx2t99_prefetch_tune): New tune constants.
9860 (tune_params *_tunings): Update all tunings (no functional change).
9861 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
9862 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
9863 from tunings structures.
9865 2017-06-16 Jakub Jelinek <jakub@redhat.com>
9868 * ubsan.c (instrument_null): Add T argument, use it instead
9869 of computing it based on IS_LHS.
9870 (instrument_object_size): Likewise.
9871 (pass_ubsan::execute): Adjust instrument_null and
9872 instrument_object_size callers to pass gimple_get_lhs or
9873 gimple_assign_rhs1 result to it. Use instrument_null instead of
9874 calling get_base_address and instrument_mem_ref. Handle
9875 aggregate call arguments for object-size sanitization.
9877 2017-06-16 Yury Gribov <tetra2005@gmail.com>
9879 PR tree-optimization/81089
9880 * tree-vrp.c (is_masked_range_test): Validate operands of
9883 2017-06-15 Martin Sebor <msebor@redhat.com>
9886 * dumpfile.c (dump_register): Avoid calling memset to initialize
9887 a class with a default ctor.
9888 * gcc.c (struct compiler): Remove const qualification.
9889 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
9890 * hash-table.h: Ditto.
9891 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
9893 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
9894 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
9896 * params.h (struct param_info): Make struct members non-const.
9897 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
9898 with default initialization.
9899 * vec.h (vec_copy_construct, vec_default_construct): New helper
9901 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
9902 with vec_copy_construct.
9903 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
9904 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
9905 * doc/invoke.texi (-Wclass-memaccess): Document.
9907 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9909 * emit-rtl.h (is_leaf): Update comment about local
9912 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
9915 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
9916 for a variable to have a section before checking if the section has a
9918 Set section to.persistent if persistent attribute is set.
9919 Warn if .persistent attribute is used on an automatic variable.
9921 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
9923 PR rtl-optimization/80474
9924 * reorg.c (update_block): Do not ignore instructions in a delay slot.
9926 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
9928 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
9931 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
9933 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
9934 (casesi): Emit bounds checking as RTL.
9935 (casesi_internal_mips16_<mode>): Remove bounds checking.
9937 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
9939 * config/xtensa/xtensa.c (xtensa_option_override): Append
9940 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
9941 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
9942 xtensa_doloop_hooks): Define unconditionally.
9943 (xtensa_reorg_loops): Only call reorg_loops in the presence of
9945 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
9946 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
9947 for it in xtensa_option_override.
9948 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
9949 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
9951 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
9953 * doc/cppopts.texi: Document '-' special value to -MF.
9955 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
9957 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
9958 (cortex_a53_fconst): Likewise.
9959 (cortex_a53_fpmul): Likewise.
9960 (cortex_a53_f_load_64): Likewise.
9961 (cortex_a53_f_load_many): Likewise.
9962 (cortex_a53_advsimd_alu): Likewise.
9963 (cortex_a53_advsimd_alu_q): Likewise.
9964 (cortex_a53_advsimd_mul): Likewise.
9965 (cortex_a53_advsimd_mul_q): Likewise.
9966 (fpmac bypass): Add new bypass for fpmac-fpmac case.
9967 Add missing fmul, r2f_cvt and fconst cases.
9969 2017-06-14 Richard Biener <rguenther@suse.de>
9972 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
9974 (fold_binary_loc): When associating do not treat pre-existing
9975 TREE_OVERFLOW on literal constants as a reason to allow
9976 TREE_OVERFLOW on associated literal constants.
9978 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9980 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
9981 (MASK_FEATURES): New macro.
9982 * config/sparc/sparc.c (sparc_option_override): Remove the special
9983 handling of -mfpu and generalize it to all MASK_FEATURES switches.
9985 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9987 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
9988 a division of 0 if non-call exceptions are enabled.
9990 2017-06-14 Andrew Pinski <apinski@cavium.com>
9991 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9994 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
9995 Improve vector initialization code gen for only variable case.
9997 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
9999 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
10001 2017-06-14 Richard Biener <rguenther@suse.de>
10003 PR tree-optimization/81083
10004 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
10007 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10009 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
10010 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
10011 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
10012 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
10013 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
10014 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
10016 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10018 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
10019 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
10021 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10023 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
10025 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10027 * config/rs6000/t-rtems: Don't handle SPE.
10029 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10031 * config/rs6000/t-linux: Don't handle SPE.
10033 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10035 * config/rs6000/eabispe.h: Delete file.
10037 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10039 * config/rs6000/t-spe: Delete file.
10041 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10043 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
10044 (rs6000_legitimate_offset_address_p): Return false for anything in
10045 V2SImode or V2SFmode.
10047 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
10049 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
10050 except V2SF and V2SI. Rearrange the vector modes, and add comments.
10051 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
10053 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
10054 (rs6000_legitimate_offset_address_p): Ditto.
10055 (rs6000_emit_move): Ditto.
10056 (rs6000_init_builtins): Remove V4HI_type_node.
10058 2017-06-13 Martin Liska <mliska@suse.cz>
10061 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
10062 (gate_asan): Likewise.
10063 * asan.h (asan_no_sanitize_address_p): Remove the function.
10064 (sanitize_flags_p): New function.
10065 * builtins.def: Fix coding style.
10066 * common.opt: Use renamed enum value.
10067 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
10068 * doc/extend.texi: Document no_sanitize attribute.
10069 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
10070 to SANITIZE_UNDEFINED_NONDEFAULT.
10071 * gcc.c (sanitize_spec_function): Use the renamed enum value.
10072 * gimple-fold.c (optimize_atomic_compare_exchange_p):
10073 Use sanitize_flags_p.
10074 * gimplify.c (gimplify_function_tree): Likewise.
10075 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
10076 * opts.c (parse_no_sanitize_attribute): New function.
10077 (common_handle_option): Use renamed enum value.
10078 * opts.h (parse_no_sanitize_attribute): Declare.
10079 * tree.c (sanitize_flags_p): New function.
10080 * tree.h: Declared here.
10081 * tsan.c: Use sanitize_flags_p.
10082 * ubsan.c (ubsan_expand_null_ifn): Likewise.
10083 (instrument_mem_ref): Likewise.
10084 (instrument_bool_enum_load): Likewise.
10085 (do_ubsan_in_current_function): Remove the function.
10086 (pass_ubsan::execute): Use sanitize_flags_p.
10087 * ubsan.h: Remove do_ubsan_in_current_function
10088 * tree-cfg.c (print_no_sanitize_attr_value): New function.
10089 (dump_function_to_file): Use it here.
10091 2017-06-13 Martin Jambor <mjambor@suse.cz>
10093 PR tree-optimization/80803
10094 PR tree-optimization/81063
10095 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
10096 (propagate_subaccesses_across_link): Enqueue subtree whenever
10097 necessary instead of relying on the caller.
10099 2017-06-13 Martin Jambor <mjambor@suse.cz>
10101 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
10102 that have a first_link.
10103 (sort_and_splice_var_accesses): Do not check first_link before
10105 (subtree_mark_written_and_enqueue): Likewise.
10106 (propagate_all_subaccesses): Likewise and do not stop at first
10107 parent with a first_link.
10109 2017-06-13 Martin Jambor <mjambor@suse.cz>
10111 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
10114 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10116 * match.pd: New pattern.
10118 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10120 * tree-vrp.c (is_masked_range_test): New function.
10121 (register_edge_assert_for): Determine ranges for
10124 2017-06-13 Yury Gribov <tetra2005@gmail.com>
10126 PR tree-optimization/67328
10127 * fold-const.c (maskable_range_p): New function.
10128 (build_range_check): Generate bittests if possible.
10130 2017-06-13 Martin Liska <mliska@suse.cz>
10132 * gimple-pretty-print.c (dump_probability): Add new argument.
10133 (dump_edge_probability): Dump both probability and count.
10134 (dump_gimple_label): Likewise.
10135 (dump_gimple_bb_header): Likewise.
10137 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
10140 * config/avr/avr-devices.c: Fix indentation.
10141 * config/avr/gen-avr-mmcu-specs.c: Dito.
10143 2017-06-13 Richard Biener <rguenther@suse.de>
10145 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
10146 instead get vector type from stmt_info.
10147 (vectorizable_reduction): Adjust. Remove dead code.
10149 2017-06-13 Richard Biener <rguenther@suse.de>
10151 PR middle-end/81065
10152 * fold-const.c (extract_muldiv_1): Remove bogus distribution
10153 case of C * (x * C2 + C3).
10154 (fold_addr_of_array_ref_difference): Properly fold index difference.
10156 2017-06-12 David S. Miller <davem@davemloft.net>
10159 * config/sparc/sparc.md (return expander): Emit frame blockage if
10160 function uses alloca.
10162 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
10164 * combine.c (make_field_assignment): Check len rather than the mode
10165 precision when calling force_to_mode.
10167 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
10169 Support multilibs and devices that see flash in RAM address range.
10172 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
10173 (avr_mcu_t) <flash_pm_offset>: New field.
10174 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
10175 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
10176 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
10177 (AVR_TINY_PM_OFFSET): Remove macro.
10178 * config/avr/avr.opt (-mshort-calls): New option.
10179 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
10180 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
10181 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
10182 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
10183 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
10184 instead of avr_arch->have_jmp_call.
10185 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
10186 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
10187 avr_arch->flash_pm_offset to define.
10188 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
10189 new field flash_pm_offset. Add entry for avrxmega3.
10190 (avr_texinfo): Add entry for avrxmega3.
10191 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
10192 attiny212, attiny214,
10193 attiny412, attiny414, attiny416, attiny417,
10194 attiny814, attiny816, attiny817,
10195 attiny1614, attiny1616, attiny1617,
10196 attiny3214, attiny3216, attiny3217.
10197 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
10198 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
10199 (avr_print_operand_address) [AVR_TINY]: Same.
10200 (avr_asm_init_sections) <readonly_data_section>: Only patch
10201 callback if avr_arch->flash_pm_offset = 0.
10202 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
10203 for rodata if avr_arch->flash_pm_offset != 0.
10204 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
10205 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
10206 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
10207 (m_options): Append opt_rcall.
10208 (m_dirnames): Append dir_rcall.
10209 * config/avr/t-multilib: Regenerate.
10211 * configure.ac [target=avr]: Check whether avrxmega3 default
10212 linker description file works as needed.
10213 * configure: Regenerate.
10214 * doc/avr-mmcu.texi: Regenerate.
10215 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
10216 <__AVR_ARCH__>: Document avrxmega3 and 103.
10217 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
10218 <__AVR_SHORT_CALLS__>: Document it.
10219 <__AVR_PM_BASE_ADDRESS__>: Document it.
10220 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
10221 (AVR Variable Attributes) <progmem>: Document this is
10222 not needed for avrxmega3.
10223 (AVR Named Address Spaces) <__flash>: Dito.
10225 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
10227 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
10229 2017-06-12 Doug Rupp <rupp@adacore.com>
10231 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
10232 Append vxworks-stdint.h to the tm_file list.
10233 * config/vxworks-stdint.h: New file.
10235 2017-06-12 Martin Liska <mliska@suse.cz>
10237 PR tree-optimization/81041
10238 * tree-profile.c (gimple_gen_ic_func_profiler):
10239 Create an extra BB in profile-generate
10240 (gimple_gen_time_profiler): Likewise.
10242 2017-06-12 Jakub Jelinek <jakub@redhat.com>
10244 PR tree-optimization/81003
10245 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
10246 (update_range_test): Use it instead of force_gimple_operand_gsi.
10248 2017-06-12 Richard Biener <rguenther@suse.de>
10250 PR tree-optimization/81053
10251 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
10252 with backedge value not defined in loop. Simplify def stmt
10255 2017-06-11 Tom de Vries <tom@codesourcery.com>
10258 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
10260 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
10261 nvptx_cannot_force_const_mem.
10263 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10265 * opts.c (finish_options): Move test for flag_split_stack after
10266 it has been initialized.
10268 2017-06-11 Jason Merrill <jason@redhat.com>
10270 * tree.h (id_equal): New.
10271 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
10272 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
10273 instead of strcmp of IDENTIFIER_POINTER.
10275 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10277 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
10278 (mark_all_inlined_calls_cdtor): Fix formating.
10279 (inline_transform): Rescale profile before inlining.
10281 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10283 * cgraph.h (cgraph_edge::clone): Update prototype.
10284 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
10285 (cgraph_node::create_clone): Update.
10286 (cgraph_node::create_version_clone): Update.
10287 * tree-inline.c (copy_bb): Update.
10288 (expand_call_inline): Update.
10290 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
10292 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
10293 factored out from ...
10294 (rs6000_emit_prologue): ... here.
10296 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
10298 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
10299 factored out from ...
10300 (rs6000_emit_prologue): ... here.
10302 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10304 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
10306 (handle_missing_profiles): Fix computation of tp_first_run.
10307 (counts_to_freqs): Do not touch freqs when count is 0.
10309 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
10311 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
10314 2017-06-10 Tom de Vries <tom@codesourcery.com>
10316 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
10317 attributes): Document signal effective target.
10319 2017-06-10 Tom de Vries <tom@codesourcery.com>
10321 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
10322 Document effective target stack_size.
10324 2017-06-09 David Malcolm <dmalcolm@redhat.com>
10326 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
10327 to the edit_context if they can be auto-applied.
10329 2017-06-9 Ian Lance Taylor <iant@golang.org>
10331 * opts.c (finish_options): If -fsplit-stack, disable implicit
10332 -forder-blocks-and-partition.
10333 * doc/invoke.texi (Optimize Options): Document that when using
10334 -fsplit-stack -forder-blocks-and-partition is not implicitly
10337 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10339 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
10340 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
10341 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
10342 * builtins.def (abort, trap, unreachable): Declare cold.
10343 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
10344 * tree-core.h (ECF_COLD): New.
10345 * tree.c (set_call_expr_flags): Handle ECF_COLD.
10346 (build_common_builtin_nodes): Mark unreachable and abort as cold.
10348 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10350 * predict.c (unlikely_executed_stmt_p): Cleanup.
10352 2017-06-09 Richard Biener <rguenther@suse.de>
10354 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
10355 model if the ref is always written to.
10357 2017-06-09 Tamar Christina <tamar.christina@arm.com>
10359 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
10361 2017-06-09 Tamar Christina <tamar.christina@arm.com>
10363 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
10366 2017-06-09 Tom de Vries <tom@codesourcery.com>
10369 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
10370 "target cannot support label values" when encountering LABEL_REF.
10372 2017-06-09 Martin Liska <mliska@suse.cz>
10374 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
10375 (gimple_gen_ic_func_profiler): Emit direct comparison
10376 of __gcov_indirect_call_callee with NULL.
10377 (gimple_gen_time_profiler): Change probability from
10378 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
10380 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10382 * profile.c (edge_gcov_counts): Turn to pointer.
10383 (compute_branch_probabilities, compute_branch_probabilities): Update.
10384 (branch_prob): Do not clear edge_gcov_count.
10385 * profile.h (edge_gcov_counts): Turn to pointer.
10386 (edge_gcov_count): Update.
10388 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
10390 * gimple.h (gimple_check_failed): Mark cold.
10392 2017-06-09 Richard Biener <rguenther@suse.de>
10394 PR tree-optimization/66623
10395 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
10396 refactor check_reduction into two parts, properly computing
10397 whether we have to check reduction validity for outer loop
10400 2017-06-09 Richard Biener <rguenther@suse.de>
10402 PR tree-optimization/79483
10403 * graphite-scop-detection.c (order): New global.
10404 (get_order): Compute bb to order mapping that satisfies code
10405 generation constraints.
10406 (cmp_pbbs): New helper.
10407 (build_scops): Start domwalk at entry block, sort generated
10410 2017-06-09 Richard Biener <rguenther@suse.de>
10412 PR middle-end/81007
10413 * ipa-polymorphic-call.c
10414 (ipa_polymorphic_call_context::restrict_to_inner_class):
10415 Skip FIELD_DECLs with error_mark_node type.
10416 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
10419 2017-06-09 Martin Liska <mliska@suse.cz>
10421 * predict.c (struct branch_predictor): New struct.
10422 (test_prediction_value_range): New test.
10423 (predict_c_tests): New function.
10424 * selftest-run-tests.c (selftest::run_tests): Run the function.
10425 * selftest.h: Declare new tests.
10427 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
10430 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
10431 gen_add3_insn did not fail.
10432 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
10433 r0, construct that number in a temporary reg and add that reg to r0.
10434 If asked to put the result in r0 as well, fail.
10436 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
10438 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
10439 for early expansion of vec_eqv.
10441 2017-06-08 Jakub Jelinek <jakub@redhat.com>
10443 PR middle-end/81005
10444 * ubsan.c (instrument_null): Avoid pointless code temporary.
10445 (pass_ubsan::execute): Instrument aggregate arguments of calls.
10447 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
10451 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
10454 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
10457 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10459 * predict.c (unlikely_executed_edge_p): Move ahead.
10460 (probably_never_executed_edge_p): Use it.
10462 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10464 PR middle-end/79988
10465 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
10466 gimple_call_builtin_p call.
10468 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10470 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
10471 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
10472 rtl_check_failed_type2, rtl_check_failed_code1,
10473 rtl_check_failed_code2, rtl_check_failed_code_mode,
10474 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
10475 rtvec_check_failed_bounds, rtl_check_failed_flag,
10476 _fatal_insn_not_found, _fatal_insn): Likewise.
10477 * tree.h (tree_contains_struct_check_failed,
10478 tree_check_failed, tree_not_check_failed,
10479 tree_class_check_failed, tree_range_check_failed,
10480 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
10481 tree_vec_elt_check_failed, phi_node_elt_check_failed,
10482 tree_operand_check_failed, omp_clause_check_failed,
10483 omp_clause_operand_check_failed, omp_clause_range_check_failed):
10486 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10488 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
10489 flag_branch_probabilities.
10490 * ipa-inline.c (edge_badness): Likewise.
10491 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
10492 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
10493 * predict.c (maybe_hot_frequency_p): Likewise.
10494 (probably_never_executed): Likewise.
10495 * sched-ebb.c (schedule_ebbs): Likewise.
10496 * sched-rgn.c (find_single_block_region): Likewise.
10497 * tracer.c (tail_duplicate): Likewise.
10499 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10501 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
10502 longer requires x_flag_profile_use.
10504 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10506 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
10507 instead of flag_reorder_blocks_and_partition.
10508 * dbxout.c (dbxout_function_end): Likewise.
10509 * dwarf2out.c (gen_subprogram_die): Likewise.
10510 * haifa-sched.c (sched_create_recovery_edges): Likewise.
10511 * hw-doloop.c (reorg_loops): Likewise.
10512 * varasm.c (assemble_start_function,
10513 assemble_end_function): Likewise.
10514 (decide_function_section): Do not check for
10515 flag_reorder_blocks_and_partition.
10517 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
10519 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
10521 (chkp_get_hard_register_fake_addr_expr): Ditto.
10522 (chkp_build_addr_expr): Add check for hard reg case.
10523 (chkp_parse_array_and_component_ref): Ditto.
10524 (chkp_find_bounds_1): Ditto.
10525 (chkp_process_stmt): Don't generate bounds store for
10528 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10530 * predict.c (maybe_hot_bb_p): Do not check profile status.
10531 (maybe_hot_edge_p): Likewise.
10532 (probably_never_executed): Check for zero counts even if profile
10534 (unlikely_executed_edge_p): New function.
10535 (unlikely_executed_stmt_p): New function.
10536 (unlikely_executed_bb_p): New function.
10537 (set_even_probabilities): Use unlikely predicates.
10538 (combine_predictions_for_bb): Likewise.
10539 (predict_paths_for_bb): Likewise.
10540 (predict_paths_leading_to_edge): Likewise.
10541 (determine_unlikely_bbs): New function.
10542 (estimate_bb_frequencies): Use it.
10543 (compute_function_frequency): Use zero counts even if profile is
10545 * profile-count.h: Fix typo.
10547 2017-08-08 Julia Koval <julia.koval@intel.com>
10549 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
10550 _mm512_mask_cvtsepi16_storeu_epi8,
10551 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
10552 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
10553 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
10554 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
10555 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
10556 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
10557 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
10558 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
10559 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
10560 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
10561 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
10562 __builtin_ia32_pmovuswb256mem_mask,
10563 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
10564 __builtin_ia32_pmovwb512mem_mask): New builtins.
10566 2017-08-08 Julia Koval <julia.koval@intel.com>
10568 PR target/73350,80862
10569 * config/i386/subst.md (round): Fix round pattern.
10570 * config/i386/i386.c (ix86_erase_embedded_rounding):
10571 Fix erasing rounding for the fixed pattern.
10573 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
10575 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
10577 2017-06-08 Martin Liska <mliska@suse.cz>
10579 PR gcov-profile/80911
10580 * gcov.c (block_info::block_info): New constructor.
10582 2017-06-07 Carl Love <cel@us.ibm.com>
10584 * config/rs6000/rs6000-c: The return type of the following
10585 built-in functions was implemented as int not long long. Fix sign
10586 of return value for the unsigned version of vec_mulo and vec_mule.
10587 vector unsigned long long vec_bperm (vector unsigned long long,
10588 vector unsigned char)
10589 vector signed long long vec_mule (vector signed int,
10591 vector unsigned long long vec_mule (vector unsigned int,
10592 vector unsigned int)
10593 vector signed long long vec_mulo (vector signed int,
10595 vector unsigned long long vec_mulo (vector unsigned int,
10596 vector unsigned int)
10597 * doc/extend.texi: Fix the documentation for the built-in
10600 2017-06-07 Carl Love <cel@us.ibm.com>
10603 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
10606 2017-06-07 Carl Love <cel@us.ibm.com>
10608 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
10609 support, Generate doublehv for signed int/float for BE case only.
10611 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
10613 * doc/invoke.texi (mcx16): Rewrite.
10615 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10617 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
10618 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
10619 *mov<mode>_softfloat, and an anonymous splitter): Use
10620 nonimmediate_operand instead of rs6000_nonimmediate_operand.
10622 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10624 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
10626 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
10627 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
10628 (rs6000_debug_reg_global): Adjust.
10629 (rs6000_init_hard_regno_mode_ok): Adjust.
10630 (rs6000_dbx_register_number): Adjust.
10631 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
10632 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
10633 Remove SPE_ACC and SPEFSCR.
10634 (REG_ALLOC_ORDER): Ditto.
10635 (FRAME_POINTER_REGNUM): Change to 111.
10636 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
10637 (REG_CLASS_NAMES): Ditto.
10638 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
10639 (REGISTER_NAMES): Ditto.
10640 (ADDITIONAL_REG_NAMES): Ditto.
10641 (rs6000_reg_names): Ditto.
10642 * config/rs6000/rs6000.md: Renumber some register number
10645 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10647 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
10649 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
10650 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
10652 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
10653 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
10654 Delete the SPE high registers.
10655 (REG_ALLOC_ORDER): Ditto.
10656 (enum reg_class): Remove SPE_HIGH_REGS.
10657 (REG_CLASS_NAMES): Ditto.
10658 (REG_CLASS_CONTENTS): Delete the SPE high registers.
10659 (REGISTER_NAMES): Ditto.
10660 (rs6000_reg_names): Ditto.
10661 * doc/tm.texi.in: Remove SPE as example.
10662 * doc/tm.texi: Regenerate.
10664 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10666 * config/rs6000/8540.md (ppc8540_brinc): Delete.
10667 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
10668 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
10669 * config/rs6000/rs6000.md (type): Remove "brinc".
10671 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10673 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
10674 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
10675 * config/rs6000/linuxspe.h: Delete file.
10676 * config/rs6000/rs6000.md: Don't include spe.md.
10677 * config/rs6000/spe.h: Delete file.
10678 * config/rs6000/spe.md: Delete file.
10679 * config/rs6000/t-rs6000: Remove spe.md.
10681 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10683 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
10684 (reg_or_none500mem_operand): Delete.
10685 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
10686 instead of reg_or_none500mem_operand.
10688 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10690 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
10691 handling of SPE flags.
10692 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
10694 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10696 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
10698 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
10699 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
10700 paired_divv2sf3): Similar.
10701 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
10702 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
10703 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
10705 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
10706 Rename the paired_* instruction patterns.
10707 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
10709 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
10710 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
10711 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
10712 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
10713 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
10714 PAIRED_VECTOR_MODE.
10715 (struct machine_function): Delete field spe_insn_chain_scanned_p.
10716 (spe_func_has_64bit_regs_p): Delete.
10717 (spe_expand_predicate_builtin): Delete.
10718 (spe_expand_evsel_builtin): Delete.
10719 (TARGET_DWARF_REGISTER_SPAN): Do not define.
10720 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
10721 (invalid_e500_subreg): Delete.
10722 (rs6000_legitimize_address): Always force_reg op2 as well, for
10723 paired single memory accesses.
10724 (rs6000_member_type_forces_blk): Delete.
10725 (rs6000_spe_function_arg): Delete.
10726 (rs6000_expand_unop_builtin): Delete SPE handling.
10727 (rs6000_expand_binop_builtin): Ditto.
10728 (spe_expand_stv_builtin): Delete.
10729 (bdesc_2arg_spe): Delete.
10730 (spe_expand_builtin): Delete.
10731 (spe_expand_predicate_builtin): Delete.
10732 (spe_expand_evsel_builtin): Delete.
10733 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
10734 (spe_init_builtins): Delete.
10735 (spe_func_has_64bit_regs_p): Delete.
10736 (savres_routine_name): Delete "info" parameter. Adjust callers.
10737 (rs6000_emit_stack_reset): Ditto.
10738 (rs6000_dwarf_register_span): Delete.
10739 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
10740 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
10741 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
10743 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
10745 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
10746 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
10747 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
10748 mulv2sf3, divv2sf3): Delete expanders.
10750 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10752 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
10754 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10756 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
10757 * config/rs6000/rs6000.c: Ditto.
10759 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10761 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
10762 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
10763 comparison_operator.
10765 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10767 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
10768 * config/rs6000/rs6000.opt: Ditto.
10769 * config/rs6000/t-rtems: Ditto.
10771 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10773 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
10774 TARGET_E500_SINGLE by 0, simplify.
10775 * config/rs6000/rs6000.c: Ditto.
10776 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
10777 (spe_build_register_parallel): Delete.
10778 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
10779 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
10780 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
10781 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
10782 (E500_CONVERT): Delete.
10783 * config/rs6000/spe.md: Remove many patterns and all define_constants.
10785 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
10787 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
10788 * config/rs6000/dfp.md: Ditto.
10789 (negdd2, *negdd2_fpr): Merge.
10790 (absdd2, *absdd2_fpr): Merge.
10791 (negtd2, *negtd2_fpr): Merge.
10792 (abstd2, *abstd2_fpr): Merge.
10793 * config/rs6000/e500.h: Delete file.
10794 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
10795 TARGET_FPRS by 1 and simplify.
10796 * config/rs6000/rs6000-c.c: Ditto.
10797 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
10798 TARGET_DF_SPE by 0.
10799 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
10801 * config/rs6000/rs6000.md: Ditto.
10802 (floatdidf2, *floatdidf2_fpr): Merge.
10803 (move_from_CR_gt_bit): Delete.
10804 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
10805 (E500_CR_IOR_COMPARE): Delete.
10806 (All patterns that require !TARGET_FPRS): Delete.
10807 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
10809 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10811 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
10813 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10815 * graphds.c (add_edge): Intitialize edge's attached data.
10816 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
10817 pointer parameter. Call pointed function on each edge during
10818 graph traversing. Skip traversing the edge when the function
10820 (graphds_dfs, graphds_scc): Ditto.
10821 (for_each_edge): New parameter. Pass the new parameter to callback
10823 * graphds.h (skip_edge_callback): New function pointer type.
10824 (graphds_dfs, graphds_scc): New function pointer parameter.
10825 (graphds_edge_callback, for_each_edge): New parameter.
10827 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10829 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
10830 out code checking if runtime alias check is possible to below ...
10831 Call the new function.
10832 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
10833 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
10835 2017-06-07 Marek Polacek <polacek@redhat.com>
10838 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
10839 TYPE_OVERFLOW_WRAPS check.
10841 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10843 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
10844 if versioning is required.
10845 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
10846 peeling with the check for versioning.
10848 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10850 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
10851 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
10852 Set true to new parameter if new ssa variable is defined.
10853 (vect_gen_vector_loop_niters): Refactor. Set range information
10854 for the new vector loop bound variable.
10855 (vect_do_peeling): Ditto.
10857 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10859 * tree-affine.c (ssa.h): Include header file.
10860 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
10861 has wrapping overflow behavior.
10863 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10865 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
10867 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10869 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
10870 (tree_to_aff_combination): ... here.
10872 2017-06-07 Bin Cheng <bin.cheng@arm.com>
10874 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
10875 reg_pressure model function.
10876 (ivopts_global_cost_for_size): Delete.
10877 (determine_set_costs, iv_ca_recount_cost): Call new model function
10878 ivopts_estimate_reg_pressure.
10880 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10882 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
10883 expensive than udiv. Remove floating point cases from mod.
10885 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10887 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
10888 Increase idiv cost.
10890 2017-06-07 Tamar Christina <tamar.christina@arm.com>
10892 * config/aarch64/aarch64.md
10893 (copysignsf3): Fix mask generation.
10895 2017-06-07 Jakub Jelinek <jakub@redhat.com>
10897 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
10899 (class dump_manager): Add register_dumps method.
10900 * dumpfile.c: Include langhooks.h.
10901 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
10902 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
10903 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
10904 (dump_manager::dump_register): Start with 512 entries instead of 32.
10905 (dump_manager::register_dumps): New method.
10906 * toplev.c (general_init): Instead of invoking register_dumps
10907 langhook, invoke register_dumps method on the dump manager.
10908 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
10911 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
10913 * doc/md.texi: Clarify the restrictions on a define_insn condition.
10914 Say that # requires an associated define_split to exist, and that
10915 the define_split must be suitable for use after register allocation.
10917 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
10919 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
10920 (compute_outgoing_frequencies): Also initialize zero counts.
10921 (find_many_sub_basic_blocks): Do not produce uninitialized profile
10922 around loops; preserve more of profile when nothing changes.
10924 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
10926 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
10928 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
10929 * config/arm/arm-cpu-cdata.h: Regenerate.
10930 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
10931 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
10932 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
10933 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
10934 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
10936 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
10938 * config/arm/t-rmprofile: Likewise.
10939 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
10941 2017-06-06 David S. Miller <davem@davemloft.net>
10944 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
10945 blockage if function uses alloca.
10947 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10949 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
10950 New "uid" fields to hold pretty-print IDs of group and ref.
10951 Memory references are now identified as <group_id>:<ref_id>
10952 instead of using [random] addresses.
10953 (dump_mem_details): Simplify, no functional change.
10954 (dump_mem_ref): Simplify and make output more concise.
10955 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
10956 (find_or_create_group): Initialize group uid.
10957 (record_ref): Initialize ref uid. Improve debug output.
10958 (prune_group_by_reuse, should_issue_prefetch_p,)
10959 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
10960 (mark_nontemporal_store, determine_loop_nest_reuse):
10961 Improve debug output.
10963 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
10965 * dbgcnt.def (prefetch): New debug counter.
10966 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
10967 (schedule_prefetches): Stop issueing prefetches if debug counter
10970 2017-06-06 Tom de Vries <tom@codesourcery.com>
10972 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
10973 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
10975 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10977 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
10978 Use aarch64_reg_or_zero predicate for operand 4.
10979 (aarch64_compare_and_swap<mode> define_insn_and_split):
10980 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
10981 (aarch64_store_exclusive<mode>): Likewise for operand 2.
10983 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
10985 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
10986 (arm_compute_save_core_reg_mask): This.
10987 (thumb1_compute_save_reg_mask): Rename into ...
10988 (thumb1_compute_save_core_reg_mask): This.
10989 (arm_compute_save_reg0_reg12_mask): Adapt comment.
10990 (arm_compute_frame_layout): Likewise.
10992 2017-06-06 Richard Biener <rguenther@suse.de>
10994 PR tree-optimization/80974
10995 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
10996 keep or clear leaders SSA info.
10998 2017-06-06 Tom de Vries <tom@codesourcery.com>
11000 * config/nvptx/nvptx.c (split_mode_p): New function.
11001 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
11003 2017-06-06 Tom de Vries <tom@codesourcery.com>
11005 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
11007 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11010 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
11013 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11015 * shrink-wrap.c (handle_simple_exit): Update profile.
11016 (try_shrink_wrapping): Upate profile.
11018 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11020 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
11021 (tree_guess_outgoing_edge_probabilities): New.
11022 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
11023 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
11025 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11027 * ipa-split.c (split_function): Initialize return bb profile.
11029 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
11031 * profile.c (compute_branch_probabilities): Also initialize
11032 EXIT_BLOCK profile.
11034 2017-06-06 Richard Biener <rguenther@suse.de>
11036 PR tree-optimization/80928
11037 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
11038 (vect_analyze_loop_operations): Properly guard analysis for
11040 (vect_transform_loop): Likewise.
11041 (vect_analyze_loop_2): Also reset SLP type on PHIs.
11042 (vect_model_induction_cost): Do not cost for pure SLP.
11043 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
11044 of induction in inner loop vectorization.
11045 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
11046 (vect_get_and_check_slp_defs): Handle vect_induction_def.
11047 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
11049 (vect_analyze_slp_cost_1): Cost induction.
11050 (vect_detect_hybrid_slp_stmts): Handle PHIs.
11051 (vect_get_slp_vect_defs): Likewise.
11052 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
11053 (vect_transform_stmt): Handle SLP reductions.
11054 * tree-vectorizer.h (vectorizable_induction): Adjust.
11056 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
11058 * config/rs6000/rs6000.c (make_resolver_func): Update
11059 init_lowered_empty_function call.
11061 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
11063 * doc/invoke.texi: Document the -fprofile-abs-path option.
11064 * common.opt (fprofile-abs-path): New option.
11065 * gcov-io.h (gcov_write_filename): Declare.
11066 * gcov-io.c (gcov_write_filename): New function.
11067 * coverage.c (coverage_begin_function): Use gcov_write_filename.
11068 * profile.c (output_location): Likewise.
11070 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11072 * shring-wrap.c: Revert accidental commit.
11074 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
11076 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
11078 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11080 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
11082 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
11083 profile in callgraph edge.
11084 * profile-count.h (apply_probability): If THIS is 0, then result is 0
11085 (apply_scale): Likewise.
11086 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
11087 Also scale profile when inlining function with zero profile.
11088 (initialize_cfun): Update exit block profile even when it is zero.
11089 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
11090 when profile is read.
11092 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
11094 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
11095 (CLONE_*): New constants to define the processors we can generate
11096 code for with the target_clone attribute.
11097 (rs6000_clone_map): New array to identify which clone processors
11098 the current program is running on.
11099 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
11100 target_clone attribute.
11101 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
11102 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
11103 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
11104 (cpu_expand_builtin): Add support for target_clone attribute.
11105 (rs6000_valid_attribute_p): Allow "default" attribute.
11106 (get_decl_name): New debug function to simplify printing the
11107 current function name in debugging statements.
11108 (rs6000_clone_priority): New functions to support the target_clone
11109 attribute, and be able to generate code to switch between ISA 2.05
11110 through ISA 3.0 (power6 through power9).
11111 (rs6000_compare_version_priority): Likewise.
11112 (rs6000_get_function_versions_dispatcher): Likewise.
11113 (make_resolver_func): Likewise.
11114 (add_condition_to_bb): Likewise.
11115 (dispatch_function_versions): Likewise.
11116 (rs6000_generate_version_dispatcher_body): Likewise.
11117 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
11118 (fusion_gpr_load_p): Fix a spacing issue.
11119 * doc/extend.texi (Common Function Attributes): Document that the
11120 PowerPC supports the target_clone attribute.
11122 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
11124 * config/arm/arm.h: explain F symbol found in description of ARM
11125 register allocation in its legend.
11127 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11129 * config/mips/frame-header-opt.c: Include profile-count.h.
11130 * config/riscv/riscv.c: Include profile-count.h
11132 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
11134 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
11136 (sm_set_flag_if_changed): Add bbs field.
11137 (execute_sm_if_changed_flag_set): Pass BBS.
11138 (execute_sm): Update.
11140 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11142 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
11145 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11147 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
11148 (peephole2): New peephole2 to emit the above.
11149 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
11151 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11153 * config/aarch64/aarch64.c (define_peephole2 above
11154 *sub_<shift>_<mode>): New peephole.
11156 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
11158 * config/i386/i386.c (make_resolver_func): Update.
11159 * Makefile.in: Add profile-count.h and profile-count.o
11160 * auto-profile.c (afdo_indirect_call): Update to new API.
11161 (afdo_set_bb_count): Update.
11162 (afdo_propagate_edge): Update.
11163 (afdo_propagate_circuit): Update.
11164 (afdo_calculate_branch_prob): Update.
11165 (afdo_annotate_cfg): Update.
11166 * basic-block.h: Include profile-count.h
11167 (struct edge_def): Turn count to profile_count.
11168 (struct basic_block_def): Likewie.
11169 (REG_BR_PROB_BASE): Move to profile-count.h
11170 (RDIV): Move to profile-count.h
11171 * bb-reorder.c (max_entry_count): Turn to profile_count.
11172 (find_traces): Update.
11173 (rotate_loop):Update.
11174 (connect_traces):Update.
11175 (sanitize_hot_paths):Update.
11176 * bt-load.c (migrate_btr_defs): Update.
11177 * cfg.c (RDIV): Remove.
11178 (init_flow): Use alloc_block.
11179 (alloc_block): Uninitialize count.
11180 (unchecked_make_edge): Uninitialize count.
11181 (check_bb_profile): Update.
11182 (dump_edge_info): Update.
11183 (dump_bb_info): Update.
11184 (update_bb_profile_for_threading): Update.
11185 (scale_bbs_frequencies_int): Update.
11186 (scale_bbs_frequencies_gcov_type): Update.
11187 (scale_bbs_frequencies_profile_count): New.
11188 * cfg.h (update_bb_profile_for_threading): Update.
11189 (scale_bbs_frequencies_profile_count): Declare.
11190 * cfgbuild.c (compute_outgoing_frequencies): Update.
11191 (find_many_sub_basic_blocks): Update.
11192 * cfgcleanup.c (try_forward_edges): Update.
11193 (try_crossjump_to_edge): Update.
11194 * cfgexpand.c (expand_gimple_tailcall): Update.
11195 (construct_exit_block): Update.
11196 * cfghooks.c (verify_flow_info): Update.
11197 (dump_bb_for_graph): Update.
11198 (split_edge): Update.
11199 (make_forwarder_block): Update.
11200 (duplicate_block): Update.
11201 (account_profile_record): Update.
11202 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
11203 (get_estimated_loop_iterations): Update.
11204 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
11205 (single_likely_exit): Update.
11206 * cfgloopmanip.c (scale_loop_profile): Update.
11208 (set_zero_probability): Update.
11209 (lv_adjust_loop_entry_edge): Update.
11210 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
11211 (purge_dead_edges): Update.
11212 (rtl_account_profile_record): Update.
11213 * cgraph.c (cgraph_node::create): Uninitialize count.
11214 (symbol_table::create_edge): Uninitialize count.
11215 (cgraph_update_edges_for_call_stmt_node): Update.
11216 (cgraph_edge::dump_edge_flags): Update.
11217 (cgraph_node::dump): Update.
11218 (cgraph_edge::maybe_hot_p): Update.
11219 * cgraph.h: Include profile-count.h
11220 (create_clone), create_edge, create_indirect_edge): Update.
11221 (cgraph_node): Turn count to profile_count.
11222 (cgraph_edge0: Likewise.
11223 (make_speculative, clone): Update.
11224 (create_edge): Update.
11225 (init_lowered_empty_function): Update.
11226 * cgraphclones.c (cgraph_edge::clone): Update.
11227 (duplicate_thunk_for_node): Update.
11228 (cgraph_node::create_clone): Update.
11229 * cgraphunit.c (cgraph_node::analyze): Update.
11230 (cgraph_node::expand_thunk): Update.
11231 * final.c (dump_basic_block_info): Update.
11232 * gimple-streamer-in.c (input_bb): Update.
11233 * gimple-streamer-out.c (output_bb): Update.
11234 * graphite.c (print_global_statistics): Update.
11235 (print_graphite_scop_statistics): Update.
11236 * hsa-brig.c: Include basic-block.h.
11237 * hsa-dump.c: Include basic-block.h.
11238 * hsa-gen.c (T sum_slice): Update.
11239 (convert_switch_statements):Update.
11240 * hsa-regalloc.c: Include basic-block.h.
11241 * ipa-chkp.c (chkp_produce_thunks): Update.
11242 * ipa-cp.c (struct caller_statistics): Update.
11243 (init_caller_stats): Update.
11244 (gather_caller_stats): Update.
11245 (ipcp_cloning_candidate_p): Update.
11246 (good_cloning_opportunity_p): Update.
11247 (get_info_about_necessary_edges): Update.
11248 (dump_profile_updates): Update.
11249 (update_profiling_info): Update.
11250 (update_specialized_profile): Update.
11251 (perhaps_add_new_callers): Update.
11252 (decide_about_value): Update.
11253 (ipa_cp_c_finalize): Update.
11254 * ipa-devirt.c (struct odr_type_warn_count): Update.
11255 (struct decl_warn_count): Update.
11256 (struct final_warning_record): Update.
11257 (possible_polymorphic_call_targets): Update.
11258 (ipa_devirt): Update.
11259 * ipa-fnsummary.c (redirect_to_unreachable): Update.
11260 * ipa-icf.c (sem_function::merge): Update.
11261 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
11262 * ipa-inline.c (compute_uninlined_call_time): Update.
11263 (compute_inlined_call_time): Update.
11264 (want_inline_small_function_p): Update.
11265 (want_inline_self_recursive_call_p): Update.
11266 (edge_badness): Update.
11267 (lookup_recursive_calls): Update.
11268 (recursive_inlining): Update.
11269 (inline_small_functions): Update.
11270 (dump_overall_stats): Update.
11271 (dump_inline_stats): Update.
11272 * ipa-profile.c (ipa_profile_generate_summary): Update.
11273 (ipa_propagate_frequency): Update.
11274 (ipa_profile): Update.
11275 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
11276 * ipa-utils.c (ipa_merge_profiles): Update.
11277 * loop-doloop.c (doloop_modify): Update.
11278 * loop-unroll.c (report_unroll): Update.
11279 (unroll_loop_runtime_iterations): Update.
11280 * lto-cgraph.c (lto_output_edge): Update.
11281 (lto_output_node): Update.
11282 (input_node): Update.
11283 (input_edge): Update.
11284 (merge_profile_summaries): Update.
11285 * lto-streamer-in.c (input_cfg): Update.
11286 * lto-streamer-out.c (output_cfg): Update.
11287 * mcf.c (create_fixup_graph): Update.
11288 (adjust_cfg_counts): Update.
11289 (sum_edge_counts): Update.
11290 * modulo-sched.c (sms_schedule): Update.
11291 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
11292 * predict.c (maybe_hot_count_p): Update.
11293 (probably_never_executed): Update.
11294 (dump_prediction): Update.
11295 (combine_predictions_for_bb): Update.
11296 (propagate_freq): Update.
11297 (handle_missing_profiles): Update.
11298 (counts_to_freqs): Update.
11299 (rebuild_frequencies): Update.
11300 (force_edge_cold): Update.
11301 * predict.h: Include profile-count.h
11302 (maybe_hot_count_p, counts_to_freqs): UPdate.
11303 * print-rtl-function.c: Do not include cfg.h
11304 * print-rtl.c: Include basic-block.h
11305 * profile-count.c: New file.
11306 * profile-count.h: New file.
11307 * profile.c (is_edge_inconsistent): Update.
11308 (correct_negative_edge_counts): Update.
11309 (is_inconsistent): Update.
11310 (set_bb_counts): Update.
11311 (read_profile_edge_counts): Update.
11312 (compute_frequency_overlap): Update.
11313 (compute_branch_probabilities): Update; Initialize and deinitialize
11315 (branch_prob): Update.
11316 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
11317 (edge_gcov_count): New.
11318 (bb_gcov_count): New.
11319 * shrink-wrap.c (try_shrink_wrapping): Update.
11320 * tracer.c (better_p): Update.
11321 * trans-mem.c (expand_transaction): Update.
11322 (ipa_tm_insert_irr_call): Update.
11323 (ipa_tm_insert_gettmclone_call): Update.
11324 * tree-call-cdce.c: Update.
11325 * tree-cfg.c (gimple_duplicate_sese_region): Update.
11326 (gimple_duplicate_sese_tail): Update.
11327 (gimple_account_profile_record): Update.
11328 (execute_fixup_cfg): Update.
11329 * tree-inline.c (copy_bb): Update.
11330 (copy_edges_for_bb): Update.
11331 (initialize_cfun): Update.
11332 (freqs_to_counts): Update.
11333 (copy_cfg_body): Update.
11334 (expand_call_inline): Update.
11335 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
11336 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
11337 (try_unroll_loop_completely): Update.
11338 (try_peel_loop): Update.
11339 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
11340 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
11341 * tree-ssa-loop-split.c (connect_loops): Update.
11342 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
11343 * tree-ssa-reassoc.c (branch_fixup): Update.
11344 * tree-ssa-tail-merge.c (replace_block_by): Update.
11345 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
11346 (compute_path_counts): Update.
11347 (update_profile): Update.
11348 (recompute_probabilities): Update.
11349 (update_joiner_offpath_counts): Update.
11350 (estimated_freqs_path): Update.
11351 (freqs_to_counts_path): Update.
11352 (clear_counts_path): Update.
11353 (ssa_fix_duplicate_block_edges): Update.
11354 (duplicate_thread_path): Update.
11355 * tree-switch-conversion.c (case_bit_test_cmp): Update.
11356 (struct switch_conv_info): Update.
11357 * tree-tailcall.c (decrease_profile): Update.
11358 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
11359 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
11360 * value-prof.c (check_counter): Update.
11361 (gimple_divmod_fixed_value): Update.
11362 (gimple_mod_pow2): Update.
11363 (gimple_mod_subtract): Update.
11364 (gimple_ic_transform): Update.
11365 (gimple_stringop_fixed_value): Update.
11366 * value-prof.h (gimple_ic): Update.
11368 2017-06-02 Carl Love <cel@us.ibm.com>
11370 * config/rs6000/rs6000-c: Add support for built-in functions
11371 vector double vec_doublee (vector signed int);
11372 vector double vec_doublee (vector unsigned int);
11373 vector double vec_doublee (vector float);
11374 vector double vec_doubleh (vector signed int);
11375 vector double vec_doubleh (vector unsigned int);
11376 vector double vec_doubleh (vector float);
11377 vector double vec_doublel (vector signed int);
11378 vector double vec_doublel (vector unsigned int);
11379 vector double vec_doublel (vector float);
11380 vector double vec_doubleo (vector signed int);
11381 vector double vec_doubleo (vector unsigned int);
11382 vector double vec_doubleo (vector float);.
11383 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
11384 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
11386 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
11387 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
11388 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
11390 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
11391 vec_doublel, vec_doubleh.
11392 * doc/extend.texi: Update the built-in documentation file for the
11393 new built-in functions.
11395 2017-06-02 David Malcolm <dmalcolm@redhat.com>
11398 * ipa-inline-analysis.c (free_growth_caches): Set
11399 edge_removal_hook_holder to NULL after removing it.
11401 2017-06-02 Sudakshina Das <sudi.das@arm.com>
11403 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
11404 comparision with zero.
11406 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
11407 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11408 for early expansion of vec_min and vec_max builtins.
11409 (builtin_function_type): Add min/max unsigned variants to those
11410 identified as having unsigned arguments.
11412 2017-06-02 Olivier Hainque <hainque@adacore.com>
11414 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
11416 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11418 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
11419 Use VALL_F16 iterator rather than VALL.
11421 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11423 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
11424 Emit CBNZ inside loop when doing a strong exchange and comparing
11425 against zero. Generate the CC flags after the loop.
11427 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
11429 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
11430 (dl_section_ref): New.
11431 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
11432 On AIX, append an expression to subtract the size of the
11433 section length to dl_section_ref.
11435 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
11437 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
11438 for early expansion of vector absolute builtins.
11440 2017-06-02 Richard Biener <rguenther@suse.de>
11442 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
11443 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
11445 2017-06-02 Richard Biener <rguenther@suse.de>
11447 PR tree-optimization/80948
11448 * tree-tailcall.c (find_tail_calls): Track stmts to move in
11449 stmt order as well.
11451 2017-06-02 Richard Biener <rguenther@suse.de>
11453 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
11455 * tree-vect-stmts.c (process_use): Do not mark backedge defs
11456 for inductions as relevant.
11458 2017-06-02 Richard Biener <rguenther@suse.de>
11460 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
11461 (vectorizable_induction): ... this. Remove dead code.
11463 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
11465 * builtins. (expand_builtin_alloca): Remove second parameter and
11466 infer its value from the first parameter instead.
11467 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
11469 2017-06-02 Jakub Jelinek <jakub@redhat.com>
11471 PR rtl-optimization/80903
11472 * loop-doloop.c (add_test): Unshare sequence.
11474 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
11476 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
11478 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
11480 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
11482 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
11483 xlogue_layout::get_instance, logue_layout::xlogue_layout,
11484 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
11485 (xlogue_layout::get_stub_rtx): Make static.
11486 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
11487 (xlogue_layout::compute_stub_managed_regs): Rename to...
11488 (xlogue_layout::count_stub_managed_regs): ...this.
11489 (xlogue_layout::is_stub_managed_reg): New function.
11490 (xlogue_layout::m_stub_names): Rename to...
11491 (xlogue_layout::s_stub_names): ...this, make static.
11492 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
11493 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
11494 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
11495 xlogue_layout::s_stub_names): Instantiate statics.
11496 (stub_managed_regs): Remove.
11497 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
11498 (disable_call_ms2sysv_xlogues): Rename to...
11499 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
11500 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
11502 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
11503 change after reload_completed.
11504 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
11506 (ix86_expand_prologue): Likewise.
11507 (ix86_expand_epilogue): Likewise.
11508 (ix86_expand_split_stack_prologue): Likewise.
11509 (ix86_compute_frame_layout): Remove frame parameter ...
11510 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
11511 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
11513 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
11514 (ix86_frame): Move from here ...
11515 * config/i386/i386.h (ix86_frame): ... to here.
11516 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
11517 complete ix86_frame data structure instead. Remove some_ld_name.
11519 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
11521 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
11522 symbols that hold a DECL_VALUE_EXPR.
11524 2017-06-01 Martin Jambor <mjambor@suse.cz>
11526 PR tree-optimization/80898
11527 * tree-sra.c (process_subtree_disqualification): Removed.
11528 (disqualify_candidate): Do not acll
11529 process_subtree_disqualification.
11530 (subtree_mark_written_and_enqueue): New function.
11531 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
11532 RHS has been disqualified and re-queue LHS if necessary. Apart
11533 from that, ignore disqualified RHS.
11535 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11537 * config/s390/s390.c (s390_emit_epilogue): Disable early return
11538 address fetch for z10 or later.
11540 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11542 * config/arc/arc.md (tst_movb): Add guard when splitting.
11544 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11546 * config/arc/arc.c (arc_can_eliminate): Test against
11547 arc_frame_pointer_needed.
11549 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11551 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
11552 to prevent store reordering.
11553 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
11554 (type): Add block type.
11555 (stack_tie): Define special instruction to be used in
11558 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11560 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
11561 constraint. It is not valid for the pattern.
11562 (noncommutative_binary_comparison): Likewise.
11564 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11566 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
11569 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11571 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
11572 be used by the reg-alloc.
11574 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11576 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
11577 reg-alloc when having mul64 or mul32x16 instructions.
11578 (mulsidi3): Likewise.
11579 (umulsidi3): Likewise.
11580 (mulsi32x16): New pattern.
11581 (mulsi64): Likewise.
11582 (mulsidi64): Likewise.
11583 (umulsidi64): Likewise.
11584 (MUL32x16_REG): Define.
11585 (mul64_600): Use MUL32x16_REG.
11586 (mac64_600): Likewise.
11587 (umul64_600): Likewise.
11588 (umac64_600): Likewise.
11590 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
11592 * config/arc/arc.md (mulsi3_700): Make it commutative.
11594 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
11596 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
11598 (*sign_extendsidi2_insn): Likewise for movstosw.
11600 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
11602 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
11603 the type of the input discriminant value. Convert the
11604 discriminant value of signedness vary.
11606 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
11608 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
11609 Add to -Wall section.
11611 2017-06-01 Richard Biener <rguenther@suse.de>
11613 PR middle-end/66313
11614 * fold-const.c (fold_plusminus_mult_expr): If the factored
11615 factor may be zero use a wrapping type for the inner operation.
11616 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
11617 and handle moved defs.
11618 (process_assignment): Properly guard the unary op case. Return a
11619 tri-state indicating that moving the stmt before the call may allow
11620 to continue. Pass through to_move.
11621 (find_tail_calls): Handle moving unrelated defs before
11624 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
11627 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
11628 splitter result in the canonical way.
11630 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
11632 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
11633 also for 32bit target. Update insn attributes.
11634 (zero-extendsidi2 splitter): Allow all registers for operand 1.
11636 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
11638 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
11639 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
11640 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
11641 (_mm_maskz_min_ss): New intrinsics.
11643 2017-05-31 Martin Liska <mliska@suse.cz>
11645 * tree-vect-loop.c (vect_create_epilog_for_reduction):
11646 Change comment style to one we normally use.
11647 (vectorizable_reduction): Likewise.
11648 (vectorizable_induction): Likewise.
11649 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11650 (vectorizable_call): Likewise.
11651 (vectorizable_simd_clone_call): Likewise.
11652 (vectorizable_conversion): Likewise.
11653 (vectorizable_assignment): Likewise.
11654 (vectorizable_shift): Likewise.
11655 (vectorizable_operation): Likewise.
11656 (vectorizable_store): Likewise.
11657 (vectorizable_load): Likewise.
11658 * tree-vectorizer.h: Likewise.
11660 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
11662 * passes.c (emergency_dump_function): New.
11663 * tree-pass.h (emergency_dump_function): Declare.
11664 * plugin.c (plugins_internal_error_function): Remove.
11665 * plugin.h (plugins_internal_error_function): Remove declaration.
11666 * toplev.c (internal_error_function): New static function. Use it...
11667 (general_init): ...here.
11669 2017-05-31 Graham Markall <graham.markall@embecosm.com>
11671 * config/arc/arc.c (arc_print_operand): Handle constant operands.
11672 (arc_rtx_costs): Add costs for new patterns.
11673 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
11674 * config/arc/predicates.md: Add _1_2_3_operand predicate.
11676 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11678 * tree-ssa-strlen.c (get_next_strinfo): New function.
11679 (get_stridx_plus_constant): Use it.
11680 (zero_length_string): Likewise.
11681 (adjust_related_strinfos): Likewise.
11682 (adjust_last_stmt): Likewise.
11684 2017-05-31 Richard Biener <rguenther@suse.de>
11687 * config/i386/i386.c (ix86_expand_builtin): Remove assert
11688 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
11690 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
11692 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
11693 loop_vinfo argument and use of dependence distance vectors.
11694 Check instead whether the two references differ only in their
11695 initial value and assume that they have the same alignment if the
11696 difference is a multiple of the vector alignment.
11697 (vect_analyze_data_refs_alignment): Update call accordingly.
11699 2017-05-31 Martin Liska <mliska@suse.cz>
11702 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
11704 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11706 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
11707 (create_intersect_range_checks): Move from ...
11708 * tree-data-ref.c (create_intersect_range_checks_index)
11709 (create_intersect_range_checks): ... to here.
11710 (create_runtime_alias_checks): New function factored from ...
11711 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
11712 here. Call above function.
11713 * tree-data-ref.h (create_runtime_alias_checks): New function.
11715 2017-05-31 Bin Cheng <bin.cheng@arm.com>
11717 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
11718 segment length for dr_b and compute it in wide_int.
11720 2017-05-31 Richard Biener <rguenther@suse.de>
11722 PR tree-optimization/80906
11723 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
11724 and pass through iv_map.
11725 (copy_bb_and_scalar_dependences): Adjust.
11726 (translate_pending_phi_nodes): Likewise.
11727 (copy_loop_close_phi_args): Handle code-generating IVs instead
11730 2017-05-30 David Malcolm <dmalcolm@redhat.com>
11732 * diagnostic-color.c (color_dict): Add "type-diff".
11733 (parse_gcc_colors): Update comment.
11734 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
11735 -fdiagnostics-show-template-tree and -fno-elide-type.
11736 (GCC_COLORS): Add type-diff to example.
11738 (-fdiagnostics-show-template-tree): New.
11739 (-fno-elide-type): New.
11740 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
11741 the pp_format_decoder callback. Call any m_format_postprocessor's
11743 (pretty_printer::pretty_printer): Initialize
11744 m_format_postprocessor.
11745 (pretty_printer::~pretty_printer): Delete any
11746 m_format_postprocessor.
11747 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
11748 (class format_postprocessor): New class.
11749 (struct pretty_printer::format_decoder): Document the new parameters.
11750 (struct pretty_printer::m_format_postprocessor): New field.
11751 * tree-diagnostic.c (default_tree_printer): Update for new
11752 bool and const char ** params.
11753 * tree-diagnostic.h (default_tree_printer): Likewise.
11755 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
11757 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
11758 (lwa_operand): Delete rs6000_gen_cell_microcode test.
11759 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
11760 rs6000_gen_cell_microcode code.
11761 (rs6000_final_prescan_insn): Delete.
11762 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
11763 "warn-cell-microcode" entries.
11764 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
11765 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
11766 throughout. Change cc_reg_not_micro_cr0_operand to
11767 cc_reg_not_cr0_operand throughout.
11768 (*extendhi<mode>2_noload): Delete.
11769 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
11770 (mwarn-cell-microcode): Delete.
11771 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
11772 -mgen-cell-microcode and -mwarn-cell-microcode.
11774 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
11777 * config/i386/constraints.md (Yd): New constraint.
11779 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
11780 and (?Yd, r) alternatives. Update insn attributes.
11781 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
11782 and (?*Yd, r) alternatives. Update insn attributes.
11783 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
11785 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
11787 * gimplify.c (gimplify_modify_expr): Don't create a
11788 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
11791 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
11793 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
11795 2017-05-30 Richard Biener <rguenther@suse.de>
11797 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
11798 and reduc_def fields.
11799 (STMT_VINFO_REDUC_TYPE): New define.
11800 (STMT_VINFO_REDUC_DEF): Likewise.
11801 (vect_force_simple_reduction): Adjust prototype.
11802 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
11803 (vect_is_simple_reduction): Remove check_reduction argument.
11804 (vect_force_simple_reduction): Adjust and set
11805 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11806 (vectorizable_reduction): Do not re-do reduction analysis
11807 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
11808 * tree-parloops.c (gather_scalar_reductions): Adjust.
11810 2017-05-30 Richard Biener <rguenther@suse.de>
11812 PR middle-end/80901
11813 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
11816 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
11818 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
11819 Introduce unknown_misalignment parameter and remove vf.
11820 (vect_peeling_hash_get_lowest_cost):
11821 Pass unknown_misalignment parameter.
11822 (vect_enhance_data_refs_alignment):
11823 Fix unsupportable data ref treatment.
11825 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11827 * tree-vect-data-refs.c (vect_get_data_access_cost):
11828 Workaround for SLP handling.
11829 (vect_enhance_data_refs_alignment):
11830 Compute costs for doing no peeling at all, compare to the best
11831 peeling costs so far and avoid peeling if cheaper.
11833 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11835 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
11836 Return peeling info and set costs to zero for unlimited cost
11838 (vect_enhance_data_refs_alignment): Also inspect all datarefs
11839 with unknown misalignment. Compute and costs for unknown
11840 misalignment, compare them to the costs for known misalignment
11841 and choose the cheapest for peeling.
11843 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11845 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
11846 (vect_get_peeling_costs_all_drs): Create function.
11847 (vect_peeling_hash_get_lowest_cost):
11848 Use vect_get_peeling_costs_all_drs.
11849 (vect_peeling_supportable): Create function.
11850 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
11852 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
11854 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
11855 DR_HAS_NEGATIVE_STEP.
11856 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
11857 (vect_enhance_data_refs_alignment): Use.
11858 (vect_duplicate_ssa_name_ptr_info): Use.
11859 * tree-vectorizer.h (dr_misalignment): Use.
11860 (known_alignment_for_access_p): Use.
11862 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11865 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
11867 (has_section_name): New function.
11869 2017-05-30 Martin Liska <mliska@suse.cz>
11872 * auto-profile.c (get_function_decl_from_block): Fix
11875 2017-05-30 Richard Biener <rguenther@suse.de>
11877 PR middle-end/80876
11878 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
11880 2017-05-30 Martin Liska <mliska@suse.cz>
11882 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
11883 * dumpfile.h (struct dump_file_info): Remove ctors.
11885 2017-05-30 Martin Liska <mliska@suse.cz>
11887 * predict.def: Fix GNU coding style.
11889 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11891 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
11892 Mark 'to' argument with ATTRIBUTE_UNUSED.
11894 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
11896 * config/xtensa/xtensa.c (xtensa_emit_call): Use
11897 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
11898 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
11901 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
11903 * doc/install.texi (Options specification): Restore entry of
11904 --enable-sjlj-exceptions.
11906 2017-05-27 Michael Eager <eager@eagercon.com>
11909 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
11911 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
11913 * config/microblaze/microblaze.h
11914 (FIXED_REGISTERS): Update in macro.
11915 (CALL_USED_REGISTERS): Update in macro.
11917 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
11919 * doc/install.texi: Add links to macOS binary distributions.
11921 2017-05-27 Jakub Jelinek <jakub@redhat.com>
11925 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
11927 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
11929 2017-05-26 Martin Liska <mliska@suse.cz>
11931 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
11933 2017-05-26 Martin Liska <mliska@suse.cz>
11935 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
11936 always leading ';; '.
11937 (dump_bb_info): Likewise.
11938 (brief_dump_cfg): Likewise.
11939 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
11940 * dumpfile.c: Remove usage of TDF_VERBOSE.
11941 * dumpfile.h (enum dump_kind): Likewise.
11942 (dump_gimple_bb_header): Do not use TDF_COMMENT.
11943 * print-tree.c (debug_verbose): Remove.
11944 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
11945 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
11946 * tree-diagnostic.c (default_tree_printer): Replace
11947 TDF_DIAGNOSTIC with TDF_SLIM.
11949 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11951 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
11952 in parameter loop, rather than loop_vinfo.
11953 (create_intersect_range_checks): Ditto.
11954 (vect_create_cond_for_alias_checks): Update call to above functions.
11956 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11958 PR tree-optimization/80815
11959 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
11960 for merging runtime alias checks. Handle negative DR_STEPs.
11962 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11964 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
11966 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
11967 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
11968 out code pruning runtime alias checks.
11969 * tree-data-ref.c (prune_runtime_alias_test_list): New function
11970 factored out from above.
11971 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
11973 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
11975 (prune_runtime_alias_test_list): New decalaration.
11977 2017-05-26 Bin Cheng <bin.cheng@arm.com>
11979 * tree-vect-data-refs.c (compare_tree): Rename and move ...
11980 * tree-data-ref.c (data_ref_compare_tree): ... to here.
11981 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
11982 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
11983 (operator==, comp_dr_with_seg_len_pair): Ditto.
11984 (vect_prune_runtime_alias_test_list): Ditto.
11986 2017-05-26 Martin Liska <mliska@suse.cz>
11989 * params.def: Bound partial-inlining-entry-probability param.
11991 2017-05-26 Marek Polacek <polacek@redhat.com>
11994 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
11997 2017-05-26 Richard Biener <rguenther@suse.de>
11999 PR tree-optimization/80842
12000 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
12003 2017-05-26 Richard Biener <rguenther@suse.de>
12005 PR tree-optimization/80844
12006 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
12008 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
12010 * doc/md.texi (Machine Constraints): Update x86 family
12011 machine constraints section to match 'config/i386/constraints.md'.
12013 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
12015 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
12017 2017-05-25 Nathan Sidwell <nathan@acm.org>
12019 * doc/invoke.texi (--enable-languages): Update documentation.
12021 2017-05-25 Martin Liska <mliska@suse.cz>
12023 * dumpfile.c: Add TDF_FOLDING.
12024 * dumpfile.h (enum dump_kind): Likewise.
12025 * genmatch.c (dt_simplify::gen_1): Use it.
12027 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12029 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
12031 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12033 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
12034 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
12036 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
12038 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
12039 * match.pd (X == C): Rewrite it here.
12040 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
12041 with_certain_nonzero_bits2): New predicates.
12042 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
12044 2017-05-24 Nathan Sidwell <nathan@acm.org>
12046 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
12049 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
12052 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
12054 * config/powerpcspe: New port. Files are copied from the rs6000
12055 port, with "rs6000" in filenames replaced by "powerpcspe".
12057 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
12059 PR rtl-optimization/80754
12060 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
12062 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
12064 * config/sparc/sparc.md (length): Return the correct value for -mflat
12065 sibcalls to match output_sibcall.
12067 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
12071 * config/rs6000/rs6000.c (struct machine_function): Add new field
12073 (rs6000_get_separate_components): Init that field, use it.
12074 (rs6000_components_for_bb): Use the field.
12076 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
12078 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
12080 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
12082 PR middle-end/80823
12083 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
12085 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12088 * config/s390/s390.c (s390_check_qrst_address): Check incoming
12089 address against address_operand predicate.
12090 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
12092 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
12094 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
12095 parameters passed indirectly.
12097 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
12099 * config/i386/i386.md (*movdi_internal): Remove SSE4
12100 alternative 18 (?r, *v). Update insn attributes.
12101 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
12102 Update insn attributes.
12103 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
12104 Update insn attributes.
12105 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
12106 alternative 1 (r, v). Remove isa attribute.
12107 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
12108 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
12109 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
12111 2017-05-23 Tom de Vries <tom@codesourcery.com>
12113 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
12116 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12118 * cgraphunit.c (symbol_table::process_new_functions): Update.
12119 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
12120 (inline_generate_summary): Rename to ...
12121 (ipa_fn_summary_generate): ... this one.
12122 (inline_read_summary): Rename to ...
12123 (ipa_fn_summary_read): ... this one.
12124 (inline_write_summary): Rename to ...
12125 (ipa_fn_summary_write): ... this one.
12126 (inline_free_summary): Rename to ...
12127 (ipa_free_fn_summary): ... this one.
12128 (pass_data_local_fn_summary, pass_local_fn_summary,
12129 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
12130 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
12131 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
12132 make_pass_ipa_fn_summary): New.
12133 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
12134 inline_write_summary, inline_free_summary): Remove.
12135 (ipa_free_fn_summary) : New.
12136 * ipa-inline.c (ipa_inline): Update.
12137 (pass_ipa_inline): Do not generate summaries.
12138 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
12140 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
12141 and add pass_ipa_fn_summary.
12142 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
12144 (make_pass_inline_parameters): Remove.
12146 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
12148 * omp-low.c (struct omp_context): Remove "default_kind" member.
12151 * omp-offload.c (execute_oacc_device_lower): Remove the
12152 parallelism dimensions function attributes for unparallelized
12153 OpenACC kernels constructs.
12155 2017-05-23 Martin Liska <mliska@suse.cz>
12157 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
12159 (cgraph_edge::make_speculative): Likewise.
12160 (cgraph_edge::resolve_speculation): Likewise.
12161 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
12162 (cgraph_node::dump): Likewise.
12163 * cgraph.h: Likewise.
12164 * cgraphunit.c (analyze_functions): Likewise.
12165 (symbol_table::compile): Likewise.
12166 * ipa-cp.c (print_all_lattices): Likewise.
12167 (determine_versionability): Likewise.
12168 (initialize_node_lattices): Likewise.
12169 (ipcp_verify_propagated_values): Likewise.
12170 (estimate_local_effects): Likewise.
12171 (update_profiling_info): Likewise.
12172 (create_specialized_node): Likewise.
12173 (perhaps_add_new_callers): Likewise.
12174 (decide_about_value): Likewise.
12175 (decide_whether_version_node): Likewise.
12176 (identify_dead_nodes): Likewise.
12177 (ipcp_store_bits_results): Likewise.
12178 * ipa-devirt.c (dump_targets): Likewise.
12179 (ipa_devirt): Likewise.
12180 * ipa-icf.c (sem_item::dump): Likewise.
12181 (sem_function::equals): Likewise.
12182 (sem_variable::equals): Likewise.
12183 (sem_item_optimizer::read_section): Likewise.
12184 (sem_item_optimizer::execute): Likewise.
12185 (congruence_class::dump): Likewise.
12186 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
12187 (dump_inline_summary): Likewise.
12188 (estimate_node_size_and_time): Likewise.
12189 (inline_analyze_function): Likewise.
12190 * ipa-inline-transform.c (inline_call): Likewise.
12191 * ipa-inline.c (report_inline_failed_reason): Likewise.
12192 (want_early_inline_function_p): Likewise.
12193 (edge_badness): Likewise.
12194 (update_edge_key): Likewise.
12195 (inline_small_functions): Likewise.
12196 * ipa-profile.c (ipa_profile): Likewise.
12197 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
12198 (ipa_make_edge_direct_to_target): Likewise.
12199 (remove_described_reference): Likewise.
12200 (ipa_impossible_devirt_target): Likewise.
12201 (propagate_controlled_uses): Likewise.
12202 (ipa_print_node_params): Likewise.
12203 (ipcp_transform_function): Likewise.
12204 * ipa-pure-const.c (pure_const_read_summary): Likewise.
12205 (propagate_pure_const): Likewise.
12206 * ipa-reference.c (generate_summary): Likewise.
12207 (read_write_all_from_decl): Likewise.
12208 (propagate): Likewise.
12209 (ipa_reference_read_optimization_summary): Likewise.
12210 * ipa-utils.c (ipa_merge_profiles): Likewise.
12211 * ipa.c (walk_polymorphic_call_targets): Likewise.
12212 (symbol_table::remove_unreachable_nodes): Likewise.
12213 (ipa_single_use): Likewise.
12214 * passes.c (execute_todo): Likewise.
12215 * predict.c (drop_profile): Likewise.
12216 * symtab.c (symtab_node::get_dump_name): New function.
12217 (symtab_node::dump_name): Likewise.
12218 (symtab_node::dump_asm_name): Likewise.
12219 (symtab_node::dump_references): Likewise.
12220 (symtab_node::dump_referring): Likewise.
12221 (symtab_node::dump_base): Likewise.
12222 (symtab_node::debug_symtab): Likewise.
12223 * tree-sra.c (convert_callers_for_node): Likewise.
12224 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
12225 * value-prof.c (init_node_map): Likewise.
12227 2017-05-23 Martin Liska <mliska@suse.cz>
12229 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
12230 and symtab_node::debug_symtab to symbol_table::debug.
12231 * cgraphunit.c (analyze_functions): Use the renamed function.
12232 (symbol_table::compile): Likewise.
12233 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
12234 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
12235 * passes.c (execute_todo): Likewise.
12236 * symtab.c (symbol_table::dump): New function.
12237 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
12239 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12241 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
12242 that nonconst implies exec.
12244 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
12246 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
12247 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
12248 (inline_edge_summary_vec): Turn into ...
12249 (ipa_call_summaries): ... this one.
12250 (redirect_to_unreachable, edge_set_predicate,
12251 evaluate_properties_for_edge, inline_summary_alloc,
12252 reset_ipa_call_summary, reset_inline_summary,
12253 inline_summary_t::duplicate): Update.
12254 (inline_edge_duplication_hook): Turn to ...
12255 (ipa_call_summary_t::duplicate): ... this one.
12256 (inline_edge_removal_hook): Turn to ...
12257 (ipa_call_summary_t::remove): ... this one.
12258 (dump_inline_edge_summary): Turn to ...
12259 (dump_ipa_call_summary): ... this one.
12260 (estimate_function_body_sizes): Update.
12261 (inline_update_callee_summaries): Update.
12262 (remap_edge_change_prob): Update.
12263 (remap_edge_summaries): Update.
12264 (inline_merge_summary): Update.
12265 (do_estimate_edge_time): Update.
12266 (inline_generate_summary): Update.
12267 (inline_read_section): Update.
12268 (inline_read_summary): Update.
12269 (inline_free_summary): Update.
12270 * ipa-inline.c (can_inline_edge_p): Update.
12271 (compute_inlined_call_time): Update.
12272 (want_inline_small_function_p): Update.
12273 (edge_badness): Update.
12274 (early_inliner): Update.
12275 * ipa-inline.h (inline_edge_summary): Turn to ...
12276 (ipa_call_summary): ... this one.
12277 (ipa_call_summary_t): New class.
12278 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
12279 (ipa_call_summaries): New.
12280 (inline_edge_summary): Remove.
12281 (estimate_edge_growth): Update.
12282 * ipa-profile.c (ipa_propagate_frequency_1): Update.
12283 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12284 * ipa-split.c (execute_split_functions): Update.
12285 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
12287 2017-05-23 Tom de Vries <tom@codesourcery.com>
12289 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
12290 attributes): Document rdrand effective target.
12292 2017-05-23 Tom de Vries <tom@codesourcery.com>
12294 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
12295 attributes): Sort alphabetically.
12297 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
12299 * config/avr/genmultilib.awk: Use gsub instead of gensub.
12301 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
12304 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
12305 V2DF/V2DI splat into two separate patterns, one that handles
12306 registers, and the other that only handles memory. Drop support
12307 for splatting from a GPR on ISA 2.07 and then splitting the
12308 splat into direct move and splat.
12309 (vsx_splat_<mode>_reg): Likewise.
12310 (vsx_splat_<mode>_mem): Likewise.
12312 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
12314 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
12316 2017-05-22 Jakub Jelinek <jakub@redhat.com>
12318 PR middle-end/80809
12319 * omp-low.c (finish_taskreg_remap): New function.
12320 (finish_taskreg_scan): If unit size of ctx->record_type
12321 is non-constant, unshare the size expression and replace
12322 decls in it with possible outer var refs.
12324 PR middle-end/80809
12325 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
12326 GOVD_SHARED rather than GOVD_PRIVATE with it.
12327 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
12328 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
12330 PR middle-end/80853
12331 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
12332 as last argument to build_outer_var_ref for pointer bases of array
12333 section reductions.
12335 2017-05-19 Martin Sebor <msebor@redhat.com>
12337 * print-tree.c (print_node): Print DECL_READ_P flag.
12339 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12341 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
12342 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
12343 * cgraph.c: Likewise.
12344 * cgraphunit.c: Likewise.
12345 * gengtype.c: Likewise.
12346 * ipa-cp.c: Likewise.
12347 * ipa-devirt.c: Likewise.
12348 * ipa-icf.c: Likewise.
12349 * ipa-predicate.c: Likewise.
12350 * ipa-profile.c: Likewise.
12351 * ipa-prop.c: Likewise.
12352 * ipa-split.c: Likewise.
12354 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
12355 edge_predicate_pool, dump_inline_hints,
12356 inline_summary::account_size_time, redirect_to_unreachable,
12357 edge_set_predicate, set_hint_predicate,
12358 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
12359 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
12360 inline_summary_t::remove, remap_hint_predicate_after_duplication,
12361 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
12362 ipa_call_summary_t::remove, initialize_growth_caches,
12363 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
12364 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
12365 mark_modified, unmodified_parm_1, unmodified_parm,
12366 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
12367 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
12368 compute_bb_predicates, will_be_nonconstant_expr_predicate,
12369 will_be_nonconstant_predicate, record_modified_bb_info,
12370 get_minimal_bb, record_modified, param_change_prob,
12371 phi_result_unknown_predicate, predicate_for_phi_result,
12372 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
12373 estimate_function_body_sizes, compute_inline_parameters,
12374 compute_inline_parameters_for_curren, pass_data_inline_parameters,
12375 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
12376 inline_update_callee_summaries, remap_edge_change_prob,
12377 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
12378 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
12379 inline_analyze_function, inline_summary_t::insert,
12380 inline_generate_summary, read_ipa_call_summary, inline_read_section,
12381 inline_read_summary, write_ipa_call_summary, inline_write_summary,
12382 inline_free_summary): Move to ipa-fnsummary.h
12383 (predicate_t): Remove.
12384 * ipa-fnsummary.c: New file.
12385 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
12386 (enum inline_hints_vals, inline_hints, agg_position_info,
12387 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
12388 inline_summaries, ipa_call_summary, ipa_call_summary_t,
12389 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
12390 dump_inline_summary, dump_inline_hints, inline_generate_summary,
12391 inline_read_summary, inline_write_summary, inline_free_summary,
12392 inline_analyze_function, initialize_inline_failed,
12393 inline_merge_summary, inline_update_overall_summary,
12394 compute_inline_parameters): Move to ipa-fnsummary.h
12395 * ipa-fnsummary.h: New file.
12396 * ipa-inline-transform.h: Include ipa-inline.h.
12397 * ipa-inline.c: LIkewise.
12399 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12401 * ipa-inline.c (edge_badness): Use inlined_time instead of
12402 inline_summaries->get.
12404 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12406 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
12408 2017-05-22 Nathan Sidwell <nathan@acm.org>
12410 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
12411 (fdump-lang): Document 'raw' option.
12412 * dumpfile.h (TDI_tu): Delete.
12413 * dumpfile.c (dump_files): Remove translation-unit.
12414 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
12416 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
12418 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
12419 command option from $(AWK) call.
12420 * config/avr/genmultilib.awk: Simplify and rewrite so that it
12421 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
12422 [FORMAT]: Remove handling of variable.
12423 * config/avr/t-multilib: Regenerate.
12425 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12427 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
12429 (dump_inline_summary): Do not print self_time.
12430 (estimate_function_body_sizes): Do not set self_time.
12431 (compute_inline_parameters): Likewise.
12432 (inline_read_section, inline_write_summary): Do not stream self_time.
12433 * ipa-inline.h (inline_summary): Drop self_time.
12435 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
12437 * ipa-inline-analysis.c (account_size_time): Rename to ...
12438 (inline_summary::account_size_time): ... this one.
12439 (reset_ipa_call_summary): Turn to ...
12440 (ipa_call_summary::reset): ... this one.
12441 (reset_inline_summary): Turn to ...
12442 (inline_summary::reset): ... this one.
12443 (inline_summary_t::remove): Update.
12444 (inline_summary_t::duplicate): Update.
12445 (ipa_call_summary_t::remove): Update.
12446 (dump_inline_summary): Update.
12447 (estimate_function_body_sizes): Update.
12448 (compute_inline_parameters): Update.
12449 (estimate_node_size_and_time): Update.
12450 (inline_merge_summary): Update.
12451 (inline_update_overall_summary): Update.
12452 (inline_read_section): Update.
12453 (inline_write_summary): Update.
12454 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
12455 add account_size_time and reset member functions.
12456 (ipa_call_summary): Add reset function.
12457 * ipa-predicate.h (predicate::operator &): Constify.
12459 2017-05-22 Richard Biener <rguenther@suse.de>
12461 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
12463 2017-05-19 Jason Merrill <jason@redhat.com>
12465 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
12467 2017-05-19 Marek Polacek <polacek@redhat.com>
12470 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
12471 TYPE_OVERFLOW_WRAPS checks.
12473 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
12475 * tree-core.h (enum omp_clause_default_kind): Add
12476 "OMP_CLAUSE_DEFAULT_PRESENT".
12477 * tree-pretty-print.c (dump_omp_clause): Handle it.
12478 * gimplify.c (enum gimplify_omp_var_data): Add
12479 "GOVD_MAP_FORCE_PRESENT".
12480 (gimplify_adjust_omp_clauses_1): Map it to
12481 "GOMP_MAP_FORCE_PRESENT".
12482 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
12484 * gimplify.c (oacc_default_clause): Clarify.
12486 2017-05-19 Nathan Sidwell <nathan@acm.org>
12488 LANG_HOOK_REGISTER_DUMPS
12489 * toplev.c (general_init): Call register dump lang hook.
12490 * doc/invoke.texi: Document -fdump-lang option family.
12491 * dumpfile.c (dump_files): Remove class dump here.
12492 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
12493 * dumpfile.h (tree_dump_index): Remove TDI_class.
12494 * langhooks-def.h (lhd_register_dumps): Declare.
12495 (LANG_HOOKS_REGISTER_DUMPS): Define.
12496 (LANG_HOOKS_INITIALIZER): Add it.
12497 * langhooks.c (lhd_register_dumps): Define.
12498 * langhooks.h (struct lang_hooks): Add register_dumps.
12500 2017-05-19 Nathan Sidwell <nathan@acm.org>
12502 * context.h (context::set_passes): New.
12503 * context.c (context::context): Do not create pass manager.
12504 * toplev.c (general_init): Create pass manager here.
12506 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
12508 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
12509 use this splitter if two add or or instructions would also work for
12510 the constant we want to generate.
12512 2017-05-19 Richard Biener <rguenther@suse.de>
12515 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
12516 predicate evaluation.
12518 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12520 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
12522 * ipa-inline.c (want_inline_small_function_p): Do not cast to
12525 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
12527 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
12528 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
12529 (inline_edge_summary_vec): Turn into ...
12530 (ipa_call_summaries): ... this one.
12531 (redirect_to_unreachable, edge_set_predicate,
12532 evaluate_properties_for_edge, inline_summary_alloc,
12533 reset_ipa_call_summary, reset_inline_summary,
12534 inline_summary_t::duplicate): Update.
12535 (inline_edge_duplication_hook): Turn to ...
12536 (ipa_call_summary_t::duplicate): ... this one.
12537 (inline_edge_removal_hook): Turn to ...
12538 (ipa_call_summary_t::remove): ... this one.
12539 (dump_inline_edge_summary): Turn to ...
12540 (dump_ipa_call_summary): ... this one.
12541 (estimate_function_body_sizes): Update.
12542 (inline_update_callee_summaries): Update.
12543 (remap_edge_change_prob): Update.
12544 (remap_edge_summaries): Update.
12545 (inline_merge_summary): Update.
12546 (do_estimate_edge_time): Update.
12547 (inline_generate_summary): Update.
12548 (inline_read_section): Update.
12549 (inline_read_summary): Update.
12550 (inline_free_summary): Update.
12551 * ipa-inline.c (can_inline_edge_p): Update.
12552 (compute_inlined_call_time): Update.
12553 (want_inline_small_function_p): Update.
12554 (edge_badness): Update.
12555 (early_inliner): Update.
12556 * ipa-inline.h (inline_edge_summary): Turn to ...
12557 (ipa_call_summary): ... this one.
12558 (ipa_call_summary_t): New class.
12559 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
12560 (ipa_call_summaries): New.
12561 (inline_edge_summary): Remove.
12562 (estimate_edge_growth): Update.
12563 * ipa-profile.c (ipa_propagate_frequency_1): Update.
12564 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
12565 * ipa-split.c (execute_split_functions): Update.
12566 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
12568 2017-05-19 Richard Biener <rguenther@suse.de>
12570 PR middle-end/80764
12571 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
12573 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
12575 * config/rs6000/rs6000.c (struct machine_function): Add field
12576 fpr_is_wrapped_separately.
12577 (rs6000_get_separate_components): Use 64 components. Handle the
12578 new FPR components.
12579 (rs6000_components_for_bb): Handle the FPR components.
12580 (rs6000_emit_prologue_components): Handle the FPR components.
12581 (rs6000_emit_epilogue_components): Handle the FPR components.
12582 (rs6000_set_handled_components): Handle the FPR components.
12583 (rs6000_emit_prologue): Don't output prologue code for those FPRs
12584 that are already separately shrink-wrapped.
12585 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
12586 that are already separately shrink-wrapped.
12588 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
12591 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
12594 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
12595 (define_peephole2 for Altivec d-form load): Add peepholes to catch
12596 cases where the register allocator uses a move and an offsettable
12597 memory operation to/from a FPR register on ISA 2.06/2.07.
12598 (define_peephole2 for Altivec d-form store): Likewise.
12600 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
12603 * config/i386/mmx.md (*mov<mode>_internal): Enable
12604 alternatives 11, 12, 13 and 14 also for 32bit targets.
12605 Remove alternatives 15, 16, 17 and 18.
12606 * config/i386/sse.md (vec_concatv2di): Change
12607 alternative (!x, *y) to (x, ?!*Yn).
12609 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
12611 * dumpfile.h (enum dump_kind): Remove stray comma.
12613 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
12615 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
12616 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
12617 predicate::num_conditions
12618 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
12619 (CHANGED): turn into predicate::changed.
12620 (agg_position_info): Move to ipa-predicate.h
12621 (add_condition, predicate::add_clause, predicate::operator &=,
12622 predicate::or_with, predicate::evaluate, predicate::probability,
12623 dump_condition, dump_clause, predicate::dump,
12624 predicate::remap_after_duplication, predicate::remap_after_inlining,
12625 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
12626 (evaluate_conditions_for_known_args): Update.
12627 (set_cond_stmt_execution_predicate): Update.
12628 * ipa-inline.h: Include ipa-predicate.h
12629 (condition, inline_param_summary, conditions, agg_position_info,
12630 predicate): Move to ipa-predicate.h
12631 * ipa-predicate.c: New file.
12632 * ipa-predicate.h: New file.
12634 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
12636 * final.c (leaf_function_p): Check we are not in a sequence.
12638 2017-05-18 Martin Liska <mliska@suse.cz>
12640 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
12641 * dumpfile.c (dump_register): Use new enum dump_kind.
12642 (get_dump_file_name): Likewise.
12643 (dump_enable_all): Likewise.
12644 (dump_switch_p_1): Likewise.
12645 (enable_rtl_dump_file): Remove usage of TDF_RTL.
12646 * dumpfile.h (enum dump_kind): New enum type.
12647 (struct dump_file_info): Create constructor and
12648 format fields and comments.
12649 * passes.c (pass_manager::register_one_dump_file):
12651 * statistics.c (statistics_early_init): Likewise.
12652 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
12653 TDF_TREE with TDF_SLIM.
12654 (gather_memory_references_ref): Likewise.
12656 2017-05-18 Martin Liska <mliska@suse.cz>
12658 * vec.h (struct vnull): Use it.
12660 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
12662 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
12663 (true_predicate, false_predicate, true_predicate_p,
12664 false_predicate_p): Remove.
12665 (single_cond_predicate, not_inlined_predicate): Turn to member function
12667 (add_condition): Update.
12668 (add_clause): Turn to...
12669 (predicate::add_clause): ... this one; update; allow passing NULL
12671 (and_predicates): Turn to ...
12672 (predicate::operator &=): ... this one.
12673 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
12674 (or_predicates): Turn to ...
12675 (predicate::or_with): ... this one.
12676 (evaluate_predicate): Turn to ...
12677 (predicate::evaluate): ... this one.
12678 (predicate_probability): Turn to ...
12679 (predicate::probability): ... this one.
12680 (dump_condition): Update.
12681 (dump_predicate): Turn to ...
12682 (predicate::dump): ... this one.
12683 (account_size_time): Update.
12684 (edge_set_predicate): Update.
12685 (set_hint_predicate): UPdate.
12686 (evaluate_conditions_for_known_args): Update.
12687 (evaluate_properties_for_edge): Update.
12688 (remap_predicate_after_duplication): Turn to...
12689 (predicate::remap_after_duplication): ... this one.
12690 (remap_hint_predicate_after_duplication): Update.
12691 (inline_summary_t::duplicate): UPdate.
12692 (dump_inline_edge_summary): Update.
12693 (dump_inline_summary): Update.
12694 (set_cond_stmt_execution_predicate): Update.
12695 (set_switch_stmt_execution_predicate): Update.
12696 (compute_bb_predicates): Update.
12697 (will_be_nonconstant_expr_predicate): Update.
12698 (will_be_nonconstant_predicate): Update.
12699 (phi_result_unknown_predicate): Update.
12700 (predicate_for_phi_result): Update.
12701 (array_index_predicate): Update.
12702 (estimate_function_body_sizes): Update.
12703 (estimate_node_size_and_time): Update.
12704 (estimate_ipcp_clone_size_and_time): Update.
12705 (remap_predicate): Rename to ...
12706 (predicate::remap_after_inlining): ... this one.
12707 (remap_hint_predicate): Update.
12708 (inline_merge_summary): Update.
12709 (inline_update_overall_summary): Update.
12710 (estimate_size_after_inlining): Update.
12711 (read_predicate): Rename to ...
12712 (predicate::stream_in): ... this one.
12713 (read_inline_edge_summary): Update.
12714 (write_predicate): Rename to ...
12715 (predicate::stream_out): ... this one.
12716 (write_inline_edge_summary): Update.
12717 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
12718 (clause_t): Turn to uint32_t
12719 (predicate): Turn to class; implement constructor and operators
12721 (size_time_entry): Update.
12722 (inline_summary): Update.
12723 (inline_edge_summary): Update.
12725 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
12727 * fold-const.c (fold_binary_loc): Move transformation...
12728 * match.pd (C - X CMP X): ... here.
12730 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
12732 * config/sparc/sparc.c (sparc_option_override): Set function
12733 alignment for -mcpu=niagara7 to 64 to match the I$ line.
12734 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
12736 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
12738 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
12740 2017-05-18 Marek Polacek <polacek@redhat.com>
12743 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
12744 (pass_ubsan::execute): Call gimple_assign_single_p instead of
12745 gimple_assign_load_p.
12747 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
12749 PR middle-end/80692
12750 * real.c (do_compare): Give decimal_do_compare preference over
12751 comparing just the signs.
12753 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
12755 * doc/md.texi (Canonicalization of Instructions): Describe the
12756 canonical form of instructions that inherently set a condition
12759 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
12761 PR middle-end/80775
12762 * tree-cfg.c: Move deletion of unreachable case statements to after
12763 the merging of consecutive case labels.
12765 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
12767 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
12768 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
12769 restoring of callee-saved registers.
12771 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
12773 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
12774 * config/visium/visium.c (single_set_and_flags): Likewise.
12775 * config/visium/visium.md (Substitutions): Likewise.
12777 2017-05-17 Martin Liska <mliska@suse.cz>
12779 * cfg.c: Introduce dump_flags_t type and
12780 use it instead of int type.
12782 * cfghooks.c: Likewise.
12783 * cfghooks.h (struct cfg_hooks): Likewise.
12784 * cfgrtl.c: Likewise.
12785 * cfgrtl.h: Likewise.
12786 * cgraph.c (cgraph_node::get_body): Likewise.
12787 * coretypes.h: Likewise.
12788 * domwalk.c: Likewise.
12789 * domwalk.h: Likewise.
12790 * dumpfile.c (struct dump_option_value_info): Likewise.
12791 (dump_enable_all): Likewise.
12792 (dump_switch_p_1): Likewise.
12793 (opt_info_switch_p): Likewise.
12794 * dumpfile.h (enum tree_dump_index): Likewise.
12795 (struct dump_file_info): Likewise.
12796 * genemit.c: Likewise.
12797 * generic-match-head.c: Likewise.
12798 * gengtype.c (open_base_files): Likewise.
12799 * gimple-pretty-print.c: Likewise.
12800 * gimple-pretty-print.h: Likewise.
12801 * graph.c (print_graph_cfg): Likewise.
12802 * graphite-scop-detection.c (dot_all_sese): Likewise.
12803 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
12804 * loop-unroll.c (report_unroll): Likewise.
12805 * passes.c (pass_manager::register_one_dump_file): Likewise.
12806 * print-tree.c: Likewise.
12807 * statistics.c: Likewise.
12808 * tree-cfg.c: Likewise.
12809 * tree-cfg.h: Likewise.
12810 * tree-dfa.c: Likewise.
12811 * tree-dfa.h: Likewise.
12812 * tree-dump.c (dump_function): Likewise.
12813 * tree-dump.h (struct dump_info): Likewise.
12814 * tree-pretty-print.c: Likewise.
12815 * tree-pretty-print.h: Likewise.
12816 * tree-ssa-live.c: Likewise.
12817 * tree-ssa-live.h: Likewise.
12818 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
12819 * tree-vect-loop.c: Likewise.
12820 * tree-vect-slp.c: Likewise.
12822 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
12823 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12825 PR tree-optimization/80457
12826 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
12827 of all arguments to a statement as scalar_to_vec operations.
12828 (vectorizable_call): Adjust call to vect_model_simple_cost for
12830 (vectorizable_conversion): Likewise.
12831 (vectorizable_assignment): Likewise.
12832 (vectorizable_shift): Likewise.
12833 (vectorizable_operation): Likewise.
12834 (vectorizable_comparison): Likewise.
12835 (vect_is_simple_cond): Record the def types for operands.
12836 (vectorizable_condition): Likewise, call vect_model_simple_cost.
12837 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
12838 for statement argument count.
12840 2017-05-16 Carl Love <cel@us.ibm.com>
12842 * config/rs6000/rs6000-c: Add support for built-in functions
12843 vector unsigned long long vec_bperm (vector unsigned long long,
12844 vector unsigned char)
12845 vector signed long long vec_mule (vector signed int,
12847 vector unsigned long long vec_mule (vector unsigned int,
12848 vector unsigned int)
12849 vector signed long long vec_mulo (vector signed int,
12851 vector unsigned long long vec_mulo (vector unsigned int,
12852 vector unsigned int)
12853 vector signed char vec_sldw (vector signed char,
12854 vector signed char,
12856 vector unsigned char vec_sldw (vector unsigned char,
12857 vector unsigned char,
12859 vector signed short vec_sldw (vector signed short,
12860 vector signed short,
12862 vector unsigned short vec_sldw (vector unsigned short,
12863 vector unsigned short,
12865 vector signed int vec_sldw (vector signed int,
12868 vector unsigned int vec_sldw (vector unsigned int,
12869 vector unsigned int,
12871 vector signed long long vec_sldw (vector signed long long,
12872 vector signed long long,
12874 vector unsigned long long vec_sldw (vector unsigned long long,
12875 vector unsigned long long,
12877 * config/rs6000/rs6000-c: Add support for built-in functions
12878 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
12879 * config/rs6000/altivec.h: Add defintion for vec_sldw.
12880 * doc/extend.texi: Update the built-in documentation for the
12881 new built-in functions.
12883 2017-05-16 Marek Polacek <polacek@redhat.com>
12887 * tree.c (save_expr): Don't fold the expression.
12889 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
12891 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
12892 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
12893 and (?*y,m). Update insn attributes.
12895 2017-05-16 Martin Liska <mliska@suse.cz>
12897 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
12898 flags argument of print_gimple_stmt, print_gimple_expr,
12899 print_generic_stmt and print_generic_expr.
12900 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
12901 * coretypes.h: Likewise.
12902 * except.c (dump_eh_tree): Likewise.
12903 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
12904 * gimple-pretty-print.h: Likewise.
12905 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
12906 (backprop::push_to_worklist): Likewise.
12907 (backprop::pop_from_worklist): Likewise.
12908 (backprop::process_use): Likewise.
12909 (backprop::intersect_uses): Likewise.
12910 (note_replacement): Likewise.
12911 * gimple-ssa-store-merging.c
12912 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
12913 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
12914 (pass_store_merging::execute): Likewise.
12915 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
12916 (ssa_base_cand_dump_callback): Likewise.
12917 (dump_incr_vec): Likewise.
12918 (replace_refs): Likewise.
12919 (replace_mult_candidate): Likewise.
12920 (create_add_on_incoming_edge): Likewise.
12921 (create_phi_basis): Likewise.
12922 (insert_initializers): Likewise.
12923 (all_phi_incrs_profitable): Likewise.
12924 (introduce_cast_before_cand): Likewise.
12925 (replace_one_candidate): Likewise.
12926 * gimplify.c (gimplify_expr): Likewise.
12927 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
12928 (set_rename): Likewise.
12929 (rename_uses): Likewise.
12930 (copy_loop_phi_nodes): Likewise.
12931 (add_close_phis_to_merge_points): Likewise.
12932 (copy_loop_close_phi_args): Likewise.
12933 (copy_cond_phi_args): Likewise.
12934 (graphite_copy_stmts_from_block): Likewise.
12935 (translate_pending_phi_nodes): Likewise.
12936 * graphite-poly.c (print_pdr): Likewise.
12937 (dump_gbb_cases): Likewise.
12938 (dump_gbb_conditions): Likewise.
12939 (print_scop_params): Likewise.
12940 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
12941 (build_cross_bb_scalars_use): Likewise.
12942 (gather_bbs::before_dom_children): Likewise.
12943 * hsa-dump.c (dump_hsa_immed): Likewise.
12944 * ipa-cp.c (print_ipcp_constant_value): Likewise.
12945 (get_replacement_map): Likewise.
12946 * ipa-inline-analysis.c (dump_condition): Likewise.
12947 (estimate_function_body_sizes): Likewise.
12948 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
12949 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
12950 * ipa-prop.c (ipa_dump_param): Likewise.
12951 (ipa_print_node_jump_functions_for_edge): Likewise.
12952 (ipa_modify_call_arguments): Likewise.
12953 (ipa_modify_expr): Likewise.
12954 (ipa_dump_param_adjustments): Likewise.
12955 (ipa_dump_agg_replacement_values): Likewise.
12956 (ipcp_modif_dom_walker::before_dom_children): Likewise.
12957 * ipa-pure-const.c (check_stmt): Likewise.
12958 (pass_nothrow::execute): Likewise.
12959 * ipa-split.c (execute_split_functions): Likewise.
12960 * omp-offload.c (dump_oacc_loop_part): Likewise.
12961 (dump_oacc_loop): Likewise.
12962 * trans-mem.c (tm_log_emit): Likewise.
12963 (tm_memopt_accumulate_memops): Likewise.
12964 (dump_tm_memopt_set): Likewise.
12965 (dump_tm_memopt_transform): Likewise.
12966 * tree-cfg.c (gimple_verify_flow_info): Likewise.
12967 (print_loop): Likewise.
12968 * tree-chkp-opt.c (chkp_print_addr): Likewise.
12969 (chkp_gather_checks_info): Likewise.
12970 (chkp_get_check_result): Likewise.
12971 (chkp_remove_check_if_pass): Likewise.
12972 (chkp_use_outer_bounds_if_possible): Likewise.
12973 (chkp_reduce_bounds_lifetime): Likewise.
12974 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
12975 (chkp_mark_completed_bounds): Likewise.
12976 (chkp_register_incomplete_bounds): Likewise.
12977 (chkp_mark_invalid_bounds): Likewise.
12978 (chkp_maybe_copy_and_register_bounds): Likewise.
12979 (chkp_build_returned_bound): Likewise.
12980 (chkp_get_bound_for_parm): Likewise.
12981 (chkp_build_bndldx): Likewise.
12982 (chkp_get_bounds_by_definition): Likewise.
12983 (chkp_generate_extern_var_bounds): Likewise.
12984 (chkp_get_bounds_for_decl_addr): Likewise.
12985 * tree-chrec.c (chrec_apply): Likewise.
12986 * tree-data-ref.c (dump_data_reference): Likewise.
12987 (dump_subscript): Likewise.
12988 (dump_data_dependence_relation): Likewise.
12989 (analyze_overlapping_iterations): Likewise.
12990 * tree-inline.c (expand_call_inline): Likewise.
12991 (tree_function_versioning): Likewise.
12992 * tree-into-ssa.c (dump_defs_stack): Likewise.
12993 (dump_currdefs): Likewise.
12994 (dump_names_replaced_by): Likewise.
12995 (dump_update_ssa): Likewise.
12996 (update_ssa): Likewise.
12997 * tree-object-size.c (pass_object_sizes::execute): Likewise.
12998 * tree-parloops.c (build_new_reduction): Likewise.
12999 (try_create_reduction_list): Likewise.
13000 (ref_conflicts_with_region): Likewise.
13001 (oacc_entry_exit_ok_1): Likewise.
13002 (oacc_entry_exit_single_gang): Likewise.
13003 * tree-pretty-print.h: Likewise.
13004 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
13005 (get_scalar_evolution): Likewise.
13006 (add_to_evolution): Likewise.
13007 (get_loop_exit_condition): Likewise.
13008 (analyze_evolution_in_loop): Likewise.
13009 (analyze_initial_condition): Likewise.
13010 (analyze_scalar_evolution): Likewise.
13011 (instantiate_scev): Likewise.
13012 (number_of_latch_executions): Likewise.
13013 (gather_chrec_stats): Likewise.
13014 (final_value_replacement_loop): Likewise.
13015 (scev_const_prop): Likewise.
13016 * tree-sra.c (dump_access): Likewise.
13017 (disqualify_candidate): Likewise.
13018 (create_access): Likewise.
13019 (reject): Likewise.
13020 (maybe_add_sra_candidate): Likewise.
13021 (create_access_replacement): Likewise.
13022 (analyze_access_subtree): Likewise.
13023 (analyze_all_variable_accesses): Likewise.
13024 (sra_modify_assign): Likewise.
13025 (initialize_constant_pool_replacements): Likewise.
13026 (find_param_candidates): Likewise.
13027 (decide_one_param_reduction): Likewise.
13028 (replace_removed_params_ssa_names): Likewise.
13029 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
13030 * tree-ssa-copy.c (dump_copy_of): Likewise.
13031 (copy_prop_visit_cond_stmt): Likewise.
13032 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
13033 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
13034 (record_equivalences_from_stmt): Likewise.
13035 * tree-ssa-dse.c (compute_trims): Likewise.
13036 (delete_dead_call): Likewise.
13037 (delete_dead_assignment): Likewise.
13038 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
13039 (forward_propagate_into_cond): Likewise.
13040 (pass_forwprop::execute): Likewise.
13041 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
13042 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
13044 (move_computations_worker): Likewise.
13045 (execute_sm): Likewise.
13046 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
13047 (remove_exits_and_undefined_stmts): Likewise.
13048 (remove_redundant_iv_tests): Likewise.
13049 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
13050 (adjust_iv_update_pos): Likewise.
13051 * tree-ssa-math-opts.c (bswap_replace): Likewise.
13052 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
13053 (value_replacement): Likewise.
13054 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
13055 * tree-ssa-pre.c (print_pre_expr): Likewise.
13056 (get_representative_for): Likewise.
13057 (create_expression_by_pieces): Likewise.
13058 (insert_into_preds_of_block): Likewise.
13059 (eliminate_insert): Likewise.
13060 (eliminate_dom_walker::before_dom_children): Likewise.
13061 (eliminate): Likewise.
13062 (remove_dead_inserted_code): Likewise.
13063 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
13064 * tree-ssa-reassoc.c (get_rank): Likewise.
13065 (eliminate_duplicate_pair): Likewise.
13066 (eliminate_plus_minus_pair): Likewise.
13067 (eliminate_not_pairs): Likewise.
13068 (undistribute_ops_list): Likewise.
13069 (eliminate_redundant_comparison): Likewise.
13070 (update_range_test): Likewise.
13071 (optimize_range_tests_var_bound): Likewise.
13072 (optimize_vec_cond_expr): Likewise.
13073 (rewrite_expr_tree): Likewise.
13074 (rewrite_expr_tree_parallel): Likewise.
13075 (linearize_expr): Likewise.
13076 (break_up_subtract): Likewise.
13077 (linearize_expr_tree): Likewise.
13078 (attempt_builtin_powi): Likewise.
13079 (attempt_builtin_copysign): Likewise.
13080 (transform_stmt_to_copy): Likewise.
13081 (transform_stmt_to_multiply): Likewise.
13082 (dump_ops_vector): Likewise.
13083 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
13084 (print_scc): Likewise.
13085 (set_ssa_val_to): Likewise.
13086 (visit_reference_op_store): Likewise.
13087 (visit_use): Likewise.
13088 (sccvn_dom_walker::before_dom_children): Likewise.
13089 (run_scc_vn): Likewise.
13090 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
13092 (expr_hash_elt::print): Likewise.
13093 (const_and_copies::pop_to_marker): Likewise.
13094 (const_and_copies::record_const_or_copy_raw): Likewise.
13095 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
13096 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
13097 (dump_predicates): Likewise.
13098 (find_uninit_use): Likewise.
13099 (warn_uninitialized_phi): Likewise.
13100 (pass_late_warn_uninitialized::execute): Likewise.
13101 * tree-ssa.c (verify_vssa): Likewise.
13102 (verify_ssa): Likewise.
13103 (maybe_optimize_var): Likewise.
13104 * tree-vrp.c (dump_value_range): Likewise.
13105 (dump_all_value_ranges): Likewise.
13106 (dump_asserts_for): Likewise.
13107 (register_edge_assert_for_2): Likewise.
13108 (vrp_visit_cond_stmt): Likewise.
13109 (vrp_visit_switch_stmt): Likewise.
13110 (vrp_visit_stmt): Likewise.
13111 (vrp_visit_phi_node): Likewise.
13112 (simplify_cond_using_ranges_1): Likewise.
13113 (fold_predicate_in): Likewise.
13114 (evrp_dom_walker::before_dom_children): Likewise.
13115 (evrp_dom_walker::push_value_range): Likewise.
13116 (evrp_dom_walker::pop_value_range): Likewise.
13117 (execute_early_vrp): Likewise.
13119 2017-05-16 Richard Biener <rguenther@suse.de>
13121 * dwarf2out.c (loc_list_from_tree_1): Do not create
13122 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
13124 2017-05-16 Richard Biener <rguenther@suse.de>
13126 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
13128 (note_variable_value_in_expr): If we resolved the decl ref
13129 do not push to the stack.
13131 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
13133 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
13134 operations in fast-math mode.
13135 (vaddq_f16): Likewise.
13136 (vmul_f16): Likewise.
13137 (vmulq_f16): Likewise.
13138 (vsub_f16): Likewise.
13139 (vsubq_f16): Likewise.
13140 * config/arm/neon.md (add<mode>3): New.
13142 (fma:<VH:mode>3): New. Also remove outdated comment.
13145 2017-05-16 Martin Liska <mliska@suse.cz>
13149 * ipa-devirt.c (warn_types_mismatch): Fix typo.
13150 (odr_types_equivalent_p): Likewise.
13152 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
13154 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
13156 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
13159 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
13160 non-interunit SSE move alternatives with '?'.
13161 (zero-extendsidi peephole2): New peephole to skip intermediate
13162 general register in SSE zero-extend sequence.
13164 2017-05-15 Jeff Law <law@redhat.com>
13166 * reorg.c (relax_delay_slots): Create a new variable to hold
13167 the temporary target rather than clobbering TARGET_LABEL.
13169 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
13170 missing argument to extract_bit_field call.
13171 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
13173 2017-05-15 Martin Liska <mliska@suse.cz>
13176 * gcc.c (process_command): Do not allow empty argument of -o option.
13178 2017-05-15 Renlin Li <renlin.li@arm.com>
13180 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
13181 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
13182 * config/aarch64/constraints.md (Usf): Add long call check.
13183 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
13184 (call_value): Likewise.
13185 (sibcall): Likewise.
13186 (sibcall_value): Likewise.
13188 (call_value_insn): New.
13189 (sibcall_insn): Update rtx pattern.
13190 (sibcall_value_insn): Likewise.
13191 (call_internal): Remove.
13192 (call_value_internal): Likewise.
13193 (sibcall_internal): Likewise.
13194 (sibcall_value_internal): Likewise.
13195 (call_reg): Likewise.
13196 (call_symbol): Likewise.
13197 (call_value_reg): Likewise.
13198 (call_value_symbol): Likewise.
13200 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
13203 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
13205 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
13207 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
13208 compatible with CCGOCmode and with CCZmode.
13210 2017-05-14 Martin Sebor <msebor@redhat.com>
13212 PR middle-end/77671
13213 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
13214 (gimple_fold_builtin_snprintf): Same.
13215 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
13216 (gimple_fold_builtin_snprintf): Same.
13217 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
13218 of character types.
13219 (is_call_safe): New function.
13220 (try_substitute_return_value): Call it.
13221 (try_simplify_call): New function.
13222 (pass_sprintf_length::handle_gimple_call): Call it.
13224 2017-05-14 Martin Sebor <msebor@redhat.com>
13226 PR middle-end/80669
13227 * builtins.c (expand_builtin_stpncpy): Simplify.
13229 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
13231 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
13232 * config/i386/i386.h
13233 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
13234 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
13235 (struct machine_function): Add new members call_ms2sysv,
13236 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
13237 (struct machine_frame_state): New fields sp_realigned and
13238 sp_realigned_offset.
13239 * config/i386/i386.c
13240 (enum xlogue_stub): New enum.
13241 (enum xlogue_stub_sets): New enum.
13242 (class xlogue_layout): New class.
13243 (struct ix86_frame): New fields stack_realign_allocate_offset,
13244 stack_realign_offset and outlined_save_offset. Modify comments to
13245 detail stack layout when using out-of-line stubs.
13246 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
13247 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
13248 -mcall-ms2sysv-xlogues.
13249 (stub_managed_regs): New static variable.
13250 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
13251 registers managed by out-of-line stub.
13252 (disable_call_ms2sysv_xlogues): New function.
13253 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
13254 m->call_ms2sysv when appropriate and compute frame layout for
13256 (sp_valid_at, fp_valid_at): New inline functions.
13257 (choose_basereg): New function.
13258 (choose_baseaddr): Add align parameter, use choose_basereg and modify
13260 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
13261 Use align parameter of choose_baseaddr to generated aligned SSE movs
13263 (pro_epilogue_adjust_stack): Modify to track
13264 machine_frame_state::sp_realigned.
13265 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
13266 (ix86_nsaved_sseregs): Likewise.
13267 (ix86_emit_save_regs): Likewise.
13268 (ix86_emit_save_regs_using_mov): Likewise.
13269 (ix86_emit_save_sse_regs_using_mov): Likewise.
13270 (get_scratch_register_on_entry): Likewise.
13271 (gen_frame_set): New function.
13272 (gen_frame_load): Likewise.
13273 (gen_frame_store): Likewise.
13274 (emit_outlined_ms2sysv_save): Likewise.
13275 (emit_outlined_ms2sysv_restore): Likewise.
13276 (ix86_expand_prologue): Modify stack re-alignment code and call
13277 emit_outlined_ms2sysv_save when appropriate.
13278 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
13279 parameter rtx_insn *insn, which allows the function to be used to only
13280 generate the notes.
13281 (ix86_expand_epilogue): Modify validity checks of frame and stack
13282 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
13283 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
13284 * config/i386/predicates.md
13285 (save_multiple): New predicate.
13286 (restore_multiple): Likewise.
13287 * config/i386/sse.md
13288 (save_multiple<mode>): New pattern.
13289 (save_multiple_realign<mode>): Likewise.
13290 (restore_multiple<mode>): Likewise.
13291 (restore_multiple_and_return<mode>): Likewise.
13292 (restore_multiple_leave_return<mode>): Likewise.
13293 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
13295 2017-05-14 Julia Koval <julia.koval@intel.com>
13297 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
13298 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
13299 (__builtin_ia32_xsetbv): New builtins.
13300 * config/i386/i386.c (ix86_expand_special_args_builtin):
13302 (ix86_expand_builtin): Special expand for new intrinsics.
13303 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
13304 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
13305 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
13307 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13309 * cfganal.c (inverted_post_order_compute): Change argument type
13311 * cfganal.h (inverted_post_order_compute): Adjust prototype.
13312 * df-core.c (rest_of_handle_df_initialize): Adjust.
13313 (rest_of_handle_df_finish): Likewise.
13314 (df_analyze_1): Likewise.
13315 (df_analyze): Likewise.
13316 (loop_inverted_post_order_compute): Change argument to be a vec *.
13317 (df_analyze_loop): Adjust.
13318 (df_get_n_blocks): Likewise.
13319 (df_get_postorder): Likewise.
13320 * df.h (struct df_d): Change field to be a vec.
13321 * lcm.c (compute_laterin): Adjust.
13322 (compute_available): Likewise.
13323 * lra-lives.c (lra_create_live_ranges_1): Likewise.
13324 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
13325 * tree-ssa-pre.c (compute_antic): Likewise.
13327 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13329 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
13330 (depth_first_search::depth_first_search): Change structure init
13331 function to this constructor.
13332 (depth_first_search::add_bb): Rename function to this member.
13333 (depth_first_search::execute): Likewise.
13334 (flow_dfs_compute_reverse_finish): Adjust.
13336 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13338 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
13339 (longest_simple_path): Likewise.
13340 * shrink-wrap.c (spread_components): Likewise.
13341 (disqualify_problematic_components): Likewise.
13342 (emit_common_heads_for_components): Likewise.
13343 (emit_common_tails_for_components): Likewise.
13344 (insert_prologue_epilogue_for_components): Likewise.
13346 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13348 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
13351 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13353 * df-core.c (df_set_blocks): Start using auto_bitmap.
13354 (df_compact_blocks): Likewise.
13355 * df-problems.c (df_rd_confluence_n): Likewise.
13356 * df-scan.c (df_insn_rescan_all): Likewise.
13357 (df_process_deferred_rescans): Likewise.
13358 (df_update_entry_block_defs): Likewise.
13359 (df_update_exit_block_uses): Likewise.
13360 (df_entry_block_bitmap_verify): Likewise.
13361 (df_exit_block_bitmap_verify): Likewise.
13362 (df_scan_verify): Likewise.
13363 * lra-constraints.c (lra_constraints): Likewise.
13364 (undo_optional_reloads): Likewise.
13365 (lra_undo_inheritance): Likewise.
13366 * lra-remat.c (calculate_gen_cands): Likewise.
13367 (do_remat): Likewise.
13368 * lra-spills.c (assign_spill_hard_regs): Likewise.
13369 (spill_pseudos): Likewise.
13370 * tree-ssa-pre.c (bitmap_set_and): Likewise.
13371 (bitmap_set_subtract_values): Likewise.
13373 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13375 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
13376 management with auto_bitmap.
13377 (fix_inter_tick): Likewise.
13378 (fix_recovery_deps): Likewise.
13379 * ira.c (add_store_equivs): Likewise.
13380 (find_moveable_pseudos): Likewise.
13381 (split_live_ranges_for_shrink_wrap): Likewise.
13382 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
13383 (rtx_reuse_manager::seen_def_p): Likewise.
13384 (rtx_reuse_manager::set_seen_def): Likewise.
13385 * print-rtl.h (class rtx_reuse_manager): Likewise.
13387 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13389 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
13391 (migrate_btr_def): Likewise.
13392 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
13393 * df-core.c (loop_post_order_compute): Likewise.
13394 (loop_inverted_post_order_compute): Likewise.
13395 * hsa-common.h: Likewise.
13396 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
13397 * init-regs.c (initialize_uninitialized_regs): Likewise.
13398 * ipa-inline.c (resolve_noninline_speculation): Likewise.
13399 (inline_small_functions): Likewise.
13400 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
13401 * ira.c (combine_and_move_insns): Likewise.
13402 (build_insn_chain): Likewise.
13403 * loop-invariant.c (find_invariants): Likewise.
13404 * lower-subreg.c (propagate_pseudo_copies): Likewise.
13405 * predict.c (tree_predict_by_opcode): Likewise.
13406 (predict_paths_leading_to): Likewise.
13407 (predict_paths_leading_to_edge): Likewise.
13408 (estimate_loops_at_level): Likewise.
13409 (estimate_loops): Likewise.
13410 * shrink-wrap.c (try_shrink_wrapping): Likewise.
13411 (spread_components): Likewise.
13412 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
13413 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
13414 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
13415 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
13416 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
13417 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
13418 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
13419 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
13420 (mark_threaded_blocks): Likewise.
13421 (thread_through_all_blocks): Likewise.
13422 * tree-ssa.c (verify_ssa): Likewise.
13423 (execute_update_addresses_taken): Likewise.
13424 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
13426 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13428 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
13430 (post_order_compute): Likewise.
13431 (inverted_post_order_compute): Likewise.
13432 (pre_and_rev_post_order_compute_fn): Likewise.
13434 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13436 * genrecog.c (int_set::int_set): Explicitly construct our
13437 auto_vec base class.
13438 * vec.h (auto_vec::auto_vec): New constructor.
13440 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13442 * bitmap.h (class auto_bitmap): New constructor taking
13443 bitmap_obstack * argument.
13445 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13447 * bitmap.h (class auto_bitmap): Change type of m_bits to
13448 bitmap_head, and adjust ctor / dtor and member operators.
13450 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
13452 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
13453 when returned register mode doesn't match original mode.
13455 2017-05-12 Jeff Law <law@redhat.com>
13456 Jakub Jelinek <jakub@redhat.com>
13458 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
13459 we look for cc setter after the compare-elim changes.
13460 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
13461 within the vector to match what compare-elim now expects.
13462 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
13463 (xorsi3_flags, one_cmplsi2_flags): Likewise.
13465 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
13466 after the compare-elim changes.
13467 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
13468 the vector to match what compare-elim now expects.
13469 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
13470 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
13471 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
13472 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
13473 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
13475 * config/visium/visium.c (single_set_and_flags): Fix where
13476 we look for cc setter after the compare-elim changes.
13477 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
13478 with the vector to match what compare-elim now expects.
13479 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
13480 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
13481 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
13482 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
13483 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
13484 (neg<mode>2_insn_set_overflow): Likewise.
13486 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
13488 PR middle-end/79794
13489 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
13490 maybe_expand_insn call, set ops[0].target. If still set after call,
13491 set alt_rtl. Add extra arg to recursive calls.
13492 (extract_bit_field): Add alt_rtl argument. Pass to
13494 * expmed.h (extract_bit_field): Fix prototype.
13495 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
13496 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
13497 to extract_bit_field_calls.
13498 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
13499 Pass alt_rtl to extract_bit_field calls.
13500 * calls.c (store_unaligned_arguments_into_psuedos)
13501 load_register_parameters): Pass extra NULL to extract_bit_field calls.
13502 * optabs.c (maybe_legitimize_operand): Clear op->target when call
13504 * optabs.h (struct expand_operand): Add target bitfield.
13506 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13508 * compare-elim.c (try_eliminate_compare): Canonicalize
13509 operation with embedded compare to
13510 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
13511 (set (reg) (operation)].
13513 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
13515 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
13518 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
13519 cost of adding a carry flag for ADC instruction.
13520 [case MINUS]: Ignore the cost of subtracting a carry flag
13521 for SBB instruction.
13523 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
13525 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
13527 * config/rs6000/bmiintrin.h: New file.
13528 * config/rs6000/bmi2intrin.h: New file.
13529 * config/rs6000/x86intrin.h: New file.
13531 2017-05-12 Jeff Law <law@redhat.com>
13533 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
13536 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
13538 PR middle-end/80707
13539 * tree-cfg.c: Remove cfg edges of unreachable case statements.
13541 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13543 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
13544 early expansion of vector divide builtins.
13545 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
13546 builtins identified as having unsigned arguments.
13548 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13550 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
13551 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
13552 expansion of vector logical operations (and, andc, or, xor,
13555 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13557 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
13558 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
13560 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
13562 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
13563 early GIMPLE expansion of vector multiplies.
13565 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
13567 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
13568 TARGET_HAVE_MOVT conditional.
13569 (movt splitter): Likewise.
13571 2017-05-12 Richard Biener <rguenther@suse.de>
13573 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
13574 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13575 Fold all stmts not inplace.
13577 2017-05-12 Richard Biener <rguenther@suse.de>
13579 PR tree-optimization/80713
13580 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
13581 inserted_exprs bit for not removed stmts.
13583 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
13585 PR middle-end/69921
13586 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
13587 parallelized" attribute for parallelized OpenACC kernels.
13588 * omp-offload.c (execute_oacc_device_lower): Use it.
13590 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
13591 Set "oacc kernels" attribute.
13592 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
13593 parameter. Adjust all users.
13594 (oacc_fn_attrib_kernels_p): Remove function.
13595 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
13596 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
13597 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
13598 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
13599 assert "oacc kernels" attribute is set.
13601 2017-05-11 Carl Love <cel@us.ibm.com>
13603 * config/rs6000/rs6000-c: Add support for built-in functions
13604 vector unsigned char vec_popcnt (vector signed char)
13605 vector unsigned char vec_popcnt (vector unsigned char)
13606 vector unsigned short vec_popcnt (vector signed short)
13607 vector unsigned short vec_popcnt (vector unsigned short)
13608 vector unsigned int vec_popcnt (vector signed int)
13609 vector unsigned int vec_popcnt (vector unsigned int)
13610 vector unsigned long long vec_popcnt (vector signed long long)
13611 vector unsigned long long vec_popcnt (vector unsigned long long)
13612 vector signed long long vec_slo (vector signed long long,
13613 vector signed char)
13614 vector signed long long vec_slo (vector signed long long,
13615 vector unsigned char)
13616 vector unsigned long long vec_slo (vector unsigned long long,
13617 vector signed char)
13618 vector unsigned long long vec_slo (vector unsigned long long,
13619 vector unsigned char)
13620 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
13621 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
13622 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
13623 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
13624 * doc/extend.texi: Update the built-in documentation file for the
13625 new built-in functions.
13627 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
13629 * attribs.h (sorted_attr_string): Move machine independent
13630 functions for target clone support from the i386 port to common
13631 code. Rename ix86_function_versions to common_function_versions.
13632 Rename make_name to make_unique_name.
13633 (common_function_versions): Likewise.
13634 (make_unique_name): Likewise.
13635 (make_dispatcher_decl): Likewise.
13636 (is_function_default_version): Likewise.
13637 * attribs.c (attr_strcmp): Likewise.
13638 (sorted_attr_string): Likewise.
13639 (common_function_versions): Likewise.
13640 (make_unique_name): Likewise.
13641 (make_dispatcher_decl): Likewise.
13642 (is_function_default_version): Likewise.
13643 * config/i386/i386.c (attr_strcmp): Likewise.
13644 (sorted_attr_string): Likewise.
13645 (ix86_function_versions): Likewise.
13646 (make_name): Likewise.
13647 (make_dispatcher_decl): Likewise.
13648 (is_function_default_version): Likewise.
13649 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
13651 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13654 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
13655 Account for direct move costs for vec_construct of integer
13658 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
13661 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
13662 (UNSPEC_STX_ATOMIC): Ditto.
13663 (loaddi_via_sse): New insn.
13664 (storedi_via_sse): Ditto.
13665 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
13666 Update corresponding peephole2 patterns.
13667 (atomic_storedi_fpu): Ditto.
13669 2017-05-11 Julia Koval <julia.koval@intel.com>
13671 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
13672 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
13674 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
13675 (__builtin_ia32_rsqrt14ss_mask): New builtins.
13676 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
13678 2017-05-11 Nathan Sidwell <nathan@acm.org>
13680 * graphite-poly.c: Include dumpfile.h.
13682 * dumpfle.h (dump_function): Declare here ...
13683 * tree-dump.h (dump_function): ... not here.
13684 * dumpfile.c: #include tree-cfg.h.
13685 (dump_function): Move here from ...
13686 * tree-dump.c (dump_function): ... here.
13687 * gimplify.c: #include splay-tree.h, not tree-dump.h.
13688 * graphite-poly.c: Don't include tree-dump.h.
13689 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
13690 * print-tree.c: Likewise.
13691 * stor-layout.c: Likewise.
13692 * tree-nested.c: Likewise.
13694 * dumpfile.c (dump_start): Use TDF_FLAGS.
13695 (dump_enable_all): Fix TDF_KIND check thinko.
13697 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
13699 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
13700 array entries to represent two legal parameterizations of the
13701 overloaded __builtin_cmpb function, as represented by the
13702 P6_OV_BUILTIN_CMPB constant.
13703 (altivec_resolve_overloaded_builtin): Add special case handling
13704 for the __builtin_cmpb function, as represented by the
13705 P6_OV_BUILTIN_CMPB constant.
13706 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
13707 (BU_P6_64BIT_2): New macro.
13708 (BU_P6_OVERLOAD_2): New macro
13709 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
13710 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
13711 (CMPB): Add overload support to represent both 32-bit and 64-bit
13712 compare-bytes function.
13713 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
13714 support for TARGET_CMPB.
13715 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
13716 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
13717 documentation of the __builtin_cmpb overloaded built-in function.
13719 2017-05-11 Richard Biener <rguenther@suse.de>
13721 PR tree-optimization/80705
13722 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
13723 bases are not vectorizable.
13725 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13727 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
13728 when counting register pressure.
13730 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13732 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
13733 (iv_ca_more_deps): Renamed to ...
13734 (iv_ca_compare_deps): ... this.
13735 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
13737 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13739 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
13741 (determine_group_iv_costs): ... here.
13742 (find_inv_vars_cb): Record inv var if it's not recorded before.
13744 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13746 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
13747 (get_shiftadd_cost): Ditto.
13749 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13751 * tree-ssa-address.c: Include header file.
13752 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
13754 (add_to_parts): Refactor.
13755 (addr_to_parts): New parameter. Update use of move_hint_to_base.
13756 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
13759 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13761 PR tree-optimization/53090
13762 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
13764 (extract_cond_operands): Detect condition with IV on both sides
13765 and return COMP_IV_EXPR_2.
13766 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
13767 (rewrite_use_compare): Simplify by removing call to function
13768 extract_cond_operands.
13770 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13772 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
13773 (extract_cond_operands): Detect condition comparing against non-
13774 invariant bound and return appropriate enum value.
13775 (find_interesting_uses_cond): Update use of extract_cond_operands.
13776 Handle its return value accordingly.
13777 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
13779 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13781 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
13782 nonlinear iv_use computation in loop invariant sensitive way.
13784 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13786 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
13787 (find_iv_candidates): Call relate_compare_use_with_all_cands.
13789 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13791 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
13792 (dump_cand): Support iv_cand.inv_exprs.
13793 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
13795 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
13798 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13800 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
13802 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
13803 as local function. Include necessary header files.
13804 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
13806 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13808 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
13810 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13812 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
13813 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
13814 RSHIFT_EXPR and BIT_NOT_EXPR.
13816 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13818 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
13819 (adjust_setup_cost): New parameter supporting round up adjustment.
13820 (struct address_cost_data): Delete.
13821 (force_expr_to_var_cost): Don't bound cost with spill_cost.
13822 (split_address_cost, ptr_difference_cost): Delete.
13823 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
13824 (struct ainc_cost_data): New struct.
13825 (get_address_cost_ainc): New function.
13826 (get_address_cost, get_computation_cost): Reimplement.
13827 (determine_group_iv_cost_address): Record inv_expr for all uses of
13829 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
13830 (iv_ca_has_deps): Reimplemented to ...
13831 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
13833 (iv_ca_extend): Call iv_ca_more_deps.
13835 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13837 * tree-ssa-address.c (struct mem_address): Move to header file.
13838 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
13839 * tree-ssa-address.h (struct mem_address): Move from C file.
13840 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
13842 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13844 * tree-affine.h (aff_combination_type): New interface.
13845 (aff_combination_zero_p): Remove static.
13846 (aff_combination_const_p): New interface.
13847 (aff_combination_singleton_var_p): New interfaces.
13849 2017-05-11 Richard Biener <rguenther@suse.de>
13851 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
13852 Skip unreachable blocks and destinations.
13853 (eliminate): Move stmt removal and fixup ...
13854 (fini_eliminate): ... here. Skip inserted exprs.
13855 (pass_pre::execute): Move fini_pre after fini_eliminate.
13856 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
13857 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
13858 PRE to get rid of dead code that has invalid SSA form and
13859 split critical edges again.
13861 2017-05-11 Bin Cheng <bin.cheng@arm.com>
13863 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
13865 2017-05-11 Richard Biener <rguenther@suse.de>
13867 * passes.c (execute_function_todo): Verify loops if they are
13868 said to be up-to-date.
13869 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
13870 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
13872 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
13875 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
13876 handle calling assemble_external ourself.
13879 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
13880 modes with zero size. Enhance comment.
13882 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13884 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
13885 built-ins for vec_xl and vec_xst with short and char pointer
13888 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13890 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
13891 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
13892 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
13893 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
13894 (_mm_maskz_min_round_ss): New intrinsics.
13895 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
13896 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
13897 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
13898 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
13899 (__builtin_ia32_minss_mask_round): New builtins.
13900 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13901 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13902 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
13904 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
13905 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
13907 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
13910 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
13912 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
13913 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
13914 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
13915 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
13916 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
13917 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
13918 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
13919 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13920 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
13921 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
13922 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
13923 (__builtin_ia32_mulss_mask_round): New builtins.
13924 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
13925 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
13926 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
13928 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
13929 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
13931 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
13934 2017-05-10 Julia Koval <julia.koval@intel.com>
13936 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
13937 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
13938 (_mm256_setr_m128i): New intrinsics.
13940 2017-05-10 Julia Koval <julia.koval@intel.com>
13942 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
13943 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
13944 (_mm_maskz_rcp14_ss): New intrinsics.
13945 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
13946 (__builtin_ia32_rcp14ss_mask): New builtins.
13947 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
13949 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
13951 PR tree-optimization/51513
13952 * tree-cfg.c (gimple_seq_unreachable_p): New function.
13953 (assert_unreachable_fallthru_edge_p): Use it.
13954 (group_case_labels_stmt): Likewise.
13955 * tree-cfg.h: Prototype it.
13956 * stmt.c: Include cfghooks.h and tree-cfg.h.
13957 (emit_case_dispatch_table) <gap_label>: New local variable.
13958 Use it to fill dispatch table gaps.
13959 Test for default_label before updating probabilities.
13960 (expand_case) <default_label>: Remove unneeded initialization.
13961 Test for unreachable default case statement and remove its edge.
13962 Set default_label accordingly.
13963 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
13965 2017-05-10 Carl Love <cel@us.ibm.com>
13967 * config/rs6000/rs6000-c: Add support for built-in functions
13968 vector signed char vec_neg (vector signed char)
13969 vector signed short int vec_neg (vector short int)
13970 vector signed int vec_neg (vector signed int)
13971 vector signed long long vec_neg (vector signed long long)
13972 vector float vec_neg (vector float)
13973 vector double vec_neg (vector double)
13974 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
13976 * config/rs6000/altivec.h: Add define for vec_neg
13977 * doc/extend.texi: Update the built-in documentation for the
13978 new built-in functions.
13980 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13982 PR tree-optimization/77644
13983 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
13985 2017-05-10 Nathan Sidwell <nathan@acm.org>
13987 * dumpfile.h (TDI_lang_all): New.
13988 (TDF_KIND): New. Renumber others
13989 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
13991 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
13993 (get_dump_file_name): Adjust suffix generation.
13994 (dump_enable_all): Use TDF_KIND.
13995 * doc/invoke.texi (-fdump-lang-all): Document.
13997 * dumpfile.h: Tabify.
13999 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
14002 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
14003 Move member access before delete.
14005 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
14007 * tree-inline.c (expand_call_inline): Split block at stmt
14010 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
14013 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
14014 are now unused after splitting mov{sf,sd}_hardfloat.
14015 (f32_lr2): Likewise.
14016 (f32_lm): Likewise.
14017 (f32_lm2): Likewise.
14018 (f32_li): Likewise.
14019 (f32_li2): Likewise.
14020 (f32_lv): Likewise.
14021 (f32_sr): Likewise.
14022 (f32_sr2): Likewise.
14023 (f32_sm): Likewise.
14024 (f32_sm2): Likewise.
14025 (f32_si): Likewise.
14026 (f32_si2): Likewise.
14027 (f32_sv): Likewise.
14028 (f32_dm): Likewise.
14029 (f32_vsx): Likewise.
14030 (f32_av): Likewise.
14031 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
14032 For movsf, order stores so the VSX stores occur before the GPR
14033 store which encourages the register allocator to use a traditional
14034 FPR instead of a GPR. For movsd, order the stores so that the GPR
14035 store comes before the VSX stores to allow the power6 to work.
14036 This is due to the power6 not having a 32-bit integer store
14037 instruction from a FPR.
14038 (movsf_hardfloat): Likewise.
14039 (movsd_hardfloat): Likewise.
14041 2017-05-09 Martin Sebor <msebor@redhat.com>
14043 PR translation/80280
14044 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
14047 PR translation/80280
14048 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
14049 data member added in r247778.
14050 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
14052 2017-05-09 Nathan Sidwell <nathan@acm.org>
14054 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
14056 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
14059 2017-05-09 Marek Polacek <polacek@redhat.com>
14061 * doc/invoke.texi: Fix typo.
14063 2017-05-09 Richard Biener <rguenther@suse.de>
14065 * tree-vrp.c (vrp_val_is_max): Adjust comment.
14066 (vrp_val_is_min): Likewise.
14067 (set_value_range_to_value): Likewise.
14068 (set_value_range_to_nonnegative): Likewise.
14069 (gimple_assign_nonzero_p): Likewise.
14070 (gimple_stmt_nonzero_p): Likewise.
14071 (vrp_int_const_binop): Likewise. Remove unreachable case.
14072 (adjust_range_with_scev): Adjust comments.
14073 (compare_range_with_value): Likewise.
14074 (extract_range_from_phi_node): Likewise.
14075 (test_for_singularity): Likewise.
14077 2017-05-09 Richard Biener <rguenther@suse.de>
14079 * tree-vrp.c (get_single_symbol): Add assert that we don't
14080 get overflowed constants as invariant part.
14081 (compare_values_warnv): Add comment before the TREE_NO_WARNING
14082 checks. Use wi::cmp instead of recursing for integer constants.
14083 (compare_values): Just ignore whether we assumed undefined
14084 overflow instead of failing the compare.
14085 (extract_range_for_var_from_comparison_expr): Add comment before the
14086 TREE_NO_WARNING sets.
14087 (test_for_singularity): Likewise.
14088 (extract_range_from_comparison): Do not disable optimization
14089 when we assumed undefined overflow.
14090 (extract_range_basic): Remove init of unused var.
14092 2017-05-09 Richard Biener <rguenther@suse.de>
14094 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
14095 (extract_range_from_multiplicative_op_1): Adjust.
14096 (extract_range_from_binary_expr_1): Use int_const_binop.
14098 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
14101 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
14102 rs6000_store_data_bypass_p in seven define_bypass directives and
14103 in several comments.
14104 * config/rs6000/rs6000-protos.h: Add prototype for
14105 rs6000_store_data_bypass_p function.
14106 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
14107 function implements slightly different (rs6000-specific) semantics
14108 than store_data_bypass_p, returning false rather than aborting
14109 with assertion error when arguments do not satisfy the
14110 requirements of store data bypass.
14111 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
14112 rs6000_store_data_bypass_p.
14114 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
14116 * config/xtensa/xtensa-protos.h
14117 (xtensa_initial_elimination_offset): New declaration.
14118 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
14119 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
14120 macro definition, add case for FRAME_POINTER_REGNUM when
14121 FRAME_GROWS_DOWNWARD.
14122 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
14123 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
14124 xtensa_initial_elimination_offset.
14126 2017-05-08 Nathan Sidwell <nathan@acm.org>
14128 * doc/invoke.texi: Alphabetize -fdump options.
14130 2017-05-08 Martin Sebor <msebor@redhat.com>
14132 PR translation/80280
14133 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
14135 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
14137 * target.def (compute_frame_layout): New optional target hook.
14138 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
14139 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
14140 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
14142 * reload1.c (verify_initial_elim_offsets): Likewise.
14143 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
14144 (use_simple_return_p): Call arm_compute_frame_layout if needed.
14145 (arm_get_frame_offsets): Split up into this ...
14146 (arm_compute_frame_layout): ... and this function.
14148 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
14150 * config/aarch64/constraints.md (Usa): New constraint.
14151 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
14153 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
14155 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
14156 with_multilib_list after it has been checked.
14158 2017-05-08 Richard Biener <rguenther@suse.de>
14160 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
14161 (bitmap_set_subtract_values): Likewise.
14163 2017-05-08 Richard Biener <rguenther@suse.de>
14165 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
14166 (gimple_assign_nonzero): ... this and remove strict_overflow_p
14168 (gimple_stmt_nonzero_warnv_p): Rename to ...
14169 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
14171 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
14172 (extract_range_basic): Adjust, do not disable propagation on
14173 strict overflow sensitive simplification.
14174 (vrp_visit_cond_stmt): Likewise.
14176 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
14178 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
14179 body size unconditionally.
14181 2017-05-07 Jeff Law <law@redhat.com>
14184 2017-05-06 Jeff Law <law@redhat.com>
14185 PR tree-optimization/78496
14186 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
14189 PR tree-optimization/78496
14190 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
14191 (simplify_stmt_using_ranges): Call it.
14192 (vrp_dom_walker::before_dom_children): Extract equivalences
14193 from an ASSERT_EXPR with an equality comparison against a
14196 2017-05-06 Jeff Law <law@redhat.com>
14198 PR tree-optimization/78496
14199 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
14202 PR tree-optimization/78496
14203 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
14204 (simplify_stmt_using_ranges): Call it.
14205 (vrp_dom_walker::before_dom_children): Extract equivalences
14206 from an ASSERT_EXPR with an equality comparison against a
14209 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
14211 * lra-constraints.c (lra_copy_reg_equiv): New function.
14212 (split_reg): Use it to copy equivalence information from the
14213 original register to the spill register.
14215 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
14217 PR rtl-optimization/75964
14218 * simplify-rtx.c (simplify_const_relational_operation): Remove
14219 invalid handling of comparisons of integer ABS.
14221 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
14223 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
14224 initialize to zero.
14225 (init_regs): Remove declaration.
14226 (function_arg_advance_32): Initialize error_p as boolean variable.
14228 2017-05-05 Nathan Sidwell <nathan@acm.org>
14230 * store-motion.c (remove_reachable_equiv_notes): Reformat long
14231 lines. Use for (;;).
14233 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14235 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
14236 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
14237 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
14238 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
14239 VF=2 that require versioning.
14241 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14243 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
14246 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14248 * diagnostic.h (diagnostic_override_option_index): Convert from
14249 macro to inline function.
14251 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14253 * diagnostic.c (last_module_changed_p): New function.
14254 (set_last_module): New function.
14255 (diagnostic_report_current_module): Convert macro usage to
14256 the above functions.
14257 * diagnostic.h (diagnostic_context::last_module): Strengthen
14258 from const line_map * to const line_map_ordinary *.
14259 (diagnostic_last_module_changed): Delete macro.
14260 (diagnostic_set_last_module): Delete macro.
14262 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14264 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
14265 with diagnostic_report_diagnostic.
14266 (diagnostic_n_impl_richloc): Likewise.
14267 * diagnostic.h (report_diagnostic): Delete macro.
14268 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
14269 with diagnostic_report_diagnostic.
14270 * substring-locations.c (format_warning_va): Likewise.
14272 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14274 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
14275 save/restor of format_spec. Move option-printing code to...
14276 (print_option_information): ...this new function, and
14277 reimplement by simply printing to the pretty_printer,
14278 rather than appending to the format string.
14280 2017-05-05 David Malcolm <dmalcolm@redhat.com>
14282 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
14283 handling logic into...
14284 (update_effective_level_from_pragmas): ...this new function.
14286 2017-05-04 Andrew Waterman <andrew@sifive.com>
14288 * config/riscv/riscv.opt (mstrict-align): New option.
14289 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
14290 (SLOW_UNALIGNED_ACCESS): Define.
14291 (riscv_slow_unaligned_access): Declare.
14292 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
14294 (riscv_slow_unaligned_access): New variable.
14295 (rocket_tune_info): Set slow_unaligned_access to true.
14296 (optimize_size_tune_info): Set slow_unaligned_access to false.
14297 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
14298 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
14299 (riscv_option_override): Set riscv_slow_unaligned_access.
14300 * doc/invoke.texi: Add -mstrict-align to RISC-V.
14302 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
14304 * config/riscv/riscv.md: Unify indentation.
14306 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
14311 * config/rs6000/rs6000.md (u code attribute): Add FIX and
14313 (extendsi<mode>2): Add support for doing sign extension via
14314 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
14315 don't have ISA 3.0 instructions.
14316 (extendsi<mode>2 splitter): Likewise.
14317 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
14318 generate the normal insns since SImode can now go in vector
14319 registers. Disallow the special UNSPECs needed for previous
14320 machines to hide SImode being used. Add new insns
14321 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
14322 (fix_trunc<mode>si2_stfiwx): Likewise.
14323 (fix_trunc<mode>si2_internal): Likewise.
14324 (fixuns_trunc<mode>si2): Likewise.
14325 (fixuns_trunc<mode>si2_stfiwx): Likewise.
14326 (fctiw<u>z_<mode>_smallint): Likewise.
14327 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
14328 of floating point to 32-bit integer from doing a direct move to
14329 the GPR registers to do a store.
14330 (fctiwz_<mode>): Break long line.
14332 2017-05-05 Bin Cheng <bin.cheng@arm.com>
14334 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
14335 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
14336 (addr_list, addr_offset_valid_p): New.
14337 (split_address_groups): Check offset validity with above function.
14338 (gt-tree-ssa-loop-ivopts.h): Include header file.
14340 2017-05-05 Nathan Sidwell <nathan@acm.org>
14342 * config.gcc (arm*-*-*): Add missing 'fi'.
14344 2017-05-05 Steve Ellcey <sellcey@cavium.com>
14346 * doc/invoke.texi (-fopt-info): Explicitly say order of options
14347 included in -fopt-info does not matter.
14348 * doc/optinfo.texi (-fopt-info): Fix description of default
14349 behavour. Explicitly say order of options included in -fopt-info
14352 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
14354 * config.gcc: Allow combinations of aprofile and rmprofile values for
14355 --with-multilib-list.
14356 * config/arm/t-multilib: New file.
14357 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
14358 variables. Remove setting of ISA and floating-point ABI in
14359 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
14360 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
14361 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
14362 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
14364 * config/arm/t-rmprofile: Likewise except for the matches changes.
14365 * doc/install.texi (--with-multilib-list): Document the combination of
14366 aprofile and rmprofile values and warn about pitfalls in doing that.
14368 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
14370 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
14371 (movdi_aarch64): Likewise.
14373 2017-05-05 Jakub Jelinek <jakub@redhat.com>
14375 PR tree-optimization/80632
14376 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
14378 (build_arrays): Initialize it for virtual phis.
14379 (fix_phi_nodes): Use it for virtual phis.
14381 PR tree-optimization/80558
14382 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
14383 [x, y] op z into [x op, y op z] for op & or | if conditions
14386 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
14387 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14390 * config/arm/arm.md (use_literal_pool): Remove.
14391 (64-bit immediate split): No longer takes cost into consideration
14392 if arm_disable_literal_pool is enabled.
14393 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
14394 used when arm_disable_literal_pool is enabled.
14395 (arm_max_const_double_inline_cost): Remove use of
14396 arm_disable_literal_pool.
14397 (push_minipool_fix): Add assert.
14398 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
14399 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
14400 (no_literal_pool_sf_immediate): New.
14402 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14404 PR tree-optimization/80613
14405 * tree-ssa-dce.c (propagate_necessity): Remove cases for
14406 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
14408 2017-05-05 Richard Biener <rguenther@suse.de>
14410 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
14412 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
14414 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
14415 of this flag from insn conditions due to removal from r247495.
14417 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
14419 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
14421 (arm_early_store_addr_dep_ptr): Likewise.
14422 * config/arm/aarch-common-protos.h
14423 (arm_early_load_addr_dep_ptr): Add prototype.
14424 (arm_early_store_addr_dep_ptr): Likewise.
14425 * config/arm/cortex-a53.md: Add new bypasses.
14427 2017-05-05 Jakub Jelinek <jakub@redhat.com>
14429 * tree.c (next_type_uid): Change type to unsigned.
14430 (type_hash_canon): Decrement back next_type_uid if
14431 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
14432 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
14435 2017-05-04 Martin Sebor <msebor@redhat.com>
14437 * builtins.c: Fix a trivial typo in a comment.
14439 PR middle-end/79234
14440 * builtins.c (check_sizes): Adjust to handle reading past the end.
14441 Avoid printing excessive upper bound of ranges. Use %E to print
14442 tree nodes instead of converting them to %wu.
14443 (expand_builtin_memchr): New function.
14444 (compute_dest_size): Rename...
14445 (compute_objsize): ...to this.
14446 (expand_builtin_memcpy): Adjust.
14447 (expand_builtin_mempcpy): Adjust.
14448 (expand_builtin_strcat): Adjust.
14449 (expand_builtin_strcpy): Adjust.
14450 (check_strncat_sizes): Adjust.
14451 (expand_builtin_strncat): Adjust.
14452 (expand_builtin_strncpy): Adjust and simplify.
14453 (expand_builtin_memset): Adjust.
14454 (expand_builtin_bzero): Adjust.
14455 (expand_builtin_memcmp): Adjust.
14456 (expand_builtin): Handle memcmp.
14457 (maybe_emit_chk_warning): Check strncat just once.
14459 2017-05-04 Martin Sebor <msebor@redhat.com>
14461 PR preprocessor/79214
14462 PR middle-end/79222
14463 PR middle-end/79223
14464 * builtins.c (check_sizes): Add inlining context and issue
14465 warnings even when -Wno-system-headers is set.
14466 (check_strncat_sizes): Same.
14467 (expand_builtin_strncat): Same.
14468 (expand_builtin_memmove): New function.
14469 (expand_builtin_stpncpy): Same.
14470 (expand_builtin): Handle memmove and stpncpy.
14472 2017-05-04 Bin Cheng <bin.cheng@arm.com>
14474 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
14475 which is not used any more.
14477 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14479 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
14481 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14483 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
14484 (cortexa53_tunings): Likewise.
14485 (cortexa57_tunings): Likewise.
14486 (cortexa72_tunings): Likewise.
14487 (cortexa73_tunings): Likewise.
14489 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14491 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
14492 Set loop alignment to 8.
14494 2017-05-04 Martin Sebor <msebor@redhat.com>
14496 PR translation/80280
14497 * builtins.c (expand_builtin_object_size): Add missing quoting to
14498 %D and like directives.
14499 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
14500 (hsa_type_for_tree_type): Same.
14501 (verify_function_arguments): Same.
14502 * symtab.c (symbol_table::change_decl_assembler_name): Same.
14503 * varasm.c (get_section): Same.
14506 2017-05-04 Martin Sebor <msebor@redhat.com>
14508 PR translation/80280
14509 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
14511 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
14513 * config/aarch64/aarch64.c (generic_addrcost_table):
14514 Change HI/TI mode setting.
14516 2017-05-04 Martin Jambor <mjambor@suse.cz>
14518 PR tree-optimization/80622
14519 * tree-sra.c (comes_initialized_p): New function.
14520 (build_accesses_from_assign): Only set write lazily when
14521 comes_initialized_p is false.
14522 (analyze_access_subtree): Use comes_initialized_p.
14523 (propagate_subaccesses_across_link): Assert !comes_initialized_p
14524 instead of testing for PARM_DECL.
14526 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14528 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
14529 constraint on operand 0 to allow more general addressing modes.
14530 Adjust output template.
14531 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
14533 * config/aarch64/aarch64-protos.h
14534 (aarch64_address_valid_for_prefetch_p): Declare prototype.
14535 * config/aarch64/constraints.md (Dp): New address constraint.
14536 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
14539 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
14541 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
14542 update use of estimate_ipcp_clone_size_and_time.
14543 (estimate_local_effects): Update use of
14544 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
14545 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
14546 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
14547 Return nonspecialized time.
14549 2017-05-04 Richard Biener <rguenther@suse.de>
14551 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
14552 for the last VUSE which def dominates the PHI. Directly call
14554 (get_continuation_for_phi_1): Remove.
14556 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
14558 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
14559 to explain the use of truncating division. Cap the number of
14560 iterations to the maximum given by nb_iterations_upper_bound,
14563 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
14565 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
14566 * configure: Regenerate.
14567 * config.in: Regenerate.
14568 * config/i386/driver-mingw32.c: new file.
14569 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
14570 * config.host: Link driver-mingw32.o on MinGW host.
14571 * doc/install.texi: Document new --enable-mingw-wildcard configure
14574 2017-05-04 Marek Polacek <polacek@redhat.com>
14576 PR tree-optimization/80612
14577 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
14579 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14580 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
14582 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
14583 (movt splitter): Likewise.
14584 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
14585 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
14586 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
14587 block for Thumb-1 with MOVT.
14588 (thumb2_legitimate_address_p): Move code block ...
14589 (can_avoid_literal_pool_for_label_p): ... into this new function.
14590 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
14592 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
14593 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
14594 "M-profile targets with the MOVT instruction".
14596 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
14598 * config/arm/arm-builtins.c (arm_init_builtins): Rename
14599 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
14600 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
14602 2017-05-04 Martin Liska <mliska@suse.cz>
14604 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
14605 variable cond_code.
14607 2017-05-04 Richard Biener <rguenther@suse.de>
14609 * tree.c (array_at_struct_end_p): Handle arrays at struct
14610 end with flexarrays more conservatively. Refactor and treat
14611 arrays of arrays or aggregates more strict. Fix
14612 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
14613 * tree.c (array_at_struct_end_p): Adjust prototype.
14614 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
14615 * gimple-fold.c (get_range_strlen): Likewise.
14616 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
14618 2017-05-04 Richard Biener <rguenther@suse.de>
14620 PR tree-optimization/31130
14621 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
14623 (supports_overflow_infinity): Likewise.
14624 (is_negative_overflow_infinity): Likewise.
14625 (is_positive_overflow_infinity): Likewise.
14626 (is_overflow_infinity): Likewise.
14627 (stmt_overflow_infinity): Likewise.
14628 (overflow_infinity_range_p): Likewise.
14629 (usable_range_p): Remove as always returning true.
14630 (make_overflow_infinity): Remove.
14631 (negative_overflow_infinity): Likewise.
14632 (positive_overflow_infinity): Likewise.
14633 (avoid_overflow_infinity): Likewise.
14634 (set_value_range): Adjust accordingly.
14635 (set_value_range_to_nonnegative): Likewise, remove now unused
14636 overflow_infinity arg.
14637 (vrp_operand_equal_p): Adjust.
14638 (update_value_range): Likewise.
14639 (range_int_cst_singleton_p): Likewise.
14640 (operand_less_p): Likewise.
14641 (compare_values_warnv): Likewise.
14642 (extract_range_for_var_from_comparison_expr): Likewise.
14643 (vrp_int_const_binop): Likewise.
14644 (zero_nonzero_bits_from_vr): Likewise.
14645 (extract_range_from_multiplicative_op_1): Likewise.
14646 (extract_range_from_binary_expr_1): Likewise.
14647 (extract_range_from_unary_expr): Likewise.
14648 (extract_range_from_comparison): Likewise.
14649 (extract_range_basic): Likewise.
14650 (adjust_range_with_scev): Likewise.
14651 (compare_ranges): Likewise.
14652 (compare_range_with_value): Likewise.
14653 (dump_value_range): Likewise.
14654 (test_for_singularity): Likewise, remove strict_overflow_p parameter
14656 (simplify_cond_using_ranges): Adjust.
14658 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
14660 * brig-builtins.def: Added a builtin for class_f64.
14661 * builtin-types.def: Added a builtin type needed by class_f64.
14663 2017-05-03 Jason Merrill <jason@redhat.com>
14665 * timevar.def: Add TV_CONSTEXPR.
14667 2017-05-03 David Malcolm <dmalcolm@redhat.com>
14669 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
14671 2017-05-03 Martin Jambor <mjambor@suse.cz>
14673 * ipa-prop.c (ipa_update_after_lto_read): Removed.
14674 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
14675 * ipa-cp.c (ipcp_propagate_stage): Do not call
14676 ipa_update_after_lto_read.
14677 * ipa-inline.c (ipa_inline): Likewise.
14679 2017-05-03 Martin Jambor <mjambor@suse.cz>
14681 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
14682 tag. Added a default constructor and a destructor.
14683 (ipa_edge_args_sum_t): New class;
14684 (ipa_edge_args_sum): Declare.
14685 (ipa_edge_args_vector): Remove declaration.
14686 (IPA_EDGE_REF): Use ipa_edge_args_sum.
14687 (ipa_free_edge_args_substructures): Remove declaration.
14688 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
14689 (ipa_edge_args_info_available_for_edge_p): Likewise.
14690 * ipa-prop.c (ipa_edge_args_vector): Removed.
14691 (edge_removal_hook_holder): Likewise.
14692 (edge_duplication_hook_holder): Likewise.
14693 (ipa_edge_args_sum): New variable.
14694 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
14695 ipa_edge_args_vector.
14696 (ipa_free_edge_args_substructures): Likewise.
14697 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
14698 ipa_edge_args_vector.
14699 (ipa_edge_removal_hook): Turned into method
14700 ipa_edge_args_sum_t::remove.
14701 (ipa_edge_duplication_hook): Turned into method
14702 ipa_edge_args_sum_t::duplicate.
14703 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
14704 registering edge hooks.
14705 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
14706 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
14707 ipa_edge_args_sum instead of ipa_edge_args_vector.
14708 * ipa-profile.c (ipa_profile): Likewise.
14710 2017-05-03 Martin Jambor <mjambor@suse.cz>
14712 * symbol-summary.h (function_summary): New method exists.
14713 (function_summary::symtab_removal): Deallocate through release.
14714 (call_summary): New class.
14715 (gt_ggc_mx): New overload.
14716 (gt_pch_nx): Likewise.
14717 (gt_pch_nx): Likewise.
14719 2017-05-03 Jeff Law <law@redhat.com>
14721 PR tree-optimization/78496
14722 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
14723 from simplify_cond_using_ranges. Split off code to walk
14724 backwards through casts into ...
14725 (simplify_cond_using_ranges_2): New function.
14726 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
14727 (execute_vrp): After identifying jump threads, call
14728 simplify_cond_using_ranges_2.
14730 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
14733 * ipa-inline.h (inline_summary): Add ctor.
14734 (create_ggc): Do not use ggc_cleared_alloc.
14736 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
14737 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14739 * gcc.c (handle_braces): Support escaping in switch matching
14741 * doc/invoke.texi (Spec Files): Document it.
14742 Remove superfluous @code markup in items.
14744 2017-05-03 David Malcolm <dmalcolm@redhat.com>
14746 * diagnostic-show-locus.c (struct column_range): New struct.
14747 (get_affected_columns): New function.
14748 (get_printed_columns): New function.
14749 (struct correction): New struct.
14750 (correction::ensure_capacity): New function.
14751 (correction::ensure_terminated): New function.
14752 (struct line_corrections): New struct.
14753 (line_corrections::~line_corrections): New dtor.
14754 (line_corrections::add_hint): New function.
14755 (layout::print_trailing_fixits): Reimplement in terms of the new
14757 (selftest::test_overlapped_fixit_printing): New function.
14758 (selftest::diagnostic_show_locus_c_tests): Call it.
14760 2017-05-03 Nathan Sidwell <nathan@acm.org>
14762 Canonicalize canonical type hashing
14763 * tree.h (type_hash_canon_hash): Declare.
14764 * tree.c (type_hash_list, attribute_hash_list): Move into
14765 type_hash_canon_hash.
14766 (build_type_attribute_qual_variant): Break out hash code calc into
14767 type_hash_canon_hash.
14768 (type_hash_canon_hash): New. Generic type hash computation.
14769 (build_range_type_1, build_array_type_1, build_function_type,
14770 build_method_type_directly, build_offset_type, build_complex_type,
14771 make_vector_type): Call it.
14773 2017-05-03 Richard Biener <rguenther@suse.de>
14775 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
14776 When all DRs have unknown misaligned do not always peel
14777 when there is a store but apply the same costing model as if
14778 there were only loads.
14780 2017-05-03 Richard Biener <rguenther@suse.de>
14783 PR tree-optimization/80492
14784 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
14785 compare_base_decls returning dont-know properly.
14787 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
14789 * config/arm/iterators.md (CCSI): New mode iterator.
14790 (arch): New mode attribute.
14791 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
14792 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
14793 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
14794 code iterator for success result mode.
14795 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
14796 the corresponding new insn generators.
14798 2017-05-03 Bin Cheng <bin.cheng@arm.com>
14801 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14802 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14804 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
14806 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
14807 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
14808 (DDR_A): Wrap DDR argument in brackets.
14809 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
14810 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
14811 (DDR_REVERSED_P): Likewise.
14813 2017-05-03 Jakub Jelinek <jakub@redhat.com>
14815 PR tree-optimization/79472
14816 * tree-switch-conversion.c (struct switch_conv_info): Add
14817 contiguous_range and default_case_nonstandard fields.
14818 (collect_switch_conv_info): Compute contiguous_range and
14819 default_case_nonstandard fields, don't clear final_bb if
14820 contiguous_range and only the default case doesn't have the required
14822 (check_all_empty_except_final): Set default_case_nonstandard instead
14823 of failing if contiguous_range and the default case doesn't have empty
14825 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
14826 and only the default case doesn't have the required constants. Skip
14828 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
14829 if default_case_nonstandard.
14830 (build_constructors): Build constant 1 just once. Assert that default
14831 values aren't inserted in between cases if contiguous_range. Skip
14833 (build_arrays): Skip virtual phis.
14834 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
14835 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
14836 Handle virtual phis.
14837 (gen_inbound_check): Handle default_case_nonstandard case.
14838 (process_switch): Adjust check_final_bb caller. Call
14839 gather_default_values with the first non-default case instead of
14840 default case if default_case_nonstandard.
14842 2017-05-02 Nathan Sidwell <nathan@acm.org>
14844 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
14845 check. Fix formatting.
14847 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
14849 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
14850 errors when comparing specialized and unspecialized times.
14852 2017-05-02 David Malcolm <dmalcolm@redhat.com>
14854 * diagnostic-show-locus.c
14855 (layout::should_print_annotation_line_p): Make private.
14856 (layout::print_annotation_line): Make private.
14857 (layout::annotation_line_showed_range_p): Make private.
14858 (layout::show_ruler): Make private.
14859 (layout::print_source_line): Make private. Pass in line and
14860 line_width, rather than calling location_get_source_line. Drop
14862 (layout::print_leading_fixits): New method.
14863 (layout::print_any_fixits): Rename to...
14864 (layout::print_trailing_fixits): ...this, and make private.
14865 Don't print newline fixits.
14866 (diagnostic_show_locus): Move logic for printing one row into...
14867 (layout::print_line): ...this new function. Move the
14868 location_get_source_line call and error-handling from
14869 print_source_line to here. Call print_leading_fixits, and rename
14870 print_any_fixits to print_trailing_fixits.
14871 (selftest::test_fixit_insert_containing_newline): Update now that
14872 newlines are partially supported.
14873 (selftest::test_fixit_insert_containing_newline_2): New test.
14874 (selftest::test_fixit_replace_containing_newline): Update comments.
14875 (selftest::diagnostic_show_locus_c_tests): Call the new test.
14876 * edit-context.c (class added_line): New class.
14877 (class edited_line): Describe newline handling in comment.
14878 (edited_line::actually_edited_p): New method.
14879 (edited_line::print_content): Delete redundant decl.
14880 (edited_line::m_predecessors): New field.
14881 (edited_file::print_content): Call edited_line::print_content.
14882 (edited_file::print_diff): Update to support newlines.
14883 (edited_file::print_diff_hunk): Likewise.
14884 (edited_file::print_run_of_changed_lines): New function.
14885 (edited_file::print_diff_line): Convert to...
14886 (print_diff_line): ...this.
14887 (edited_file::get_effective_line_count): New function.
14888 (edited_line::edited_line): Initialize new field m_predecessors.
14889 (edited_line::~edited_line): Clean up m_predecessors.
14890 (edited_line::apply_fixit): Handle newlines.
14891 (edited_line::get_effective_line_count): New function.
14892 (edited_line::print_content): New function.
14893 (edited_line::print_diff_lines): New function.
14894 (selftest::test_applying_fixits_insert_containing_newline): New
14896 (selftest::test_applying_fixits_replace_containing_newline): New
14898 (selftest::insert_line): New function.
14899 (selftest::test_applying_fixits_multiple_lines): Add example of
14901 (selftest::edit_context_c_tests): Call the new tests.
14903 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14905 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
14906 parameter cand. Update dump information.
14907 (get_computation_cost): Update uses.
14909 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14911 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
14912 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
14913 (get_computation_at, rewrite_use_address): Update use of
14914 get_computation_aff.
14916 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14918 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
14919 (get_computation): Delete.
14920 (get_computation_cost): Implement like get_computation_cost_at.
14921 Use get_computation_at.
14922 (get_computation_cost_at): Delete.
14923 (rewrite_use_nonlinear_expr): Use get_computation_at.
14924 (rewrite_use_compare, remove_unused_ivs): Ditto.
14926 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14928 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
14930 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14932 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
14933 (ivopts_global_cost_for_size): Rename parameter and update uses.
14934 (iv_ca_recount_cost): Update uses.
14935 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
14936 candidates seperately in n_invs and n_cands.
14937 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
14939 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14941 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
14942 (find_inv_vars_cb): New.
14943 (find_depends): Renamed to ...
14944 (find_inv_vars): ... this.
14945 (add_candidate_1, force_var_cost): Call find_inv_vars.
14946 (split_address_cost, determine_group_iv_cost_cond): Ditto.
14948 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14950 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
14951 inv_vars. Add inv_exprs.
14952 (struct iv_cand): Rename depends_on to inv_vars.
14953 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
14954 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
14955 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
14956 (dump_cand): Dump inv_vars.
14957 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
14958 (record_invariant, find_depends, add_candidate_1): Ditto.
14959 (set_group_iv_cost, force_var_cost): Ditto.
14960 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
14961 (get_computation_cost_at, get_computation_cost): Ditto.
14962 (determine_group_iv_cost_generic): Ditto.
14963 (determine_group_iv_cost_address): Ditto.
14964 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
14965 (determine_group_iv_costs): Ditto.
14966 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
14967 (iv_ca_set_remove_invariants): Renamed to ...
14968 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
14969 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
14970 (iv_ca_set_add_invariants): Renamed to ...
14971 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
14972 (iv_ca_set_cp): Use iv_ca_set_add_invs.
14973 (iv_ca_has_deps): Support inv_vars and inv_exprs.
14974 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
14975 (create_new_ivs): Remove useless dump.
14977 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14979 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
14981 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
14982 (iv_ca_set_no_cp, create_new_iv): Ditto.
14984 2017-05-02 Bin Cheng <bin.cheng@arm.com>
14986 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
14988 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
14990 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
14991 function tree_check2.
14993 2017-05-02 Martin Liska <mliska@suse.cz>
14995 * doc/gcov.texi: Add missing preposition.
14996 * gcov.c (function_info::function_info): Properly fill up
14997 all member variables.
14999 2017-05-02 Tamar Christina <tamar.christina@arm.com>
15001 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
15003 2017-05-02 Tamar Christina <tamar.christina@arm.com>
15005 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
15007 2017-05-02 Martin Liska <mliska@suse.cz>
15010 * lto-streamer-in.c (lto_read_tree_1): Remove
15011 LTO_STREAMER_DEBUG.
15012 * lto-streamer.c (struct tree_hash_entry): Likewise.
15013 (struct tree_entry_hasher): Likewise.
15014 (tree_entry_hasher::hash): Likewise.
15015 (tree_entry_hasher::equal): Likewise.
15016 (lto_streamer_init): Likewise.
15017 (lto_orig_address_map): Likewise.
15018 (lto_orig_address_get): Likewise.
15019 (lto_orig_address_remove): Likewise.
15020 * lto-streamer.h: Likewise.
15021 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
15022 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
15024 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
15026 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
15027 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
15028 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
15029 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
15030 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
15031 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
15032 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
15033 (mm_maskz_sub_ss): New intrinsics.
15034 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15035 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
15036 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
15037 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
15038 (__builtin_ia32_subss_mask_round): New builtins.
15039 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
15040 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
15041 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
15043 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
15044 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
15046 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
15049 2017-05-02 Martin Jambor <mjambor@suse.cz>
15051 PR tree-optimization/78687
15052 * tree-sra.c (access): New field parent.
15053 (process_subtree_disqualification): New function.
15054 (disqualify_candidate): Call it.
15055 (build_accesses_from_assign): Reset write flag if creating an
15057 (build_access_subtree): Fill in parent field and also prpagate
15058 down grp_write flag.
15059 (create_artificial_child_access): New parameter set_grp_write, set
15060 grp_write to its value.
15061 (propagate_subaccesses_across_link): Also propagate grp_write flag
15063 (propagate_all_subaccesses): Push the closest parent back to work
15064 queue if add_access_to_work_queue returned true.
15066 2017-05-02 Richard Biener <rguenther@suse.de>
15068 * common.opt (fstrict-overflow): Alias negative to fwrapv.
15069 * doc/invoke.texi (fstrict-overflow): Remove all traces of
15070 -fstrict-overflow documentation.
15071 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
15072 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
15073 flag_strict_overflow.
15074 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
15075 * lto-opts.c (lto_write_options): Do not stream it.
15076 * lto-wrapper.c (merge_and_complain): Do not handle it.
15077 * opts.c (default_options_table): Do not set -fstrict-overflow.
15078 (finish_options): Likewise do not clear it when sanitizing.
15079 * simplify-rtx.c (simplify_const_relational_operation): Do not
15080 test flag_strict_overflow.
15082 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
15084 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
15085 using enabled attribute.
15086 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
15087 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
15088 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
15089 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
15090 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
15091 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
15092 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
15093 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
15094 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
15095 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
15097 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
15099 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
15101 2017-05-02 Richard Biener <rguenther@suse.de>
15103 PR tree-optimization/80591
15105 2017-04-10 Richard Biener <rguenther@suse.de>
15107 * tree-ssa-structalias.c (find_func_aliases): Properly handle
15110 2017-05-02 Richard Biener <rguenther@suse.de>
15112 PR tree-optimization/80549
15113 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
15114 (cleanup_tree_cfg_noloop): Create forwarders to known loop
15115 headers if they do not have a preheader.
15117 2017-05-02 Martin Liska <mliska@suse.cz>
15120 * common.opt: Fix typo.
15121 * doc/invoke.texi: Likewise.
15123 2017-05-01 Jan Beulich <jbeulich@suse.com>
15125 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
15126 swapping, add (x,x,m,x,n) alternative.
15128 2017-05-01 Nathan Sidwell <nathan@acm.org>
15130 * calls.c (combine_pending_stack_adjustment_and_call): Remove
15131 unnecessary unadjusted_alignment check.
15133 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
15136 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
15138 * gimplify.c (gimplify_cilk_detach): New function.
15139 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
15140 * tree-core.h: Document EXPR_CILK_SPAWN.
15141 * tree.h (EXPR_CILK_SPAWN): Define.
15143 2017-05-01 David Malcolm <dmalcolm@redhat.com>
15145 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
15146 to use new fixit_hint representation, using the "replace" logic.
15147 (get_line_span_for_fixit_hint): Likewise.
15148 (layout::print_any_fixits): Likewise.
15149 (selftest::test_one_liner_many_fixits): Rename to...
15150 (selftest::test_one_liner_many_fixits_1): ...this, and update
15151 comment and expected output to reflect that the multiple fix-it
15152 hints are now consolidated into one insertion.
15153 (selftest::test_one_liner_many_fixits_2): New test.
15154 (selftest::test_diagnostic_show_locus_one_liner): Update for
15156 (selftest::test_fixit_consolidation): Update for fix-it API
15158 * diagnostic.c (print_parseable_fixits): Likewise.
15159 * edit-context.c (edited_line::m_line_events): Convert from
15160 auto_vec <line_event *> to auto_vec <line_event>.
15161 (class line_event): Convert from abstract base class to a concrete
15162 class, taking over the role of replace_event.
15163 (class insert_event): Delete.
15164 (class replace_event): Rename to class line_event. Convert to
15166 (edit_context::add_fixits): Reimplement.
15167 (edit_context::apply_insert): Delete.
15168 (edit_context::apply_replace): Rename to...
15169 (edit_context::apply_fixit): ...this. Convert to half-open range.
15170 (edited_file::apply_insert): Delete.
15171 (edited_file::apply_replace): Rename to...
15172 (edited_file::apply_fixit): ...this.
15173 (edited_line::~edited_line): Drop deletion of events.
15174 (edited_line::apply_insert): Delete.
15175 (edited_line::apply_replace): Rename to...
15176 (edited_line::apply_fixit): ...this. Convert to half-open range.
15177 Update for change to type of m_line_events.
15178 * edit-context.h (edit_context::apply_insert): Delete.
15179 (edit_context::apply_replace): Rename to...
15180 (edit_context::apply_fixit): ...this.
15182 2017-05-01 Martin Sebor <msebor@redhat.com>
15184 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
15187 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
15190 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
15191 __get_cpuid_max returns 0.
15192 (__get_cpuid_count): Ditto.
15194 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
15196 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
15197 replacement expression is another instance of one of its arguments.
15199 2017-05-01 Jakub Jelinek <jakub@redhat.com>
15202 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
15203 check for stack push/pop autoinc.
15204 * config/i386/i386.c (ix86_agi_dependent): Return false
15205 if the only reason why modified_in_p returned true is that
15206 addr is SP based and set_insn is a push or pop.
15208 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
15210 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
15213 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
15216 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
15217 (account_size_time): Use two predicates - exec_pred and
15219 (evaluate_conditions_for_known_args): Compute both clause and
15221 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
15222 (inline_summary_t::duplicate): Update.
15223 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
15225 (compute_inline_parameters): Likewise.
15226 (estimate_edge_size_and_time): Update caluclation of time.
15227 (estimate_node_size_and_time): Compute both time and nonspecialized
15229 (estimate_ipcp_clone_size_and_time): Update.
15230 (inline_merge_summary): Update.
15231 (do_estimate_edge_time): Update.
15232 (do_estimate_edge_size): Update.
15233 (do_estimate_edge_hints): Update.
15234 (inline_read_section, inline_write_summary): Stream both new predicates.
15235 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
15237 (compute_inlined_call_time): Cleanup.
15238 (big_speedup_p): Update.
15239 (edge_badness): Update.
15240 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
15241 (size_time_entry): Replace predicate by exec_predicate and
15242 nonconst_predicate.
15243 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
15244 (estimate_edge_time): Return also nonspec_time.
15245 (reset_edge_growth_cache): Update.
15247 2017-04-29 Jakub Jelinek <jakub@redhat.com>
15249 PR rtl-optimization/80491
15250 * ifcvt.c (noce_process_if_block): When looking for x setter
15251 with missing else_bb, don't check only the insn right before
15252 cond_earliest, but look for the last insn that x is modified in
15253 within the same bb.
15255 PR rtl-optimization/80491
15256 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
15258 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
15260 PR tree-optimization/80487
15261 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
15263 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15265 PR tree-optimization/79697
15266 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
15267 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
15268 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
15270 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
15271 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
15273 2017-04-28 Martin Sebor <msebor@redhat.com>
15275 PR tree-optimization/80523
15276 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
15277 (init_target_to_host_charmap, target_to_host, target_strtol10): New
15279 (maybe_warn, format_directive, parse_directive): Use new functions.
15280 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
15282 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
15284 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
15286 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
15288 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
15289 target_header_dir): Set correctly.
15290 * configure: Regenerated.
15291 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
15292 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
15293 instead of SYSTEM_HEADER_DIR.
15295 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
15297 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
15298 (estimate_local_effects): Likewise.
15299 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
15300 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
15301 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
15302 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
15303 do_estimate_edge_time, estimate_edge_time): Likewise.
15304 * ipa-inline-analysis.c (estimate_node_size_and_time,
15305 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
15306 (estimate_time_after_inlining): Remove.
15308 2017-04-28 Martin Liska <mliska@suse.cz>
15310 * doc/gcov.texi: Enhance documentation of gcov.
15312 2017-04-28 Martin Liska <mliska@suse.cz>
15314 * doc/gcov.texi: Sort options in alphabetic order.
15315 * doc/gcov-dump.texi: Likewise.
15316 * doc/gcov-tool.texi: Likewise.
15317 * gcov.c (print_usage): Likewise.
15318 * gcov-dump.c (print_usage): Likewise.
15319 * gcov-tool.c (print_merge_usage_message): Likewise.
15320 (print_rewrite_usage_message): Likewise.
15321 (print_overlap_usage_message): Likewise.
15323 2017-04-28 Martin Liska <mliska@suse.cz>
15325 PR gcov-profile/53915
15326 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
15328 2017-04-28 Martin Liska <mliska@suse.cz>
15330 PR gcov-profile/79891
15331 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
15332 is marked by compiler as living on a line.
15333 (get_cycles_count): Remove usage of the union.
15334 (output_intermediate_file): Likewise.
15335 (find_source): Fix GNU coding style.
15336 (accumulate_line_counts): Remove old non-all block mode.
15337 (output_lines): Remove usage of the union.
15338 * profile.c (output_location): Include all BBs, even if
15339 belonging to a same line (and file) as a previous BB.
15341 2017-04-28 Martin Liska <mliska@suse.cz>
15343 * gcov.c (process_args): Handle new argument 'w'.
15344 (read_graph_file): Assign ID to BBs.
15345 (output_branch_count): Display BB # if verbose flag is set.
15346 (output_lines): Likewise for arcs.
15347 (print_usage): Add '--verbose' option help.
15348 * doc/gcov.texi: Document --verbose (-w) option.
15350 2017-04-28 Martin Liska <mliska@suse.cz>
15352 * gcov.c (struct block_location_info): New struct.
15353 (process_file): Fill up the new structure.
15354 (read_graph_file): Replace usage of encoding by the newly added
15356 (add_line_counts): Likewise.
15357 (accumulate_line_counts): Remove usage of the union.
15358 (function_info::function_info): New function.
15359 (function_info::~function_info): Likewise.
15360 (process_file): Call delete instead of release_function.
15361 (release_function): Release the function.
15362 (release_structures): Call delete instead of release_function.
15363 (solve_flow_graph): Replace usage of num_blocks.
15364 (find_exception_blocks): Likewise.
15365 (output_lines): Fix GNU coding style.
15367 2017-04-28 Martin Liska <mliska@suse.cz>
15370 * coverage.c (coverage_remove_note_file): New function.
15371 * coverage.h: Declare the function.
15372 * toplev.c (finalize): Clean if an error has been seen.
15374 2017-04-28 Martin Liska <mliska@suse.cz>
15376 PR gcov-profile/80031
15377 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
15378 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
15379 * gcov.c (read_graph_file): Read just number of blocks.
15380 * profile.c (branch_prob): Do not stream 0 flags per a basic
15383 2017-04-28 Martin Liska <mliska@suse.cz>
15385 * gcov-dump.c (tag_*): Add new argument to declarations.
15386 (dump_gcov_file): Likewise.
15387 (tag_blocks): Add and use new argument depth.
15388 (tag_arcs): Likewise.
15389 (tag_lines): Likewise.
15390 (tag_counters): Likewise.
15391 (tag_summary): Likewise.
15392 (dump_working_sets): Use depth to do a proper indentation.
15394 2017-04-28 Jakub Jelinek <jakub@redhat.com>
15397 * cgraph.h (symtab_node::debug_symtab): No longer inline.
15398 * symtab.c (symtab_node::debug_symtab): Move definition here.
15400 2017-04-28 Richard Biener <rguenther@suse.de>
15402 * lto-streamer.h (LTO_major_version): Bump to 7.
15404 2017-04-28 Richard Biener <rguenther@suse.de>
15406 * tree-vrp.c (assert_info): New struct.
15407 (add_assert_info): New helper.
15408 (register_edge_assert_for_2): Refactor to add asserts to a vector
15410 (register_edge_assert_for_1): Likewise.
15411 (register_edge_assert_for): Likewise.
15412 (finish_register_edge_assert_for): New helper actually registering
15413 asserts where live on edge.
15414 (find_conditional_asserts): Adjust.
15415 (find_switch_asserts): Likewise.
15416 (evrp_dom_walker::try_find_new_range): Generalize.
15417 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
15419 2017-04-27 Marek Polacek <polacek@redhat.com>
15422 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
15423 arg10 and arg11 to itype.
15425 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
15427 * doc/extend.texi (Object Size Checking): Improve grammar.
15429 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
15432 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
15433 that the logic for permitting reciprocal estimates matches that
15436 2017-04-27 Jakub Jelinek <jakub@redhat.com>
15439 * tree.c (type_cache_hasher::equal): Only compare
15440 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
15441 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
15442 non-aggregate element types.
15443 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
15444 about the flag on ARRAY_TYPEs in the comment, formatting fix.
15446 2017-04-27 Richard Biener <rguenther@suse.de>
15448 PR middle-end/80533
15449 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
15450 stripping ARRAY_REFs from MEM_EXPR make sure we're not
15451 keeping a reference to a trailing array.
15453 2017-04-27 Richard Biener <rguenther@suse.de>
15455 PR middle-end/80539
15456 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
15457 being in loop-closed SSA form conservatively.
15458 (chrec_fold_multiply_poly_poly): Likewise.
15460 2017-04-27 Tamar Christina <tamar.christina@arm.com>
15462 PR middle-end/79665
15463 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
15464 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
15466 2017-04-27 Jakub Jelinek <jakub@redhat.com>
15469 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
15470 (aarch64_function_arg_alignment): Return unsigned int again, but still
15471 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
15472 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
15473 Don't emit -Wpsabi note.
15474 (aarch64_function_arg_boundary): Likewise.
15475 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
15478 2017-04-26 Nathan Sidwell <nathan@acm.org>
15480 * tree.h (crc32_unsigned_n): Declare.
15481 (crc32_unsigned, crc32_unsigned): Make inline.
15482 * tree.c (crc32_unsigned_bits): Replace with ...
15483 (crc32_unsigned_n): ... this.
15484 (crc32_unsigned, crc32_byte): Remove.
15485 (crc32_string): Remove unnecessary braces.
15487 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15489 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
15490 * ipa-inline-analysis.c (MAX_TIME): Remove.
15491 (account_size_time): Use sreal for time.
15492 (dump_inline_summary): Update.
15493 (estimate_function_body_sizes): Update.
15494 (estimate_edge_size_and_time): Update.
15495 (estimate_calls_size_and_time): Update.
15496 (estimate_node_size_and_time): Update.
15497 (inline_merge_summary): Update.
15498 (inline_update_overall_summary): Update.
15499 (estimate_time_after_inlining): Update.
15500 (inline_read_section): Update.
15501 (inline_write_summary): Update.
15502 * ipa-inline.c (compute_uninlined_call_time): Update.
15503 (compute_inlined_call_time): Update.
15504 (recursive_inlining): Update.
15505 (inline_small_functions): Update.
15506 (dump_overall_stats): Update.
15507 * ipa-inline.h: Include sreal.h.
15508 (size_time_entry): Turn time to sreal.
15509 (inline_summary): Turn self_time nad time to sreal.
15511 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
15513 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
15515 (sreal::stream_out, sreal::stream_in): New.
15516 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
15518 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15520 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
15523 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
15526 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
15528 (dimode_scalar_chain::compute_convert_gain): Ditto.
15529 (dimode_scalar_chain::make_vector_copies): Ditto.
15530 (dimode_scalar_chain::convert_reg): Ditto.
15531 (dimode_scalar_chain::convert_insn): Ditto.
15532 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
15533 (VI248_AVX512BW_1): New mode iterator.
15534 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
15535 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
15538 2017-04-25 Martin Sebor <msebor@redhat.com>
15540 PR tree-optimization/80497
15541 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
15542 constants are representable in HOST_WIDE_INT.
15543 (parse_directive): Ditto.
15545 2017-04-25 Martin Sebor <msebor@redhat.com>
15548 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
15549 (new_zero_array): Adjust signature.
15550 (dom_info::dom_init): Used unsigned rather that size_t.
15551 (dom_info::dom_info): Same.
15553 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15554 Jakub Jelinek <jakub@redhat.com>
15557 * config/arm/arm.c: Include gimple.h.
15558 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
15559 returns negative, increment ncrn only if it returned positive.
15560 (arm_needs_doubleword_align): Return int instead of bool,
15561 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
15562 members, but if there is any such non-FIELD_DECL
15563 > PARM_BOUNDARY aligned decl, return -1 instead of false.
15564 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
15565 returns negative, increment nregs only if it returned positive.
15566 (arm_setup_incoming_varargs): Likewise.
15567 (arm_function_arg_boundary): Emit -Wpsabi note if
15568 arm_needs_doubleword_align returns negative, return
15569 DOUBLEWORD_ALIGNMENT only if it returned positive.
15571 2017-04-25 Marek Polacek <polacek@redhat.com>
15574 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
15575 first argument to type.
15577 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
15580 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
15581 type checks to test for compatibility instead of equality.
15583 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
15584 Jakub Jelinek <jakub@redhat.com>
15587 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
15589 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
15590 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
15591 the alignment computation, but return their maximum in warn_alignment.
15592 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
15593 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
15595 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
15596 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
15599 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15601 * config/arc/simdext.md (dmpyh): Fix typo.
15603 2017-04-25 Richard Biener <rguenther@suse.de>
15605 PR tree-optimization/80492
15606 * alias.c (compare_base_decls): Handle registers with asm
15607 specification conservatively.
15608 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
15609 compare_base_decls returning dont-know properly.
15611 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15613 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
15614 (legitimate_offset_address_p): New function.
15615 (arc_legitimate_address_p): Use above function.
15617 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15619 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
15621 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15623 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
15624 ACCH registers whenever they are available.
15626 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15628 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
15629 double regs fix when not used.
15631 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15633 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
15635 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
15636 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
15638 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15640 * config/arc/arc.c (arc_output_addsi): Check for h-register class
15641 when emitting short ADD instructions.
15643 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
15645 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
15647 (cmpsi_cc_c_insn): Likewise.
15648 (cbranchsi4_scratch): Compute proper instruction length using
15649 compact_hreg_operand.
15650 * config/arc/predicates.md (compact_hreg_operand): New predicate.
15652 2017-04-25 Richard Biener <rguenther@suse.de>
15654 PR middle-end/80509
15655 * passes.c (pass_manager::pass_manager): Initialize
15656 m_name_to_pass_map.
15658 2017-04-25 Richard Biener <rguenther@suse.de>
15660 PR tree-optimization/79201
15661 * tree-ssa-sink.c (statement_sink_location): Handle calls.
15663 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15666 * config/s390/vector.md: Split MEM->GPR vector moves for
15667 non-s_operand addresses.
15669 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15672 * config/s390/predicates.md (reload_const_wide_int_operand): New
15674 * config/s390/s390.md ("movti"): Remove d/P alternative.
15675 ("movti_bigconst"): New pattern definition.
15677 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15680 * s390-protos.h (s390_expand_cs_hqi): Removed.
15681 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
15682 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
15683 modes as well as CCZ1mode and CCZmode.
15684 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
15685 signature of s390_emit_compare_and_swap.
15686 (s390_expand_cs_hqi): Likewise, make static.
15687 (s390_expand_cs_tdsi): Generate an explicit compare before trying
15688 compare-and-swap, in some cases.
15689 (s390_expand_cs): Wrapper function.
15690 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
15692 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
15693 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
15694 patterns for small and large integers. Forbid symref memory operands.
15695 Move expander to s390.c. Require cc register.
15696 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
15697 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
15698 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
15699 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
15700 symref memory operands. Remove CC mode and call s390_match_ccmode
15702 ("atomic_exchange<mode>"): Allow and implement all integer modes.
15704 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15706 * config/s390/s390.md (define_peephole2): New peephole to help
15707 combining the load-and-test pattern with volatile memory.
15709 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
15711 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
15712 with CCZmode for TARGET_Z196.
15714 2017-04-25 Jakub Jelinek <jakub@redhat.com>
15716 PR rtl-optimization/80501
15717 * combine.c (make_compound_operation_int): Set subreg_code to SET
15718 even for AND with mask of the sign bit of mode.
15720 PR rtl-optimization/80500
15721 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
15722 sum's initial value.
15724 2017-04-25 Julian Brown <julian@codesourcery.com>
15725 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15727 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
15729 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
15731 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
15733 2017-04-25 Julian Brown <julian@codesourcery.com>
15734 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15736 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
15737 (thunderx2t99_sha): New Reservation.
15739 2017-04-25 Julian Brown <julian@codesourcery.com>
15740 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15742 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
15743 type for 1-element load.
15745 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
15747 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
15749 2017-04-24 Martin Jambor <mjambor@suse.cz>
15751 PR tree-optimization/80293
15752 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
15753 char arrays not totally scalarizable if it is false.
15754 (analyze_all_variable_accesses): Pass correct value in the new
15755 parameter. Add a statistics counter.
15757 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
15759 PR middle-end/79931
15760 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
15762 2017-04-24 Richard Biener <rguenther@suse.de>
15764 PR tree-optimization/80494
15765 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
15766 out for complex types.
15768 2017-04-24 Richard Biener <rguenther@suse.de>
15770 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
15771 * tree-ssa-sccvn.c (print_scc): Print SCC size.
15772 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
15773 (DFS): Adjust and never fail.
15774 (sccvn_dom_walker::fail): Remove.
15775 (sccvn_dom_walker::before_dom_children): Adjust.
15776 (run_scc_vn): Likewise and never fail.
15777 * tree-ssa-pre.c (pass_pre::execute): Adjust.
15778 (pass_fre::execute): Likewise.
15780 2017-04-24 Richard Biener <rguenther@suse.de>
15782 PR tree-optimization/79725
15783 * tree-ssa-sink.c (statement_sink_location): Return whether
15784 failure reason was zero uses. Move that check later.
15785 (sink_code_in_bb): Deal with zero uses by removing the stmt
15788 2017-04-24 Richard Biener <rguenther@suse.de>
15791 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
15792 pointer-based references.
15794 2017-04-24 Richard Biener <rguenther@suse.de>
15797 * pass_manager.h (pass_manager::operator new): Remove.
15798 (pass_manager::operator delete): Likewise.
15799 * passes.c (pass_manager::operator new): Remove.
15800 (pass_manager::operator delete): Likewise.
15801 (pass_manager::pass_manager): Zero individual pass members.
15803 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
15806 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
15807 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
15808 Check "XEXP (src, 1)" operand here.
15809 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
15810 Check "XEXP (src, 1)" operand here.
15811 (dimode_scalar_chain::make_vector_copies): Detect count register
15812 of a shift instruction. Zero extend count register from QImode
15813 to DImode to satisfy vector shift pattern count operand predicate.
15814 Substitute vector shift count operand with a DImode copy.
15815 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
15818 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
15820 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
15821 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15822 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
15823 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
15824 (UNSPEC_NOREX_MEM): Remove definition.
15826 2017-04-21 Richard Biener <rguenther@suse.de>
15828 PR tree-optimization/79547
15829 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
15830 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
15831 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
15832 without any constraints.
15834 2017-04-21 Richard Biener <rguenther@suse.de>
15836 PR tree-optimization/78847
15837 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
15839 2017-04-21 Richard Biener <rguenther@suse.de>
15841 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
15842 (build_distinct_type_copy): Likewise.
15843 (build_variant_type_copy): Likewise.
15844 * tree.c (build_qualified_type): Pass down mem-stat info.
15845 (build_distinct_type_copy): Likewise.
15846 (build_variant_type_copy): Likewise.
15848 2017-04-21 Richard Biener <rguenther@suse.de>
15850 PR tree-optimization/80237
15851 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
15853 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
15854 for a simplified result.
15856 2016-04-21 Richard Biener <rguenther@suse.de>
15858 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
15859 sth as strict as a simple_iv but a chrec without symbols and an
15860 operand defined in the loop we are peeling (and not some subloop).
15861 (propagate_constants_for_unrolling): Propagate all constants.
15863 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15866 * config/i386/i386.c (print_reg): Remove assert for disalowed
15867 regno values, call output_operand_lossage instead.
15869 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
15872 * config/i386/constraints.md (Yc): New register constraint.
15873 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
15874 Use Yc constraint for alternative 2 of operand 0. Remove
15875 preferred_for_speed attribute.
15877 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15879 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
15880 lastprivate clauses in SIMT case.
15882 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
15884 * doc/invoke.texi (-Wextra-semi): Document new warning option.
15886 2017-04-20 Richard Biener <rguenther@suse.de>
15888 PR tree-optimization/57796
15889 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
15890 as N scalar stores.
15891 (vect_model_load_cost): Cost gathers as N scalar loads.
15893 2017-04-20 Richard Biener <rguenther@suse.de>
15895 * ggc-page.c (ggc_allocated_p): Rename to ...
15896 (safe_lookup_page_table_entry): ... this and return the lookup
15898 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
15900 2017-04-20 Richard Biener <rguenther@suse.de>
15902 PR tree-optimization/80453
15903 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
15904 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
15905 from the conditions.
15906 (vn_phi_eq): Pass them down.
15907 (vn_phi_lookup): Record them.
15908 (vn_phi_insert): Likewise.
15910 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
15912 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
15913 uninitialized variable warning to avoid buffer overrun.
15915 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
15918 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
15919 is suppressed for '{ 0 }' in C.
15921 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15923 * BASE-VER: Set to 8.0.0.
15925 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
15927 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
15928 priority .init_array and .fini_array section with SECTION_NOTYPE
15931 2017-04-20 Jakub Jelinek <jakub@redhat.com>
15933 PR middle-end/80423
15934 * tree.h (build_array_type): Add typeless_storage default argument.
15935 * tree.c (type_cache_hasher::equal): Also compare
15936 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
15937 (build_array_type): Add typeless_storage argument, set
15938 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
15940 (build_nonshared_array_type): Adjust build_array_type_1 caller.
15941 (build_array_type): Likewise. Add typeless_storage argument.
15943 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
15944 Jakub Jelinek <jakub@redhat.com>
15946 PR tree-optimization/80426
15947 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
15948 operation on symbolic operands, also compute the overflow for the
15949 invariant part when the operation degenerates into a negation.
15951 2017-04-19 Jakub Jelinek <jakub@redhat.com>
15954 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
15955 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
15958 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
15960 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
15963 * config/avr/avr.c (tree.h): Include it.
15964 (cgraph.h): Include it.
15965 (avr_encode_section_info): Don't warn for uninitialized progmem
15966 variable if it's just an alias.
15968 2017-04-19 Richard Biener <rguenther@suse.de>
15971 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
15972 when needed by AutoPGO.
15974 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
15977 * doc/lto.texi: Remove an extra 'that'.
15979 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
15981 PR rtl-optimization/80429
15982 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
15983 are only used in debug insns.
15985 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
15986 Vladimir Makarov <vmakarov@redhat.com>
15988 * config/sparc/predicates.md (input_operand): Add comment. Return
15989 true for any memory operand when LRA is in progress.
15990 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
15992 2017-04-18 Jeff Law <law@redhat.com>
15995 * mips.md ({return,simple_return}_internal): Do not overwrite
15998 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16000 PR tree-optimization/80443
16001 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
16002 instead of adding 1, subtract -1 and similarly instead of subtracting
16005 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
16007 PR rtl-optimization/80357
16008 * haifa-sched.c (tmp_bitmap): New variable.
16009 (model_recompute): Handle duplicate use records.
16010 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
16011 (free_global_sched_pressure_data): Free it.
16013 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
16016 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
16017 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
16018 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
16019 instead of SYSTEM_HEADER_DIR.
16021 2017-04-18 Jeff Law <law@redhat.com>
16023 PR middle-end/80422
16024 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
16025 predecessors after walking up the insn chain.
16027 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16030 * dwarf2out.c (modified_type_die): Try harder not to emit internal
16031 sizetype type into debug info.
16033 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
16036 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
16037 unneeded test for TARGET_UPPER_REGS_SF.
16038 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
16040 2017-04-18 Jakub Jelinek <jakub@redhat.com>
16043 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
16044 instead of gsi_after_labels.
16046 2017-04-18 Jeff Law <law@redhat.com>
16048 * regcprop.c (maybe_mode_change): Avoid creating copies of the
16052 2017-04-13 Jeff Law <law@redhat.com>
16053 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
16054 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
16056 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
16059 * config/avr/avr.c (intl.h): Include it.
16060 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
16062 2017-04-18 Martin Liska <mliska@suse.cz>
16064 PR gcov-profile/78783
16065 * gcov-tool.c (gcov_output_files): Validate that destination
16066 file is either removed by the tool or by a user.
16068 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
16069 Guy Benyei <guybe@mellanox.com>
16071 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
16072 block, and do not negate it, the stored id is already negative.
16074 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
16076 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
16078 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
16081 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
16082 masks of options that should be turned off if the VSX vector
16083 options are turned off.
16084 (OTHER_P8_VECTOR_MASKS): Likewise.
16085 (OTHER_VSX_VECTOR_MASKS): Likewise.
16086 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
16087 rs6000_disable_incompatible_switches to validate no type switches
16089 (rs6000_incompatible_switch): New function to disallow turning on
16090 other vector options if -mno-vsx, -mno-power8-vector, or
16091 -mno-power9-vector are specified.
16093 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16095 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
16097 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16099 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
16100 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
16101 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
16102 (ARG_POINTER_CFA_OFFSET): Likewise.
16104 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
16106 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
16107 conditions to take advantage of various optimizations.
16109 2017-04-13 Jeff Law <law@redhat.com>
16111 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
16112 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
16113 (zero_extendsidi2_dext): Likewise.
16115 2017-04-13 Jakub Jelinek <jakub@redhat.com>
16118 * fold-const.c (fold_ternary_loc): Revert
16119 use op0 instead of fold_convert_loc (loc, type, arg0) part of
16122 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
16124 PR rtl-optimization/80343
16125 * lra-remat.c (update_scratch_ops): Assign original hard reg to
16126 new scratch pseudo.
16128 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
16131 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
16132 to ubsan_encode_value.
16134 2017-04-13 Jeff Law <law@redhat.com>
16136 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
16137 appearing in DEBUG_INSNs.
16139 2017-04-13 Martin Liska <mliska@suse.cz>
16141 PR gcov-profile/80413
16142 * gcov-io.c (gcov_write_string): Copy to buffer just when
16143 allocated size is greater than zero.
16145 2017-04-13 Jakub Jelinek <jakub@redhat.com>
16148 * dwarf2out.c (decls_for_scope): Ignore declarations of
16149 current_function_decl in BLOCK_NONLOCALIZED_VARS.
16151 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
16154 * ipa-visibility.c (non_local_p): Fix typos.
16155 (localize_node): When localizing symbol in same comdat group,
16156 dissolve the group only when we know external symbols are going
16158 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
16160 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16162 PR tree-optimization/79390
16163 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
16164 order does not result in usable sequence, retry with reversed operand
16170 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
16171 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
16172 op0 instead of fold_convert_loc (loc, type, arg0).
16174 2017-04-12 Jeff Law <law@redhat.com>
16176 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
16177 has a delay slot in the generated code.
16179 * config/cris/cris.md (cris_preferred_reload_class): Return
16180 GENNONACR_REGS rather than GENERAL_REGS.
16182 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16185 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
16186 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
16187 signedness of the result type.
16189 2017-04-12 Richard Biener <rguenther@suse.de>
16190 Jeff Law <law@redhat.com>
16192 PR tree-optimization/80359
16193 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
16194 trim stores to TARGET_MEM_REFs.
16196 2017-04-12 Richard Biener <rguenther@suse.de>
16198 PR tree-optimization/79390
16199 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
16200 threading case even more.
16202 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
16205 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
16206 for quad_address_p for TImode, instead of just not indexed_address.
16208 2017-04-12 Richard Biener <rguenther@suse.de>
16209 Bernd Edlinger <bernd.edlinger@hotmail.de>
16211 PR middle-end/79671
16212 * alias.c (component_uses_parent_alias_set_from): Handle
16213 TYPE_TYPELESS_STORAGE.
16214 (get_alias_set): Likewise.
16215 * tree-core.h (tree_type_common): Add typeless_storage flag.
16216 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
16217 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
16218 for types containing members with TYPE_TYPELESS_STORAGE.
16219 (place_field): Likewise.
16220 (layout_type): Likewise for ARRAY_TYPE.
16221 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
16222 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
16223 TYPE_TYPELESS_STORAGE.
16224 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
16226 2017-04-12 Jakub Jelinek <jakub@redhat.com>
16229 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
16230 first argument to type.
16232 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16236 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
16237 CONST0_RTX (mode) rather than const0_rtx where appropriate.
16238 (rs6000_expand_binop_builtin): Likewise.
16239 (rs6000_expand_ternop_builtin): Likewise; also add missing
16240 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
16241 vshasigma built-ins.
16242 * doc/extend.texi: Document that vec_xxpermdi's third argument
16243 must be a constant.
16245 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
16247 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
16248 Use shift_const cost parameter when calculating gain of STV shifts.
16250 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
16252 PR rtl-optimization/70478
16253 * lra-constraints.c (process_alt_operands): Check memory for
16254 disfavoring memory insn operand.
16256 2017-04-11 Jakub Jelinek <jakub@redhat.com>
16258 PR middle-end/80100
16259 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
16260 left shift in unsigned HOST_WIDE_INT type.
16262 PR rtl-optimization/80385
16263 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
16264 (not (neg X)) into (plus X -1) for complex or non-integral modes.
16267 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
16268 if they have any depend clauses.
16270 2017-04-11 Martin Liska <mliska@suse.cz>
16273 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
16274 * ipa-split.c (split_function): Create a local comdat symbol
16275 if caller is in a comdat group.
16277 2017-04-11 Martin Liska <mliska@suse.cz>
16280 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
16283 2017-04-11 Martin Sebor <msebor@redhat.com>
16285 PR middle-end/80364
16286 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
16287 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
16289 (directive::set_width, directive::set_precision, format_character):
16291 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
16294 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
16297 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
16298 conflict, set target->arch_name instead of target->cpu_name.
16300 2017-04-11 Richard Biener <rguenther@suse.de>
16302 PR tree-optimization/80374
16303 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
16304 build_zero_cst, remove fold_convertible_p check again.
16306 2017-04-11 Martin Liska <mliska@suse.cz>
16309 * ubsan.c (instrument_object_size): Do not instrument register
16312 2017-04-11 Jakub Jelinek <jakub@redhat.com>
16315 * config/i386/i386-builtin-types.def
16316 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
16317 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
16318 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
16319 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
16320 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
16321 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
16322 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
16323 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
16324 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
16325 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
16326 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
16327 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
16328 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
16329 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
16330 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
16331 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
16332 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
16333 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
16334 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
16335 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
16336 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
16337 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
16338 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
16339 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
16340 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
16341 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
16342 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
16343 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
16344 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
16345 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
16346 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
16347 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
16348 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
16349 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
16350 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
16351 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
16352 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
16353 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
16354 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
16355 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
16356 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
16357 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
16358 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
16359 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
16360 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
16361 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
16363 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
16364 flag to second_arg_count, handle 4 argument function type _COUNT
16365 aliases, handle second_arg_count on second argument rather than last.
16367 2017-04-10 Jeff Law <law@redhat.com>
16369 PR tree-optimization/80374
16370 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
16371 record anything if we can not convert integer_zero_node to the
16374 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
16377 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16378 Enhance special handling given to the TARGET_P9_MINMAX option in
16379 relation to certain other options.
16381 2017-04-10 Bin Cheng <bin.cheng@arm.com>
16383 PR tree-optimization/80153
16384 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
16385 remove POINTER_PLUS_EXPR's base part directly, rather than through
16388 2017-04-10 Richard Biener <rguenther@suse.de>
16389 Bin Cheng <bin.cheng@arm.com>
16391 PR tree-optimization/80153
16392 * tree-affine.c (aff_combination_to_tree): Get base pointer from
16393 the first element of pointer type aff_tree. Build result expr in
16395 (add_elt_to_tree): Convert to type unconditionally. Remove other
16396 fold_convert calls.
16397 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
16398 (rewrite_use_nonlinear_expr): Check invariant using iv information.
16400 2017-04-10 Richard Biener <rguenther@suse.de>
16402 * tree-ssa-structalias.c (find_func_aliases): Properly handle
16405 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
16407 PR rtl-optimization/70478
16408 * lra-constraints.c (curr_small_class_check): New.
16409 (update_and_check_small_class_inputs): New.
16410 (process_alt_operands): Update curr_small_class_check. Disfavor
16411 alternative insn memory operands. Check available regs for small
16414 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
16417 * config/mips/mips.opt (-mvirt): Update description.
16418 * doc/invoke.texi (-mvirt): Likewise.
16420 2017-04-10 Richard Biener <rguenther@suse.de>
16422 PR middle-end/80362
16423 * fold-const.c (fold_binary_loc): Look at unstripped ops when
16424 looking for NEGATE_EXPR in -A / -B to A / B folding.
16426 2017-04-10 Martin Liska <mliska@suse.cz>
16428 PR gcov-profile/80224
16429 * gcov.c (print_usage): Fix usage string.
16430 (get_gcov_intermediate_filename): Remove.
16431 (output_gcov_file): Use both for normal and intermediate format.
16432 (generate_results): Do not initialize special file for
16433 intermediate format.
16435 2017-04-10 Richard Biener <rguenther@suse.de>
16437 PR tree-optimization/80304
16438 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
16441 2017-04-10 Nathan Sidwell <nathan@acm.org>
16444 * config/rs6000/rs6000.c (rs6000_vector_type): New.
16445 (rs6000_init_builtins): Use it.
16447 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16449 * config/arm/arm.md (<mrc>): Add mode to SET source.
16450 (<mrrc>): Likewise.
16452 2017-04-10 Richard Biener <rguenther@suse.de>
16454 PR middle-end/80344
16455 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
16457 2017-04-10 Jakub Jelinek <jakub@redhat.com>
16460 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
16461 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
16462 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
16463 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
16464 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
16465 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
16466 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
16467 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
16468 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
16469 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
16470 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
16471 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
16472 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
16473 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
16474 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
16475 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
16476 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
16477 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
16478 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
16479 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
16480 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
16481 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
16482 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
16484 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
16486 PR rtl-optimization/70478
16487 * lra-constraints.c: Reverse the last patch.
16489 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
16491 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
16492 Add comment for WCHAR_T.
16494 2017-04-08 Martin Liska <mliska@suse.cz>
16497 2017-04-07 Martin Liska <mliska@suse.cz>
16500 * ipa-split.c (split_function): Add function part to a same comdat
16503 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16506 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
16508 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
16510 * rs6000/rs6000.c (vec_load_pendulum): Rename...
16511 (vec_pairing): ...to this.
16512 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
16513 (rs6000_sched_init): Adjust for name change.
16514 (struct rs6000_sched_context): Likewise.
16515 (rs6000_init_sched_context): Likewise.
16516 (rs6000_set_sched_context): Likewise.
16518 2017-04-07 Jakub Jelinek <jakub@redhat.com>
16524 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
16526 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
16527 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
16528 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
16530 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
16532 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
16534 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
16536 PR rtl-optimization/70703
16537 * ira-color.c (update_conflict_hard_regno_costs): Use
16538 int64_t instead of HOST_WIDE_INT.
16540 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
16542 PR rtl-optimization/70478
16543 * lra-constraints.c (process_alt_operands): Disfavor alternative
16544 insn memory operands.
16546 2017-04-07 Jeff Law <law@redhat.com>
16548 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
16549 CALL and NOTE_INSN_CALL_ARG_LOCATION.
16551 2017-04-07 Martin Liska <mliska@suse.cz>
16554 * config/aarch64/aarch64.c (aarch64_process_target_attr):
16555 Show error message instead of an ICE.
16557 2017-04-07 Martin Liska <mliska@suse.cz>
16560 * ipa-split.c (split_function): Add function part to a same comdat
16563 2017-04-07 Richard Biener <rguenther@suse.de>
16565 PR middle-end/80341
16566 * tree.c (get_unwidened): Also handle ! for_type case for
16568 * convert.c (do_narrow): Split out from ...
16569 (convert_to_integer_1): ... here. Do not pass final truncation
16570 type to get_unwidened for TRUNC_DIV_EXPR.
16572 2017-04-07 Richard Biener <rguenther@suse.de>
16574 * tree-affine.c (wide_int_ext_for_comb): Take type rather
16576 (aff_combination_const): Adjust.
16577 (aff_combination_scale): Likewise.
16578 (aff_combination_add_elt): Likewise.
16579 (aff_combination_add_cst): Likewise.
16580 (aff_combination_convert): Likewise.
16581 (add_elt_to_tree): Likewise. Remove unused argument.
16582 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
16584 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
16586 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
16588 * config/arm/arm.c (arm_default_short_enums): Use
16589 ARM_DEFAULT_SHORT_ENUMS.
16590 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
16592 2017-04-06 Jakub Jelinek <jakub@redhat.com>
16595 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
16596 members with redundant out-of-class redeclaration.
16598 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
16601 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
16602 * config/i386/i386.md (*zero_extendsidi2):
16603 Add (?*x,*x) and (?*v,*v) alternatives.
16605 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
16608 * config/i386/i386.c (ix86_expand_builtin)
16609 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
16610 mode from insn data. Convert operands to insn operand mode.
16611 Copy operands that don't satisfy insn predicate to a register.
16613 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
16615 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
16618 2017-04-06 Richard Biener <rguenther@suse.de>
16620 PR tree-optimization/80334
16621 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
16622 preserve alignment of accesses.
16624 2017-04-06 Richard Biener <rguenther@suse.de>
16626 PR tree-optimization/80262
16627 * tree-sra.c (build_ref_for_offset): Preserve address-space
16629 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
16630 Drop useless address-space information on MEM_REF offsets.
16632 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
16634 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
16636 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
16638 PR rtl-optimization/70703
16639 * ira-color.c (update_conflict_hard_regno_costs): Use
16640 HOST_WIDE_INT instead of long.
16642 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
16645 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
16646 not defined for x86_64 target. Add -mmmx target option when __SSE2__
16648 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
16649 for x86_64 target. Handle -m3dnowa option.
16651 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
16653 PR rtl-optimization/70703
16654 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
16655 (update_conflict_hard_regno_costs): Use long instead of unsigned
16656 arithmetic for cost calculation.
16658 2017-04-05 Jakub Jelinek <jakub@redhat.com>
16659 Bernd Edlinger <bernd.edlinger@hotmail.de>
16662 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
16665 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
16668 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
16669 ptr_mode with Pmode throughout.
16670 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
16671 into probe_stack_range and use DImode.
16673 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
16676 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
16677 call_eh_return is true.
16679 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16681 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
16682 Initialize last_match_fntype_index.
16684 2017-04-05 Jakub Jelinek <jakub@redhat.com>
16687 * tree-nvr.c: Include internal-fn.h.
16688 (pass_return_slot::execute): Ignore internal calls without
16691 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16692 Richard Biener <rguenther@suse.de>
16695 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
16696 captures used multiple times, except for the last use.
16697 * generic-match-head.c: Include gimplify.h.
16699 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16701 PR tree-optimization/79390
16702 * target.h (struct noce_if_info): Declare.
16703 * targhooks.h (default_noce_conversion_profitable_p): Declare.
16704 * target.def (noce_conversion_profitable_p): New target hook.
16705 * ifcvt.h (struct noce_if_info): New type, moved from ...
16706 * ifcvt.c (struct noce_if_info): ... here.
16707 (noce_conversion_profitable_p): Renamed to ...
16708 (default_noce_conversion_profitable_p): ... this. No longer
16710 (noce_try_store_flag_constants, noce_try_addcc,
16711 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
16712 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
16713 instead of noce_conversion_profitable_p.
16714 * config/i386/i386.c: Include ifcvt.h.
16715 (ix86_option_override_internal): Don't override
16716 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
16717 (ix86_noce_conversion_profitable_p): New function.
16718 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
16719 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
16720 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
16721 * doc/tm.texi: Regenerated.
16723 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16725 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
16728 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16731 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
16732 instructions for small multiply cores.
16734 2017-04-04 Jeff Law <law@redhat.com>
16736 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
16738 (mips_expand_vec_perm_const): Initialize elements in orig_perm
16739 that are not set by the loop over the elements.
16741 2017-04-04 Jakub Jelinek <jakub@redhat.com>
16744 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
16745 int mode, convert_modes it to mode as unsigned, otherwise use
16746 lowpart_subreg to mode rather than SImode.
16747 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
16748 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
16749 Use DImode instead of SImode for the shift count operand.
16750 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
16753 2017-04-04 Richard Biener <rguenther@suse.de>
16755 PR middle-end/80281
16756 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
16757 arithmetic done for the negate or the plus. Simplify.
16758 (A - (-B) -> A + B): Likewise.
16759 * fold-const.c (split_tree): Make sure to not negate pointers.
16761 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
16763 PR rtl-optimization/60818
16764 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
16765 a compare of comparisons with the thing compared if this results
16766 in a different machine mode.
16768 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
16770 * alias.c (base_alias_check): Fix typo in comment.
16771 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
16772 * cgraphunit.c (symbol_table::compile): Likewise.
16773 * collect2.c (maybe_run_lto_and_relink): Likewise.
16774 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
16775 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
16776 * config/avr/avr.c (avr_map_op_t): Likewise.
16777 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
16778 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
16779 * config/epiphany/epiphany.md (movcc): Likewise.
16780 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
16781 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
16783 * config/mips/mips.c (mips_save_restore_reg): Likewise.
16784 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
16785 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
16786 * config/sh/sh.c (sh_rtx_costs): Likewise.
16787 * fold-const.c (fold_truth_andor): Likewise.
16788 * genautomata.c (collapse_flag): Likewise.
16789 * gengtype.h (struct type::u::s): Likewise.
16790 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
16791 * input.c (FORMAT_AMOUNT): Likewise.
16792 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
16793 (known_aggs_to_agg_replacement_list): Likewise.
16794 * ipa-inline-analysis.c: Likewise.
16795 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
16796 * ipa-polymorphic-call.c
16797 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
16798 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
16799 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
16801 * modulo-sched.c (apply_reg_moves): Likewise.
16802 * omp-expand.c (build_omp_regions_1): Likewise.
16803 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
16804 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
16805 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
16806 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
16807 * value-prof.c: Likewise.
16808 * var-tracking.c (val_reset): Likewise.
16810 2017-04-03 Richard Biener <rguenther@suse.de>
16812 PR tree-optimization/80275
16813 * fold-const.c (split_address_to_core_and_offset): Handle
16816 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
16818 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
16819 descriptors is at least equal to that of functions.
16821 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16823 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
16825 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
16828 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
16829 (mov<IMOD4:mode>): New expander.
16830 (*mov<IMOD4:mode>_internal): New insn and split pattern.
16832 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
16834 PR rtl-optimization/79405
16835 * fwprop.c (propagations_left): New variable.
16836 (forward_propagate_into): Decrement it.
16837 (fwprop_init): Initialize it.
16838 (fw_prop): If the variable has reached zero, stop propagating.
16839 (fwprop_addr): Ditto.
16841 2017-03-31 Jakub Jelinek <jakub@redhat.com>
16844 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
16845 a FUNCTION_DECL, pass it as decl instead of origin to
16848 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
16850 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
16853 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
16856 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
16857 TARGET_VSX_SMALL_INTEGER.
16859 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16861 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16862 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
16864 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
16866 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
16867 extraction from odd-numbered MSA register.
16869 2017-03-31 Jakub Jelinek <jakub@redhat.com>
16871 PR middle-end/80173
16872 * expmed.c (store_bit_field_1): Don't attempt to create
16873 a word subreg out of hard registers wider than word if they
16874 have HARD_REGNO_NREGS of 1 for their mode.
16876 PR middle-end/80163
16877 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
16878 conversions to integer types wider than word and pointer.
16881 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
16882 (rtx_equal_for_cselib_p): Pass 0 to it.
16883 * cselib.c (cselib_hasher::equal): Likewise.
16884 (rtx_equal_for_cselib_1): Add depth argument. If depth
16885 is 128, don't look up VALUE locs and punt. Increment
16886 depth in recursive calls when walking VALUE locs.
16888 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
16890 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
16891 (make_gcov_file_name): Use the canonical path name for generating
16893 (read_line): Fix handling of files with ascii null bytes.
16895 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
16897 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
16898 to initialise a vector register instead
16899 of using a const_int.
16901 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16903 PR translation/80189
16904 * gimplify.c (omp_default_clause): Use %qs instead of %s in
16905 diagnostic messages.
16907 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
16910 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
16911 (dfp_diex_<mode>): Update mode of operand 1.
16912 * doc/extend.texi (dxex, dxexq): Document change to return type.
16913 (diex, diexq): Document change to argument type.
16915 2017-03-30 Martin Jambor <mjambor@suse.cz>
16918 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
16919 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
16920 it reflects the signature changes performed at the callee side.
16921 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
16922 to cgraph_build_function_type_skip_args.
16923 (build_function_decl_skip_args): Adjust call to the above function.
16925 2017-03-30 Jakub Jelinek <jakub@redhat.com>
16928 * config/i386/sse.md
16929 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
16930 register as dest whenever it is a MEM not rtx_equal_p to the
16931 corresponding dup operand, and when forcing into reg move the
16932 reg into the memory afterwards.
16933 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
16934 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
16935 for the force_reg mode.
16936 (avx512vl_vextractf128<mode>): Use register as dest either
16937 always when a MEM, or when it is a MEM not rtx_equal_p to the
16938 corresponding dup operand, or even not when it is a CONST_VECTOR
16939 depending on the mode and lo vs. hi.
16940 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
16942 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
16943 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
16944 Likewise. Require that operands[2] is even.
16945 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
16946 Remove extraneous parens. Require that operands[2] is a multiple
16948 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
16949 operands[0] is a MEM if <mask_applied>, the predicates/constraints
16950 disallow memory then.
16952 2017-03-30 Richard Biener <rguenther@suse.de>
16954 PR tree-optimization/77498
16955 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
16956 to non-constants over backedges.
16958 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
16960 PR rtl-optimization/80233
16961 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
16962 as last_combined_insn. Do not test for BARRIER_P separately.
16964 2017-03-29 Andreas Schwab <schwab@suse.de>
16967 * calls.c (prepare_call_address): Convert funexp to Pmode before
16968 copying to temp reg.
16970 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16972 PR tree-optimization/80158
16973 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
16974 Handle possible future case of more than one alternate
16976 (replace_rhs_if_not_dup): Likewise.
16977 (replace_one_candidate): Likewise.
16979 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
16981 PR rtl-optimization/80193
16982 * ira.c (ira): Do not check allocation for LRA.
16984 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
16986 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
16987 (nvptx_output_simt_exit): Declare.
16988 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
16989 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
16990 (init_softstack_frame): Move initialization of crtl->is_leaf to...
16991 (nvptx_declare_function_name): ...here. Emit declaration of local
16992 memory space buffer for omp_simt_enter insn.
16993 (nvptx_output_unisimt_switch): New.
16994 (nvptx_output_softstack_switch): New.
16995 (nvptx_output_simt_enter): New.
16996 (nvptx_output_simt_exit): New.
16997 * config/nvptx/nvptx.h (struct machine_function): New fields
16998 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
16999 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
17000 (UNSPECV_SIMT_EXIT): Ditto.
17001 (omp_simt_enter_insn): New insn.
17002 (omp_simt_enter): New expansion.
17003 (omp_simt_exit): New insn.
17004 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
17006 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
17007 (expand_GOMP_SIMT_ENTER_ALLOC): New.
17008 (expand_GOMP_SIMT_EXIT): New.
17009 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
17010 (GOMP_SIMT_ENTER_ALLOC): Ditto.
17011 (GOMP_SIMT_EXIT): Ditto.
17012 * target-insns.def (omp_simt_enter): New insn.
17013 (omp_simt_exit): Ditto.
17014 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
17016 (lower_rec_simd_input_clauses): Implement SIMT privatization.
17017 (lower_rec_input_clauses): Likewise.
17018 (lower_lastprivate_clauses): Handle SIMT privatization.
17020 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
17021 (ompdevlow_adjust_simt_enter): New.
17022 (find_simtpriv_var_op): New.
17023 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
17024 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
17026 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
17027 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
17028 (copy_decl_for_dup_finish): Ditto.
17030 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
17032 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
17035 * config/i386/i386.c (ix86_option_override_internal): Always
17036 allow -mpreferred-stack-boundary=3 for 64-bit targets.
17038 2017-03-28 Bin Cheng <bin.cheng@arm.com>
17040 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
17042 2017-03-28 Bin Cheng <bin.cheng@arm.com>
17044 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
17045 mark new edge's irreducible flag accordign to it.
17046 (vect_do_peeling): Check loop preheader edge's irreducible flag
17047 and pass it to function slpeel_add_loop_guard.
17049 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
17051 PR tree-optimization/80218
17052 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
17053 Update block frequencies and counts.
17055 2017-03-28 Richard Biener <rguenther@suse.de>
17057 PR tree-optimization/78644
17058 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
17059 of a simplification result we may not use it at all.
17061 2017-03-28 Richard Biener <rguenther@suse.de>
17064 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
17065 without arguments, generate default definition of a SSA name.
17067 2017-03-28 Richard Biener <rguenther@suse.de>
17069 PR middle-end/80222
17070 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
17071 TYPE_REF_CAN_ALIAS_ALL references.
17072 * fold-const.c (fold_indirect_ref_1): Likewise.
17074 2017-03-28 Martin Liska <mliska@suse.cz>
17077 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
17078 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
17080 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
17081 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
17083 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
17084 (EXTRA_SPECS): Define.
17085 (SUBTARGET_EXTRA_SPECS): Likewise.
17086 (SUBTARGET_CPP_SPEC): Likewise.
17087 * config/arc/elf.h (EXTRA_SPECS): Renamed to
17088 SUBTARGET_EXTRA_SPECS.
17089 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
17091 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
17093 * config/arc/simdext.md (vst64_insn): Update pattern.
17094 (vld32wh_insn): Likewise.
17095 (vld32wl_insn): Likewise.
17096 (vld64_insn): Likewise.
17097 (vld32_insn): Likewise.
17099 2017-03-28 Marek Polacek <polacek@redhat.com>
17102 * fold-const.c (fold_comparison): Use protected_set_expr_location
17103 instead of SET_EXPR_LOCATION.
17105 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
17107 * tree.c (add_expr): Avoid name lookup warning.
17109 2017-03-27 Jeff Law <law@redhat.com>
17111 PR tree-optimization/80216
17112 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
17113 function name. Limit recursion depth.
17114 (record_temporary_equivalences): Corresponding changes.
17116 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
17118 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
17121 2017-03-27 Jakub Jelinek <jakub@redhat.com>
17124 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
17126 * cfgcleanup.c (reg_note_cfa_p): New array.
17127 (insns_have_identical_cfa_notes): New function.
17128 (old_insns_match_p): Don't cross-jump in between /f
17129 and non-/f instructions. If both i1 and i2 are frame related,
17130 verify all CFA notes, their order and content.
17132 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
17135 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
17136 HImode and SImode with zero extend to DImode to one insn.
17137 (bswap<mode>2_extenddi): Likewise.
17138 (bswapsi2_extenddi): Likewise.
17139 (bswaphi2_extendsi): Likewise.
17140 (bswaphi2): Combine bswap HImode and SImode into one insn.
17141 Separate memory insns from swapping register.
17142 (bswapsi2): Likewise.
17143 (bswap<mode>2): Likewise.
17144 (bswaphi2_internal): Delete, no longer used.
17145 (bswapsi2_internal): Likewise.
17146 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
17147 store, and gpr<-gpr swap insns.
17148 (bswap<mode>2_store): Likewise.
17149 (bswaphi2_reg): Register only splitter, combine with the splitter.
17150 (bswaphi2 splitter): Likewise.
17151 (bswapsi2_reg): Likewise.
17152 (bswapsi2 splitter): Likewise.
17153 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
17154 the insns into load, store, and register/register insns.
17155 (bswapdi2_ldbrx): Likewise.
17156 (bswapdi2_load): Likewise.
17157 (bswapdi2_store): Likewise.
17158 (bswapdi2_reg): Likewise.
17160 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
17162 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
17163 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
17165 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
17168 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
17170 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
17171 special handling for target option conflicts between dform
17172 options (-mpower9-dform, -mpower9-dform-vector,
17173 -mpower9-dform-scalar) and -mno-direct-move.
17175 2017-03-27 Richard Biener <rguenther@suse.de>
17177 PR tree-optimization/80181
17178 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
17180 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17182 * config/arc/predicates.md (move_double_src_operand): Replace the
17183 call to move_double_src_operand with a call to address_operand.
17185 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17187 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
17188 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
17189 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
17191 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17193 * config/arc/predicates.md (long_immediate_loadstore_operand):
17194 Consider scaled addresses cases.
17196 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
17198 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
17199 restored when in interrupt.
17200 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
17201 doesn't have delay slot.
17203 2017-03-27 Richard Biener <rguenther@suse.de>
17206 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
17207 inlined thunk clones.
17209 2017-03-27 Jakub Jelinek <jakub@redhat.com>
17212 * asan.c (instrument_derefs): Copy over last operand from
17213 original COMPONENT_REF to the new COMPONENT_REF with
17214 DECL_BIT_FIELD_REPRESENTATIVE.
17215 * ubsan.c (instrument_object_size): Likewise.
17217 2017-03-27 Richard Biener <rguenther@suse.de>
17219 PR tree-optimization/80170
17220 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
17221 sure DR/SCEV didnt fold in constants we do not see when looking
17222 at the reference base alignment.
17224 2017-03-27 Richard Biener <rguenther@suse.de>
17226 PR middle-end/80171
17227 * gimple-fold.c (fold_ctor_reference): Properly guard against
17228 NULL return value from canonicalize_constructor_val.
17230 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
17233 * config/i386/i386.c (ix86_expand_builtin)
17234 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
17235 flags reg setting and flags reg using instructions.
17236 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
17237 clobbering instructions to zero extend op2.
17239 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
17241 * doc/install.texi (Configuration) <--with-aix-soname>:
17242 Update link to AIX ld.
17244 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
17246 PR rtl-optimization/80160
17247 PR rtl-optimization/80159
17248 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
17249 reg_alternate_class into account.
17251 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
17254 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
17255 to consider in curr_insn_transform.
17257 2017-03-24 Jakub Jelinek <jakub@redhat.com>
17259 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
17260 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
17261 and emit_mode_inner.
17263 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17265 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
17266 argument to the overloaded builtin variants. Use the new flag to
17267 deprecate certain builtin variants.
17268 * config/s390/s390-builtin-types.def: Add new builtin types.
17269 * config/s390/s390-builtins.h: Support new flags field for
17270 overloaded builtins.
17271 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
17272 (s390_macro_to_expand): Enable vector float data type.
17273 (s390_cpu_cpp_builtins_internal): Indicate support of the new
17274 builtins by incrementing the __VEC__ version number.
17275 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
17277 (s390_resolve_overloaded_builtin): Emit error messages depending
17278 on the builtin flags.
17279 * config/s390/s390.c (s390_expand_builtin): Support additional
17280 flags argument. Change error message to match the messages
17281 emitted in s390-c.c.
17282 * config/s390/s390.md: New UNSPEC_* constants.
17283 (op_type): Add new instruction types.
17284 * config/s390/vecintrin.h: Add new builtins and test data class
17286 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
17287 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
17288 (VEC_INEXACT, VEC_NOINEXACT): New constants.
17289 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
17290 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
17291 ("vec_mergel<mode>"): V_HW -> VEC_HW.
17293 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
17294 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
17295 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
17296 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
17298 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
17299 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
17300 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
17301 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
17303 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
17304 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
17305 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
17306 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
17307 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
17308 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
17309 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
17311 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
17312 ("vec_scatter_element<V_HW_4:mode>_DI")
17313 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
17314 ("vec_fpint<mode>", "vflls")
17315 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
17316 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
17317 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
17318 ("*vec_cmphe<mode>_cc"): ... these.
17320 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
17321 mode constant instead of magic value.
17323 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17325 * config/s390/s390.c (s390_expand_vec_compare): Support other
17326 vector floating point modes than just V2DF.
17327 (s390_expand_vcond): Likewise.
17328 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
17329 (s390_cannot_change_mode_class): Prevent mode changes between TF
17330 and V1TF in vector registers.
17331 * config/s390/s390.md (DF, SF): New mode attributes.
17332 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
17333 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
17334 SFmode support for VRs.
17335 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
17337 (VFT, VF_HW): New mode iterators.
17338 (vw, sdx): New mode attributes.
17339 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
17340 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
17341 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
17342 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
17343 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
17344 also the new vector floating point modes. Renaming to ...
17346 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
17347 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
17348 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
17349 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
17350 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
17351 ("vec_unordered<mode>"): ... these.
17353 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
17354 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
17355 ("*vec_extendv2df"): New insn definitions.
17357 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17359 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
17360 ("mulditi3_2", "*muldi3_sign"): New patterns.
17361 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
17362 rename the pattern definition.
17364 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17366 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
17368 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
17370 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17372 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
17373 instruction if possible.
17374 * config/s390/vector.md (vec_halfnumelts): New mode
17376 ("*vec_vllezlf<mode>"): New pattern.
17378 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17380 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
17381 ("popcountv4si2", "popcountv2di2"): Rename to ...
17382 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
17383 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
17385 ("popcount<mode>2_vxe"): New pattern.
17387 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17389 * common/config/s390/s390-common.c (processor_flags_table): Add
17391 * config.gcc: Add arch12.
17392 * config/s390/driver-native.c (s390_host_detect_local_cpu):
17393 Default to arch12 for unknown CPU model numbers.
17394 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
17395 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
17396 PROCESSOR_max sanity check.
17397 * config/s390/s390-opts.h (enum processor_type): Add
17399 * config/s390/s390.c (processor_table): Add arch12.
17400 (s390_expand_builtin): Add check for B_VXE flag.
17401 (s390_issue_rate): Add PROCESSOR_ARCH12.
17402 (s390_get_sched_attrmask): Likewise.
17403 (s390_get_unit_mask): Likewise.
17404 (s390_sched_score): Enable z13 scheduling for arch12.
17405 (s390_sched_reorder): Likewise.
17406 (s390_sched_variable_issue): Likewise.
17407 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
17409 (s390_tune_attr): Use z13 scheduling also for arch12.
17410 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
17411 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
17412 (TARGET_VXE_P): New macros.
17413 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
17414 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
17415 * config/s390/s390.opt: Add arch12 as processor_type.
17417 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17419 * config/s390/s390.md
17420 ("fixuns_truncdddi2", "fixuns_trunctddi2")
17421 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
17422 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
17424 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
17425 Rename expanders to ...
17427 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
17428 ("fixuns_truncdddi2_emu"): ... these.
17430 ("fixuns_trunc<mode>si2_emu"): New expander.
17432 ("*fixuns_truncdfdi2_z13"): Rename to ...
17433 ("*fixuns_truncdfdi2_vx"): ... this.
17435 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17437 * config/s390/2964.md: Remove the single element vector compare
17438 instructions which are no longer used.
17439 * config/s390/s390.c (s390_select_ccmode): Remove handling of
17441 (s390_canonicalize_comparison): Remove handling of DFmode
17443 (s390_expand_vec_compare_scalar): Remove function.
17444 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
17445 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
17447 ("*cmp<mode>_ccs"): Add wfcdb instruction.
17449 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17451 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
17453 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
17454 will anyway by matched by mov<mode>_64dfp.
17456 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17458 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
17459 vlef/vstef. Add missing operand to vleif.
17461 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17463 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
17464 pair for all vector types with 64 bit elements.
17465 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
17466 * config/s390/vector.md (V_HW_64): ... here.
17467 (V_128_NOSINGLE): New mode iterator.
17468 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
17469 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
17470 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
17471 ("*vec_load_pairv2di"): Change to ...
17472 ("*vec_load_pair<mode>"): ... this one.
17474 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17476 * config/s390/constraints.md: Add comments.
17477 (jKK): Reject element sizes > 8 bytes.
17478 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
17480 * config/s390/s390.md: Add the s_operand checks formerly in
17481 s390_split_ok_p to various splitters where they are still
17483 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
17484 for 128 bit vectors. Plus two splitters.
17486 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17488 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
17491 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17494 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
17495 error if the boundary argument is not constant.
17497 2017-03-24 Jakub Jelinek <jakub@redhat.com>
17499 PR rtl-optimization/80112
17500 * loop-doloop.c (doloop_condition_get): Don't check condition
17501 if cmp isn't SET with IF_THEN_ELSE src.
17503 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17505 PR tree-optimization/80158
17506 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
17507 replacing a candidate statement, also replace it for the
17508 candidate's alternate interpretation.
17509 (replace_rhs_if_not_dup): Likewise.
17510 (replace_one_candidate): Likewise.
17512 2017-03-24 Richard Biener <rguenther@suse.de>
17514 PR tree-optimization/80167
17515 * graphite-isl-ast-to-gimple.c
17516 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
17518 (translate_isl_ast_to_gimple::get_rename): Likewise.
17520 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
17522 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
17523 handling of certain combinations of target options, including the
17524 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
17525 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
17527 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17530 * config/arm/arm.md (*load_multiple): Add reload_completed to
17531 matching condition.
17533 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17534 Richard Biener <rguenth@suse.de>
17536 PR tree-optimization/79908
17537 PR tree-optimization/80136
17538 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
17539 been cast away, gimplify_and_add suffices.
17541 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
17543 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
17545 2017-03-23 Richard Biener <rguenther@suse.de>
17547 PR tree-optimization/80032
17548 * gimplify.c (gimple_push_cleanup): Forced unconditional
17549 cleanups still have to go to the conditional_cleanups
17552 2017-03-22 Jakub Jelinek <jakub@redhat.com>
17554 PR tree-optimization/80072
17555 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
17557 (next_operand_entry_id): Change type to unsigned int.
17558 (sort_by_operand_rank): Make sure to return the right return value
17559 even if unsigned fields are bigger than INT_MAX.
17560 (struct oecount): Change cnt and id type to unsigned int.
17561 (oecount_hasher::equal): Formatting fix.
17562 (oecount_cmp): Make sure to return the right return value
17563 even if unsigned fields are bigger than INT_MAX.
17564 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
17567 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
17568 TREE_READONLY on result if writing it more than once.
17571 * doc/invoke.texi (-fsanitize=thread): Document that with
17572 -fnon-call-exceptions atomics are not able to throw
17576 * tsan.c: Include tree-eh.h.
17577 (instrument_builtin_call): Call maybe_clean_eh_stmt or
17578 maybe_clean_or_replace_eh_stmt where needed.
17579 (instrument_memory_accesses): Add cfg_changed argument.
17580 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
17581 if it returned true.
17582 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
17584 PR rtl-optimization/63191
17585 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
17586 wrapper function, moved the whole old content into ...
17587 (ix86_delegitimize_address_1): ... this. New inline function.
17588 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
17589 true as last argument instead of ix86_delegitimize_address.
17591 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
17593 * config/aarch64/aarch64.c (generic_branch_cost): Copy
17594 cortexa57_branch_cost.
17596 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
17598 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
17600 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17603 * doc/md.texi (Constraints): Document wA constraint.
17604 * config/rs6000/constraints.md (wA): New.
17605 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
17606 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
17607 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
17608 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
17610 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
17613 * gimplify.c (is_oacc_declared): New function.
17614 (oacc_default_clause): Use it to set default flags for acc declared
17615 variables inside parallel regions.
17616 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
17617 declared variables.
17618 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
17619 declare attribute to any decl as necessary.
17621 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
17624 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
17625 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
17626 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
17627 (arm_arch_lpae): This.
17628 * config/arm/arm.c (arm_arch7ve): Rename into ...
17629 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
17630 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
17633 2017-03-22 Martin Liska <mliska@suse.cz>
17636 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
17637 error message instead of an ICE.
17639 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17641 * doc/extend.texi (6.11 Additional Floating Types): Revise.
17643 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
17645 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
17647 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
17650 2017-03-21 Martin Sebor <msebor@redhat.com>
17652 * doc/extend.texi: Use "cannot" instead of "can't."
17653 * doc/hostconfig.texi: Same.
17654 * doc/install.texi: Same.
17655 * doc/invoke.texi: Same.
17656 * doc/loop.texi: Same.
17657 * doc/md.texi: Same.
17658 * doc/objc.texi: Same.
17659 * doc/rtl.texi: Same.
17660 * doc/tm.texi: Same.
17661 * doc/tm.texi.in: Same.
17662 * doc/trouble.texi: Same.
17664 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
17667 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
17668 (collect_checksum_attributes): Set it.
17669 (die_checksum_ordered): Use it.
17671 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17673 PR tree-optimization/79908
17674 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
17675 change: For a VA_ARG whose LHS has been cast away, use
17676 force_gimple_operand to construct the side effects.
17678 2017-03-21 David Malcolm <dmalcolm@redhat.com>
17680 PR translation/80001
17681 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
17682 more amenable to translation.
17683 (oacc_loop_auto_partitions): Likewise.
17685 2017-03-21 Marek Polacek <polacek@redhat.com>
17686 Martin Sebor <msebor@redhat.com>
17688 PR tree-optimization/80109
17689 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
17690 on INTEGRAL_TYPE_P.
17692 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17693 Segher Boessenkool <segher@kernel.crashing.org>
17696 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
17697 check reg_used_between_p between insn and one of succ or succ2
17698 depending on if succ is artificial insn not inserted into insn
17701 2017-03-21 Martin Liska <mliska@suse.cz>
17703 PR gcov-profile/80081
17704 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
17705 * doc/gcc.texi: Include gcov-dump stuff.
17706 * doc/gcov-dump.texi: New file.
17708 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
17710 PR rtl-optimization/79150
17711 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
17712 conditional jump, if the jump is the last insn of the loop.
17714 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17715 Richard Biener <rguenth@suse.de>
17717 PR tree-optimization/79908
17718 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
17719 been cast away, use force_gimple_operand to construct the side
17722 2017-03-21 Martin Liska <mliska@suse.cz>
17724 PR libfortran/79956
17725 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
17728 2017-03-21 Brad Spengler <spender@grsecurity.net>
17731 * plugin.c (htab_hash_plugin): New function.
17732 (add_new_plugin): Use it and adjust.
17733 (parse_plugin_arg_opt): Adjust.
17734 (init_one_plugin): Likewise.
17736 2017-03-21 Richard Biener <rguenther@suse.de>
17738 PR tree-optimization/80032
17739 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
17740 if set force the cleanup to happen unconditionally.
17741 (gimplify_target_expr): Push inserted clobbers with force_uncond
17742 to avoid them being removed by control-dependent DCE.
17744 2017-03-21 Richard Biener <rguenther@suse.de>
17746 PR tree-optimization/80122
17747 * tree-inline.c (copy_bb): Do not expans va-arg packs or
17748 va_arg_pack_len when the inlined call stmt requires pack
17750 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
17752 2017-03-21 Jakub Jelinek <jakub@redhat.com>
17755 * tsan.c (instrument_builtin_call): If the memory model argument
17756 is not a constant, assume it is valid.
17759 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
17762 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17764 PR rtl-optimization/79910
17765 * combine.c (can_combine_p): Do not allow combining an I0 or I1
17766 if its dest is used by an insn before I2 (other than the combined
17767 insns themselves, which are properly handled already).
17769 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
17772 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17774 * combine.c (record_used_regs): New static function.
17775 (try_combine): Handle situations where there is an additional
17776 instruction between I2 and I3 which needs to have a LOG_LINK
17780 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17782 * combine.c (try_combine): Delete redundant i1 test. Call
17783 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17785 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17788 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
17789 alternatives 13/14.
17791 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17793 PR tree-optimization/80054
17794 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
17795 the optimization if a PHI or any of its arguments is not dominated
17796 by the candidate's basis. Use gphi* rather than gimple* as
17798 (replace_profitable_candidates): Clean up a gimple* variable that
17799 should be a gphi* variable.
17801 2017-03-20 Martin Sebor <msebor@redhat.com>
17804 * doc/extend.texi (attribute constructor): Document present limitation.
17806 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
17809 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
17810 __POWER9_VECTOR__ #ifdef control, change template definition to
17811 use Power9-specific built-in function.
17812 (vec_any_eq): Likewise.
17813 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
17814 to control outcomes from this test.
17815 (vector_ae_<mode>p): For VEC_F modes, likewise.
17817 2017-03-20 Ian Lance Taylor <iant@google.com>
17819 * config/i386/i386.c (ix86_function_regparm): Save an extra
17820 register for -fsplit-stack with DECL_STATIC_CHAIN.
17822 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17825 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
17826 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
17828 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
17830 * config/riscv/riscv.c (riscv_print_operand): Use "fence
17832 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
17835 2017-03-20 Marek Polacek <polacek@redhat.com>
17838 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
17840 2017-03-20 Richard Biener <rguenther@suse.de>
17842 PR tree-optimization/80113
17843 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
17844 allocate extra SSA name for PHI def.
17845 (add_close_phis_to_outer_loops): Likewise.
17846 (add_close_phis_to_merge_points): Likewise.
17847 (copy_loop_close_phi_args): Likewise.
17848 (copy_cond_phi_nodes): Likewise.
17850 2017-03-20 Martin Liska <mliska@suse.cz>
17852 PR middle-end/79753
17853 * tree-chkp.c (chkp_build_returned_bound): Do not build
17854 returned bounds for a LHS that's not a BOUNDED_P type.
17856 2017-03-20 Martin Liska <mliska@suse.cz>
17860 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
17861 COMPLEX_CST and VECTOR_CST.
17863 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17866 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
17867 target operand. A new splitter adds the clobber statement in case
17868 the target operand is dead anyway.
17870 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
17872 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
17873 to age-old versions of binutils and glibc.
17875 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
17877 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
17879 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17881 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
17883 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
17885 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
17886 requirement for binutils 2.13.
17888 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
17890 * combine.c (try_combine): Delete redundant i1 test. Call
17891 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
17893 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
17895 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
17896 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
17898 <riscv64-*-elf>: Re-arrange section
17899 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
17900 <riscv32-*-linux>: Likewise.
17901 <riscv64-*-elf>: Likewise
17902 <riscv64-*-linux>: Likewise.
17904 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
17907 * aarch64.opt(verbose-cost-dump): Fix typo.
17909 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
17912 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
17913 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
17915 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
17917 * reload.c (find_reloads): When reloading a nonoffsettable address,
17918 use RELOAD_OTHER for it and its address reloads.
17920 PR rtl-optimization/79910
17921 * combine.c (record_used_regs): New static function.
17922 (try_combine): Handle situations where there is an additional
17923 instruction between I2 and I3 which needs to have a LOG_LINK
17926 2017-03-17 Jeff Law <law@redhat.com>
17928 PR tree-optimization/71437
17929 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
17930 conditional in the hash table first.
17931 (vrp_dom_walker::before_dom_children): Extract condition from
17932 ASSERT_EXPR. Record condition, its inverion and any implied
17933 conditions as well.
17935 2017-03-17 Marek Polacek <polacek@redhat.com>
17936 Markus Trippelsdorf <markus@trippelsdorf.de>
17938 PR tree-optimization/80079
17939 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
17942 2017-03-17 Richard Biener <rguenther@suse.de>
17944 PR middle-end/80075
17945 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
17946 Properly verify the LHS before the RHS possibly claims to be
17948 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
17951 2017-03-17 Martin Jambor <mjambor@suse.cz>
17953 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
17954 (List of -O2 options): Likewise.
17955 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
17958 2017-03-17 Tom de Vries <tom@codesourcery.com>
17960 * gcov-dump.c (print_usage): Print bug_report_url.
17962 2017-03-17 Richard Biener <rguenther@suse.de>
17964 PR middle-end/80050
17965 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
17966 (parser::peek): Likewise.
17968 2017-03-17 Richard Biener <rguenther@suse.de>
17970 PR tree-optimization/80048
17971 * sese.c (free_sese_info): Properly release rename_map and
17972 copied_bb_map elements.
17974 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
17976 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
17977 Add linked-list forward and backlinks. Insert on
17978 construction, remove on destruction.
17979 (class pass_store_merging): Add m_stores_head field.
17980 (pass_store_merging::terminate_and_process_all_chains):
17981 Iterate over m_stores_head list.
17982 (pass_store_merging::terminate_all_aliasing_chains):
17984 (pass_store_merging::execute): Check for debug stmts first.
17985 Push new chains onto the m_stores_head stack.
17987 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
17990 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
17991 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
17992 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
17994 2017-03-16 Jeff Law <law@redhat.com>
17996 PR tree-optimization/71437
17997 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
17998 member function. Implementation moved into after_dom_children
17999 member function and into the threader's thread_outgoing_edges
18001 (dom_opt_dom_walker::after_dom_children): Simplify by moving
18002 some code into new thread_outgoing_edges.
18003 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
18004 definition. Simplify marker handling (do it here). Assume we always
18005 have the available expression and the const/copies tables.
18006 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
18008 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
18009 * tree-vrp.c (equiv_stack): No longer file scoped.
18010 (vrp_dom_walker): New class.
18011 (vrp_dom_walker::before_dom_children): New member function.
18012 (vrp_dom_walker::after_dom_children): Likewise.
18013 (identify_jump_threads): Setup domwalker. Use it rather than
18014 walking edges in a random order by hand. Simplify setup/finalization.
18015 (finalize_jump_threads): Remove.
18016 (vrp_finalize): Do not call identify_jump_threads here.
18017 (execute_vrp): Do it here instead and call thread_through_all_blocks
18020 PR tree-optimization/71437
18021 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
18023 (simplify_stmt_for_jump_threading): Add basic_block argument. All
18025 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
18026 (dom_opt_dom_walker::thread_across_edge): Remove
18027 handle_dominating_asserts argument. All callers changed.
18028 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
18029 changes. Remove calls to lhs_of_dominating_assert. Other
18030 uses of handle_dominating_asserts turn into unconditional code
18031 (simplify_control_stmt_condition_1): Likewise.
18032 (simplify_control_stmt_condition): Likewise.
18033 (thread_through_normal_block, thread_across_edge): Likewise.
18034 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
18035 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
18036 object if it is not an SSA_NAME.
18037 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
18038 before calling into the VRP specific simplifiers.
18039 (identify_jump_threads): Remove handle_dominating_asserts
18042 2017-03-16 Jakub Jelinek <jakub@redhat.com>
18045 * tree-diagnostic.c (default_tree_printer): No longer static.
18046 * tree-diagnostic.h (default_tree_printer): New prototype.
18048 2017-03-16 Tamar Christina <tamar.christina@arm.com>
18050 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
18051 Change ins into fmov.
18053 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18055 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
18056 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
18057 Use h_con constraint for operand 1.
18058 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
18059 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
18061 2017-03-15 Jeff Law <law@redhat.com>
18063 PR tree-optimization/71437
18064 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
18065 (record_temporary_equivalences): Use it.
18067 PR tree-optimization/71437
18068 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
18069 tree-ssa-scopedtables.
18070 (lookup_avail_expr, build_and_record_new_cond): Likewise.
18071 (record_conditions, record_cond, vuse_eq): Likewise.
18072 (record_edge_info): Adjust to API tweak of record_conditions.
18073 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
18074 (record_temporary_equivalences, optimize_stmt): Likewise.
18075 (eliminate_redundant_computations): Likewise.
18076 (record_equivalences_from_stmt): Likewise.
18077 * tree-ssa-scopedtables.c: Include options.h and params.h.
18078 (vuse_eq): New function, moved from tree-ssa-dom.c
18079 (build_and_record_new_cond): Likewise.
18080 (record_conditions): Likewise. Accept vector of conditions rather
18081 than edge_equivalence structure for first argument.
18082 for the first argument.
18083 (avail_exprs_stack::lookup_avail_expr): New member function, moved
18084 from tree-ssa-dom.c.
18085 (avail_exprs_stack::record_cond): Likewise.
18086 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
18087 from tree-ssa-dom.c.
18088 (avail_exprs_stack): Add new member functions lookup_avail_expr
18090 (record_conditions): Declare.
18092 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
18095 * lra-constraints.c (process_alt_operands): Increase reject for
18096 reloading an input/output operand.
18098 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
18101 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
18102 insns to convert from signed/unsigned char/short to IEEE 128-bit
18104 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
18106 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
18109 * config/i386/i386.c (ix86_vector_duplicate_value): Create
18110 subreg of inner mode for values already in registers.
18112 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
18114 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
18115 iteration reg is used after the loop.
18117 2017-03-14 Martin Sebor <msebor@redhat.com>
18119 PR tree-optimization/79800
18120 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
18121 precision in negative-positive range.
18122 (format_floating): Call non-const overload with adjusted precision.
18124 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
18127 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
18130 2017-03-14 Martin Sebor <msebor@redhat.com>
18132 PR middle-end/80020
18133 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
18134 * builtins.def (aligned_alloc): Use it.
18137 * Makefile.in (GTFILES): Add calls.c.
18138 * calls.c: Include "gt-calls.h".
18140 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
18142 PR rtl-optimization/79728
18143 * regs.h (struct target_regs): New field
18144 x_contains_allocatable_regs_of_mode.
18145 (contains_allocatable_regs_of_mode): New macro.
18146 * reginfo.c (init_reg_sets_1): Initialize it, and change
18147 contains_reg_of_mode so it includes global regs as well.
18148 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
18149 rather than contains_regs_of_mode.
18151 2017-03-14 Martin Liska <mliska@suse.cz>
18153 * doc/invoke.texi: Document options that can't be combined with
18154 -fcheck-pointer-bounds.
18156 2017-03-14 Martin Liska <mliska@suse.cz>
18158 PR middle-end/79831
18159 * doc/invoke.texi (-Wchkp): Document the option.
18161 2017-03-14 Martin Liska <mliska@suse.cz>
18163 * Makefile.in: Install gcov-dump.
18165 2017-03-14 Martin Liska <mliska@suse.cz>
18167 * multiple_target.c (expand_target_clones): Bail out for
18168 an invalid attribute.
18170 2017-03-14 Richard Biener <rguenther@suse.de>
18172 * alias.c (struct alias_set_entry): Pack properly.
18173 * cfgloop.h (struct loop): Likewise.
18174 * cse.c (struct set): Likewise.
18175 * ipa-utils.c (struct searchc_env): Likewise.
18176 * loop-invariant.c (struct invariant): Likewise.
18177 * lra-remat.c (struct cand): Likewise.
18178 * recog.c (struct change_t): Likewise.
18179 * rtl.h (struct address_info): Likewise.
18180 * symbol-summary.h (function_summary): Likewise.
18181 * tree-loop-distribution.c (struct partition): Likewise.
18182 * tree-object-size.c (struct object_size_info): Likewise.
18183 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
18184 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
18185 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
18186 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
18187 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
18188 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
18189 (struct _stmt_vec_info): Likewise.
18191 2017-03-14 Martin Liska <mliska@suse.cz>
18194 * multiple_target.c (create_dispatcher_calls): Check that
18195 a target can create a function dispatcher.
18197 2017-03-14 Martin Liska <mliska@suse.cz>
18200 * multiple_target.c (expand_target_clones): Drop local.local
18201 flag for default implementation.
18203 2017-03-14 Richard Biener <rguenther@suse.de>
18205 PR tree-optimization/80030
18206 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
18208 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
18210 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
18211 gcc_fallthrough() instead of __attribute__((fallthrough));
18213 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
18215 * doc/gcc.texi: Remove "up" link to (DIR).
18216 * doc/gccint.texi: Ditto.
18218 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
18220 * doc/install.texi (Specific) <avr>: Remove reference to
18223 2017-03-13 Jeff Law <law@redhat.com>
18225 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
18226 attribute rather than comments.
18228 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
18229 match_scratch operand is highest.
18231 2017-03-13 Martin Liska <mliska@suse.cz>
18233 PR middle-end/78339
18234 * ipa-pure-const.c (warn_function_noreturn): If the declarations
18235 is a CHKP clone, use original declaration.
18237 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18239 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
18240 (arc_conditional_register_usage): Use a different allocation order
18241 when optimizing for size.
18242 * common/config/arc/arc-common.c (arc_option_optimization_table):
18243 Section anchors default on when optimizing for size.
18245 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18247 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
18249 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18251 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
18252 * config/arc/arc.md (cpu_facility): Add cd variant.
18253 (*movqi_insn): Add code density variant.
18254 (*movhi_insn): Likewise.
18255 (*movqi_insn): Likewise.
18256 (*addsi3_mixed): Likewise.
18257 (subsi3_insn): Likewise.
18259 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18261 * config/arc/arc.md (movsi_cond_exec): Update constraint.
18263 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
18265 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
18266 expressions with MINUS and UNARY ops.
18268 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18271 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
18273 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
18274 between vec_select and vector argument.
18275 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
18276 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
18277 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
18278 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
18279 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
18280 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
18282 2017-03-13 Richard Biener <rguenther@suse.de>
18285 * params.def (vect-max-peeling-for-alignment): Fix typo.
18287 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18289 * doc/install.texi (Specific) <mips-*-*>: Remove description of
18290 issue that only occurred with binutils below 2.18.
18292 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18294 * doc/install.texi (Specific) <cris-axis-elf>: No longer
18295 refer to binutils 2.11/2.12 minimum.
18297 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
18299 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
18300 ftp.kernel.org and simplify binutils requirement.
18302 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
18304 * doc/invoke.texi (Warning Options): Fix spelling of link-time
18306 (Optimize Options): Ditto. Also remove redundancy.
18308 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18310 PR translation/79848
18311 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
18313 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
18314 to G_ to avoid double translation.
18316 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18318 PR translation/79923
18319 * auto-profile.c (get_combined_location): Convert leading
18320 character of diagnostics to lower case and remove trailing period.
18321 (read_profile): Likewise for various diagnostics.
18322 * config/arm/arm.c (arm_option_override): Remove trailing period
18323 from various diagnostics.
18324 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
18325 (msp430_expand_delay_cycles): Likewise.
18327 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18330 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
18331 full command-line argument, rather than just "str".
18332 (aarch64_validate_march): Likewise.
18333 (aarch64_validate_mtune): Likewise.
18335 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
18337 PR rtl-optimization/78911
18338 * lra-assigns.c (must_not_spill_p): New function.
18339 (spill_for): Use it.
18341 2017-03-10 Jakub Jelinek <jakub@redhat.com>
18343 PR tree-optimization/79981
18344 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
18345 ATOMIC_COMPARE_EXCHANGE ifn result.
18346 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
18347 IFN_ATOMIC_COMPARE_EXCHANGE.
18349 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18352 * opts.c (parse_sanitizer_options): Add missing question mark to
18353 "did you mean" message.
18355 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18357 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
18359 (VMULEUH_UNS): Likewise.
18360 (VMULOUB_UNS): Likewise.
18361 (VMULOUH_UNS): Likewise.
18362 * config/rs6000/rs6000.c (builtin_function_type): Remove
18363 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
18365 2017-03-10 David Malcolm <dmalcolm@redhat.com>
18368 * read-rtl-function.c (function_reader::read_rtx_operand): Update
18369 x with result of extra_parsing_for_operand_code_0.
18370 (function_reader::extra_parsing_for_operand_code_0): Convert
18371 return type from void to rtx, returning x. When reading
18372 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
18373 larger size containing struct block_symbol.
18375 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
18377 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
18378 -mfloat128-hardware without -m64.
18380 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
18383 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
18384 entries to the case statement that marks unsigned arguments to
18385 overloaded functions.
18387 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
18389 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
18390 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
18392 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
18395 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
18396 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
18398 2017-03-10 Martin Liska <mliska@suse.cz>
18402 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
18403 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
18406 2017-03-10 Olivier Hainque <hainque@adacore.com>
18408 * tree-switch-conversion (array_value_type): Start by resetting
18409 candidate type to it's main variant.
18411 2017-03-10 Jakub Jelinek <jakub@redhat.com>
18413 PR rtl-optimization/79909
18414 * combine.c (try_combine): Use simplify_replace_rtx on individual
18415 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
18416 of the whole CALL_INSN_FUNCTION_USAGE.
18418 PR tree-optimization/79972
18419 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
18420 get_range_info on SSA_NAMEs. Formatting fixes.
18422 2017-03-10 Richard Biener <rguenther@suse.de>
18423 Jakub Jelinek <jakub@redhat.com>
18425 PR tree-optimization/77975
18426 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
18427 edge to be constant.
18428 (get_val_for): For constant x return it. Formatting fix.
18429 (loop_niter_by_eval): Avoid pointless looping if the next iteration
18430 would use the same bases as the current one.
18432 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18434 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
18435 instead of vec_select for V1TImode.
18436 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
18438 (VSX_LE_128): Add V1TI to this mode iterator.
18439 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
18440 (*vsx_le_perm_store_<mode>): Likewise.
18441 (pre-reload splitter for VSX stores): Likewise.
18442 (post-reload splitter for VSX stores): Likewise.
18443 (*vsx_xxpermdi2_le_<mode>): Likewise.
18444 (*vsx_lxvd2x2_le_<mode>): Likewise.
18445 (*vsx_stxvd2x2_le_<mode>): Likewise.
18447 2017-03-09 Michael Eager <eager@eagercon.com>
18449 Correct failures with --enable-checking=yes,rtl.
18451 * config/microblaze/microblaze.c (microblaze_expand_shift):
18452 Replace GET_CODE test with CONST_INT_P and INTVAL test with
18453 test for const0_rtx.
18454 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
18455 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
18457 2017-03-09 Richard Biener <rguenther@suse.de>
18459 PR tree-optimization/79977
18460 * graphite-scop-detection.c (scop_detection::merge_sese):
18461 Handle the case of extra exits to blocks dominating the entry.
18463 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
18465 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
18468 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
18470 PR rtl-optimization/79949
18471 * lra-constraints.c (process_alt_operands): Check memory when
18472 trying to predict a cycle. Print about the overall increase.
18474 2017-03-09 Richard Biener <rguenther@suse.de>
18476 PR middle-end/79971
18477 * gimple-expr.c (useless_type_conversion_p): Preserve
18478 TYPE_SATURATING for fixed-point types.
18480 2017-03-09 Richard Biener <rguenther@suse.de>
18483 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
18484 alignment of BLKmode params.
18486 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18489 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
18490 (VALL_NO_V2Q): Likewise.
18492 * config/aarch64/aarch64-simd.md
18493 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
18495 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
18496 VALL_NO_V2Q mode iterator.
18497 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
18499 2017-03-09 Martin Liska <mliska@suse.cz>
18501 PR tree-optimization/79631
18502 * tree-chkp-opt.c (chkp_is_constant_addr): Call
18503 tree_int_cst_sign_bit just for INTEGER constants.
18505 2017-03-09 Martin Liska <mliska@suse.cz>
18509 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
18512 2017-03-09 Marek Polacek <polacek@redhat.com>
18515 * tree.c (inchash::add_expr): Handle TREE_VEC.
18517 2017-03-09 Martin Liska <mliska@suse.cz>
18520 (chkp_narrow_size_and_offset): New function.
18521 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
18522 (void chkp_parse_bit_field_ref): New function.
18523 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
18524 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
18526 2017-03-09 Martin Liska <mliska@suse.cz>
18529 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
18530 (chkp_find_bounds_1): Remove gcc_unreachable.
18532 2017-03-09 Jakub Jelinek <jakub@redhat.com>
18535 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
18536 BUILT_IN_SYNC*, determine the access type from the size suffix and
18537 always build a MEM_REF with that type. Handle forgotten
18538 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
18541 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
18542 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
18543 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
18544 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
18545 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
18546 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
18547 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
18548 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
18549 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
18550 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
18551 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
18552 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
18553 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
18554 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
18555 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
18556 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
18557 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
18558 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
18559 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
18560 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
18561 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
18562 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
18563 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
18564 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
18565 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
18566 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
18567 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
18568 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
18569 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
18570 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
18571 definitions outside of __OPTIMIZE__ guarded section.
18574 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
18575 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
18576 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
18577 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
18580 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18582 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
18583 ("vfenez<mode>"): Add missing constraints.
18585 2017-03-08 Martin Sebor <msebor@redhat.com>
18588 * config/nds32/nds32.c (nds32_option_override):
18589 Fix misspelled diagnostic.
18591 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18594 * gimplify.c (gimplify_omp_for): Replace index var in outer
18595 taskloop statement with an artificial variable and add
18596 OMP_CLAUSE_PRIVATE clause for it.
18598 2017-03-08 Richard Biener <rguenther@suse.de>
18600 PR tree-optimization/79955
18601 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
18602 for accesses that are completely outside of the variable.
18604 2017-03-08 Andrew Haley <aph@redhat.com>
18606 PR tree-optimization/79943
18607 * tree-ssa-loop-split.c (compute_new_first_bound): When
18608 calculating the new upper bound, (END-BEG) should be added, not
18611 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18613 * config/avr/avr.md (setmemhi): Make sure match_dup
18614 operand number comes before match_scratch.
18616 2017-03-08 Richard Biener <rguenther@suse.de>
18618 PR tree-optimization/79920
18619 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
18620 with ncopies == 1 to ...
18621 (vect_transform_slp_perm_load): ... here. Properly compute
18622 all element loads by iterating VF times over the group. Do
18623 not handle ncopies (computed in a broken way) in
18624 vect_create_mask_and_perm.
18626 2017-03-08 Jakub Jelinek <jakub@redhat.com>
18629 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
18630 is a uniform vector, use uniform_vector_p return value instead of
18631 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
18633 2017-03-07 Marek Polacek <polacek@redhat.com>
18635 PR middle-end/79809
18636 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
18637 (alloca_call_type): Likewise.
18639 2017-03-07 Martin Liska <mliska@suse.cz>
18641 * gcov.c (process_args): Put comment to correct location.
18643 2017-03-07 Martin Liska <mliska@suse.cz>
18645 PR middle-end/68270
18646 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
18647 Use array_at_struct_end_p instead of DECL_CHAIN (field).
18648 (chkp_narrow_bounds_for_field): Likewise.
18649 (chkp_parse_array_and_component_ref): Pass one more argument to
18652 2017-03-07 Richard Biener <rguenther@suse.de>
18654 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
18657 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
18659 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
18660 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
18662 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18665 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
18666 to end of description.
18667 (PARAM_MAX_STORES_TO_MERGE): Likewise.
18669 2017-03-07 Jakub Jelinek <jakub@redhat.com>
18671 PR rtl-optimization/79901
18672 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
18674 (*avx512f_<code><mode>3<mask_name>): ... this.
18675 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
18676 iterator instead of VI8_AVX2_AVX512BW.
18678 PR rtl-optimization/79901
18679 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
18680 min/max expander, expand it using expand_vec_cond_expr.
18683 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
18686 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18689 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
18690 to void * for PCH reasons.
18691 * dwarf2out.c (output_loc_operands, output_die): Cast
18692 v.val_vec.array to unsigned char *.
18694 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
18697 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
18700 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
18702 PR rtl-optimization/79571
18703 * lra-constraints.c (process_alt_operands): Calculate static
18704 reject and subtract it from overall when only addresses will be
18707 2017-03-06 Julia Koval <julia.koval@intel.com>
18710 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
18711 incoming stack boundary to 128 for 64-bit targets.
18713 2017-03-06 Richard Biener <rguenther@suse.de>
18715 PR tree-optimization/79894
18716 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
18717 to NULL after folding it.
18719 2017-03-06 Richard Biener <rguenther@suse.de>
18721 PR tree-optimization/79824
18722 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
18723 check disabling peeling for gaps.
18725 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
18727 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
18728 attributes): Document gettimeofday.
18730 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18732 * config/s390/s390.c (s390_option_override_internal): Set
18733 PARAM_MIN_VECT_LOOP_BOUND
18735 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
18737 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
18738 * config/s390/s390.md: Likewise.
18740 2017-03-06 Jakub Jelinek <jakub@redhat.com>
18743 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
18744 (<avx2_avx512>_perm<mode>): Rename to ...
18745 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
18747 (<avx512>_perm<mode>_mask): Rename to ...
18748 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
18750 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
18751 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
18752 instead of VI8F_256_512.
18753 (avx512f_perm<mode>): New define_expand.
18754 (avx512f_perm<mode>_mask): Likewise.
18755 (avx512f_perm<mode>_1<mask_name>): New define_insn.
18756 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
18758 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18760 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
18761 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
18763 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
18765 2017-03-06 Martin Liska <mliska@suse.cz>
18768 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
18769 when having a SSA NAME w/o VAR_DECL assigned to it.
18771 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18773 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
18774 msa_dpsub_<su>_d): Fix MODE for vec_select.
18776 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
18778 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
18780 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
18782 2017-03-06 Richard Biener <rguenther@suse.de>
18784 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
18785 * plugin.c (register_plugin_info): Likewise.
18786 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
18788 2017-03-05 Jakub Jelinek <jakub@redhat.com>
18790 * config/i386/sse.md (sse_storehps, sse_storelps,
18791 avx_<castmode><avxsizesuffix>_<castmode>,
18792 avx512f_<castmode><avxsizesuffix>_<castmode>,
18793 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
18794 in condition that at least one operand is not a MEM.
18796 2017-03-03 Jakub Jelinek <jakub@redhat.com>
18798 PR middle-end/79805
18799 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
18800 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
18802 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
18803 gimple_call_nothrow_p flag based on whether original builtin can throw.
18804 If it can, emit following stmts on the fallthrough edge.
18805 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
18806 don't create new bb if inserting just debug stmts on the edge, try to
18807 insert them on the fallthru bb or just reset debug stmts.
18809 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
18812 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
18813 restore recog_data (including the operand rtxes inside it) around
18814 the call to get_insn_template.
18816 2017-03-03 Martin Sebor <msebor@redhat.com>
18818 PR tree-optimization/79699
18819 * context.c (context::~context): Free MPFR caches to avoid
18820 a memory leak on program exit.
18822 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18824 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
18825 Use wide_int::ulow () instead of .elt (0).
18827 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18829 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
18830 (*pushxf): Limit oF constraint to 32bit targets and add oC
18831 constraint for 64bit targets.
18832 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
18833 (*pushdf): Change rmF constraint to rmC.
18835 2017-03-03 Martin Liska <mliska@suse.cz>
18837 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
18838 Remove unused variable.
18840 2017-03-03 Jakub Jelinek <jakub@redhat.com>
18843 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
18844 is a memory operand, increase num_memory.
18845 (ix86_expand_args_builtin): Likewise.
18847 2017-03-03 Jan Hubicka <jh@suse.cz>
18850 * ipa-devirt.c (maybe_record_node): Properly handle
18851 __cxa_pure_virtual visibility.
18853 2017-03-03 Martin Liska <mliska@suse.cz>
18855 PR tree-optimization/79803
18856 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
18858 (pass_loop_prefetch::execute): Disabled optimization if an
18859 assumption about L1 cache size is not met.
18861 2017-03-03 Martin Liska <mliska@suse.cz>
18863 PR rtl-optimization/79574
18864 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
18865 (hash_scan_set): Likewise.
18866 (dump_hash_table): Likewise.
18867 (hoist_code): Likewise.
18869 2017-03-03 Richard Biener <rguenther@suse.de>
18871 * fixed-value.c (fixed_from_string): Restore use of elt (1)
18872 in place of uhigh ().
18873 (fixed_convert_from_real): Likewise.
18875 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
18878 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
18880 2017-03-03 Richard Biener <rguenther@suse.de>
18882 PR middle-end/79818
18883 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
18884 TYPE_OVERFLOW_UNDEFINED check.
18886 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18888 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
18890 (vector_ae_<mode>_p): Likewise.
18891 (vector_nez_<mode>_p): Likewise.
18892 (vector_ne_v2di_p): Likewise.
18893 (vector_ae_v2di_p): Likewise.
18894 (vector_ne_<mode>_p): Likewise.
18895 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
18897 (vsx_tsqrt<mode>2_fe): Likewise.
18899 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
18902 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
18904 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18906 PR rtl-optimization/79780
18907 * cprop.c (one_cprop_pass): When second and further conditional trap
18908 in a single basic block is turned into an unconditional trap, turn it
18909 into a deleted note to avoid RTL verification failures.
18911 2017-03-02 Richard Biener <rguenther@suse.de>
18913 * fold-const.c (const_binop): Use ulow () instead of elt (0).
18915 2017-03-02 Richard Biener <rguenther@suse.de>
18917 PR tree-optimization/79345
18919 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
18920 param and abort the walk, returning -1 if it is hit.
18921 (walk_aliased_vdefs): Take a limit param and pass it on.
18922 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
18923 defaulting to 0 and return a signed int.
18924 * tree-ssa-uninit.c (struct check_defs_data): New struct.
18925 (check_defs): New helper.
18926 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
18927 about uninitialized memory.
18928 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
18929 bogus uninitialized warning.
18930 (fixed_convert_from_real): Likewise.
18932 2017-03-02 Bin Cheng <bin.cheng@arm.com>
18934 PR tree-optimization/66768
18935 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
18936 iv_use if base object can't be determined.
18938 2017-03-02 Jakub Jelinek <jakub@redhat.com>
18940 PR tree-optimization/79345
18941 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
18942 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
18943 (get_pattern_stats): Initialize it.
18944 * genemit.c (gen_expand): Verify match_scratch numbers come after
18945 match_operand/match_dup numbers.
18946 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
18947 match_scratch numbers.
18948 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
18950 * config/s390/s390.md (trunctdsd2): Likewise.
18952 2017-03-02 Richard Biener <rguenther@suse.de>
18954 * wide-int.h (wide_int_storage::operator=): Implement in terms
18957 2017-03-02 Richard Biener <rguenther@suse.de>
18959 PR tree-optimization/79777
18960 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
18961 the to insert expression to sth existing.
18963 2017-03-01 Martin Sebor <msebor@redhat.com>
18965 PR middle-end/79692
18966 * gimple-ssa-sprintf.c
18967 (directive::known_width_and_precision): New function.
18968 (format_integer): Use it.
18969 (get_mpfr_format_length): Consider the full range of precision
18970 when computing %g output with the # flag. Set the likely byte
18971 count to 3 rather than 1 when precision is indeterminate.
18972 (format_floating): Correct the lower bound of precision.
18974 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18976 * doc/invoke.texi: Document default code model for 64-bit Linux.
18978 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18981 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
18982 udiv rather than div since input pattern is unsigned.
18984 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
18986 * config/i386/i386.c (print_reg): Warn for values of
18987 unsupported size in integer register.
18989 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
18992 * config/rs6000/predicates.md (current_file_function_operand): Do
18993 not allow self calls to be local if the function is replaceable.
18995 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
18998 * config/rs6000/altivec.h (vec_ctz and others): Change the
18999 preprocessor macro that controls conditional compilation from
19000 _ARCH_PWR9 to __POWER9_VECTOR__.
19001 (vec_all_ne): Change parameterization of __altivec_scalar_pred
19002 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
19003 control (instead of _ARCH_PWR9 control) so that template
19004 definition uses power9-specific function.
19005 (vec_any_eq): Likewise.
19006 (vec_all_ne): Change macro definition to use a power9-specific
19007 expansion under #ifdef __POWER9_VECTOR__ control (instead of
19008 _ARCH_PWR9 control).
19009 (vec_any_eq) Likewise.
19010 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
19011 expansion for CMPNEF to remove support for xvcmpnesp instruction.
19012 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
19013 support for xvcmpnedp instruction.
19014 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
19015 macro expansion so that Power9 implementation of vec_all_ne does
19016 not use the AltiVec predicate framework.
19017 (VCMPNEH_P): Likewise.
19018 (VCMPNEW_P): Likewise.
19019 (VCMPNED_P): Likewise.
19020 (VCMPNEFP_P): Likewise.
19021 (VCMPNEDP_P): Likewise.
19022 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
19023 implementation of vec_any_eq to not use AltiVec predicate
19025 (VCMPAEH_P): Likewise.
19026 (VCMPAEW_P): Likewise.
19027 (VCMPAED_P): Likewise.
19028 (VCMPAEFP_P): Likewise.
19029 (VCMPAEDP_P): Likewise.
19030 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
19031 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
19032 not use the AltiVec predicate framework.
19033 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
19034 of vec_any_eq to not use AltiVec predicate framework.
19035 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
19036 support for predefined __POWER9_VECTOR__ macro to indicate that
19037 Power9 instruction selection is enabled.
19038 (altivec_overloaded_builtins): Remove extraneous
19039 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
19040 function argument types RS6000_BTI_bool_V16QI and
19041 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
19042 entry for overloaded function argument types RS6000_BTI_bool_V4SI
19043 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
19044 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
19045 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
19046 Power9 for implementations of vec_cmpne. Change the signature for
19047 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
19048 (representing vec_all_ne) to remove the previously described first
19049 argument of type RS6000_BTI_INTSI, as this was an artifact of
19050 reliance on the AltiVec predicate framework, which is no longer
19051 used in the implementation of these functions. Add
19052 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
19053 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
19054 since, unlike the AltiVec predicate framework implementation, we
19055 do not share function descriptors between vec_alle and vec_anyeq.
19056 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
19057 set of modes that receive special treatment even when
19058 TARGET_P9_VECTOR is true. The special treatment emits code that
19059 does not depend on Power9 instructions.
19060 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
19061 define_expand to not rely on AltiVec predicate framework.
19062 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
19064 (vector_ne_v2di_p): Change this define_expand to not rely on
19065 AltiVec predicate framework.
19066 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
19068 (vector_ne_<mode>_p): Change this define_expand to not rely on
19069 AltiVec predicate framework.
19070 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
19072 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
19073 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
19074 define_insn pattern.
19075 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
19076 define_insn pattern because the xvcmpne<VSs>. instruction is not
19078 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
19079 instruction is not supported.
19081 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19083 * config/nvptx/nvptx.c: Include intl.h.
19085 2017-03-01 Martin Jambor <mjambor@suse.cz>
19088 * ipa-prop.h (ipa_bits): Removed field known.
19089 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
19090 to pointers. Adjusted their comments to warn about their sharing.
19091 (ipcp_transformation_summary): Change bits to a vector of pointers.
19092 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
19093 (ipa_get_ipa_bits_for_value): Declare.
19094 * tree-vrp.h (value_range): Mark as GTY((for_user)).
19095 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
19096 (ipa_bits_hash_table): Likewise.
19097 (ipa_vr_ggc_hash_traits): Likewise.
19098 (ipa_vr_hash_table): Likewise.
19099 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
19100 being pointers and vr_known being removed.
19101 (ipa_set_jf_unknown): Likewise.
19102 (ipa_get_ipa_bits_for_value): New function.
19103 (ipa_set_jfunc_bits): Likewise.
19104 (ipa_get_value_range): New overloaded functions.
19105 (ipa_set_jfunc_vr): Likewise.
19106 (ipa_compute_jump_functions_for_edge): Use the above functions to
19107 construct bits and vr parts of jump functions.
19108 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
19109 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
19111 (ipcp_grow_transformations_if_necessary): Also allocate
19112 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
19114 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
19115 them. Fix too long lines.
19116 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
19117 vr_known being removed.
19118 (ipa_read_jump_function): Use new setter functions to construct bits
19119 and vr parts of jump functions or set them to NULL.
19120 (write_ipcp_transformation_info): Adjust for bits being pointers.
19121 (read_ipcp_transformation_info): Likewise.
19122 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
19124 Include gt-ipa-prop.h.
19125 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
19127 (ipcp_store_bits_results): Likewise.
19128 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
19129 Do not write to existing jump functions but use a temporary instead.
19131 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19134 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
19135 attempt to use its first operand as BIT_FIELD_REF base.
19137 2017-03-01 Richard Biener <rguenther@suse.de>
19139 PR middle-end/79721
19140 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
19141 interpolating formula in wrapping arithmetic.
19142 (chrec_apply): Convert chrec_evaluate return value to wanted type.
19144 2017-03-01 Jakub Jelinek <jakub@redhat.com>
19146 PR tree-optimization/79734
19147 * tree-vect-generic.c (expand_vector_condition): Optimize
19148 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
19149 Handle VEC_COND_EXPR where comparison has different inner width from
19150 type's inner width.
19152 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
19154 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
19155 markup, and similar issues. Remove @opindex entries for things
19156 that aren't options. Add missing -mmpy-option entries.
19158 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19160 PR tree-optimization/79737
19161 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
19162 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
19163 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
19164 instead of byte_size. Formatting fix.
19165 (shift_bytes_in_array_right): Formatting fix.
19167 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
19170 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
19171 condition on optimize for the leaf function test.
19173 2017-02-28 Martin Liska <mliska@suse.cz>
19176 * read-rtl-function.c (function_reader::handle_unknown_directive):
19177 Bail out when one uses -flto.
19179 2017-02-28 Martin Liska <mliska@suse.cz>
19181 * common.opt: Replace space with tabular for options of <number>
19183 * config/i386/i386.opt: Show <number> value for
19184 -mlarge-data-threshold.
19185 * opts.c (print_filtered_help): Do not display number in hexadecimal
19188 2017-02-28 Martin Liska <mliska@suse.cz>
19190 * common.opt: Fix --help=option -Q for options which are of
19193 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
19195 * config/i386/i386.c (print_reg): Error out for values
19196 of 8-bit size in invalid integer register.
19198 2017-02-28 Martin Sebor <msebor@redhat.com>
19200 PR tree-optimization/79691
19201 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
19203 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19206 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
19207 gcc_unreachable with output_operand_lossage.
19209 2017-02-28 Richard Biener <rguenther@suse.de>
19211 PR tree-optimization/79740
19212 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
19214 (visit_nary_op): Insert the nary into the hashtable if we
19215 pattern-matched sth.
19216 * tree-ssa-pre.c (eliminate_insert): Robustify.
19218 2017-02-28 Richard Biener <rguenther@suse.de>
19220 PR middle-end/79731
19221 * fold-const.c (decode_field_reference): Reject out-of-bound
19224 2017-02-28 Jakub Jelinek <jakub@redhat.com>
19226 * config/i386/i386.c: Include intl.h.
19227 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
19228 instead of just cond ? "..." : "...".
19229 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
19230 * coverage.c (read_counts_file): Likewise.
19231 * omp-offload.c: Include intl.h.
19232 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
19233 of just cond ? "..." : "...".
19234 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
19235 of just cond ? "..." : "...".
19237 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
19240 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
19242 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
19243 'tune for' CPU name.
19244 * config/arm/arm-cpu-data.h: Regenerated.
19246 2017-02-28 Richard Biener <rguenther@suse.de>
19248 PR tree-optimization/79732
19249 * tree-inline.c (expand_call_inline): Do not shadow var.
19251 2017-02-28 Richard Biener <rguenther@suse.de>
19253 PR tree-optimization/79723
19254 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
19255 address-space properly.
19257 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
19259 * doc/optinfo.texi (Optimization groups): Fix option used for
19261 * doc/invoke.texi (-fopt-info): Document "omp".
19262 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
19263 (OPTGROUP_ALL): Add OPTGROUP_OMP.
19264 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
19265 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
19266 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
19268 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
19270 * dumpfile.c (optgroup_options): Instead of "openmp", associate
19271 OPTGROUP_OMP with "omp".
19273 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
19276 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
19277 for arithmetic shift of unsigned V2DI.
19279 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
19281 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
19282 arc/linux.h headers.
19283 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
19284 (LINK_SPEC): Likewise.
19285 (ARC_TLS_EXTRA_START_SPEC): Likewise.
19286 (EXTRA_SPECS): Likewise.
19287 (STARTFILE_SPEC): Likewise.
19288 (ENDFILE_SPEC): Likewise.
19289 (LIB_SPEC): Likewise.
19290 (TARGET_SDATA_DEFAULT): Likewise.
19291 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
19292 (MULTILIB_DEFAULTS): Likewise.
19293 (DWARF2_UNWIND_INFO): Likewise.
19294 * config/arc/big.h: New file.
19295 * config/arc/elf.h: Likewise.
19296 * config/arc/linux.h: Likewise.
19297 * config/arc/t-uClibc: Remove.
19299 2017-02-27 Bin Cheng <bin.cheng@arm.com>
19301 PR tree-optimization/77536
19302 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
19303 (tree_transform_and_unroll_loop): Use above function to compute the
19304 estimated niter of unrolled loop and use it when scaling profile.
19305 Also use count info rather than frequency if it's non-zero.
19306 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
19307 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
19308 (vect_transform_loop): Call above function.
19310 2017-02-27 Richard Biener <rguenther@suse.de>
19312 PR tree-optimization/45397
19313 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
19314 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
19315 (visit_nary_op): Add pattern matching for CSEing sign-changed
19316 or truncated operations with wider ones.
19318 2017-02-27 Richard Biener <rguenther@suse.de>
19320 PR tree-optimization/79690
19321 * tree-vect-stmts.c (vectorizable_store): Use vector type
19322 built from the DR with address-space.
19324 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
19326 * doc/invoke.texi (Optimize Options): Refine the description
19327 of asan-use-after-return.
19329 2017-02-25 Alan Modra <amodra@gmail.com>
19331 PR rtl-optimization/79584
19332 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
19333 base, not ad->base_term, the reg within base. Remove assertion
19334 that ad->base == ad->base_term. Replace gen_int_mode using
19335 bogus mode with const0_rtx.
19337 2017-02-25 Jakub Jelinek <jakub@redhat.com>
19339 PR middle-end/79396
19340 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
19341 FMA_EXPR like tcc_binary or tcc_unary.
19343 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
19346 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
19348 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
19349 (output_loc_operands): Handle DW_OP_call_ref and
19350 DW_OP_GNU_variable_value.
19351 (struct variable_value_struct): New type.
19352 (struct variable_value_hasher): Likewise.
19353 (variable_value_hash): New variable.
19354 (string_types): Remove.
19355 (copy_loc_descr): New function.
19356 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
19357 (prepend_loc_descr_to_each): New function.
19358 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
19359 instead of add_loc_descr_to_each if the first argument is single
19360 location list and the second has multiple.
19361 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
19362 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
19363 when looking for variable value which doesn't have other location info.
19364 (loc_list_from_tree): Formatting fix.
19365 (gen_array_type_die): Simplify DW_AT_string_length handling.
19366 (adjust_string_types): Remove.
19367 (gen_subprogram_die): Don't call adjust_string_types nor test/set
19368 string_types. Call resolve_variable_values.
19369 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
19370 (resolve_addr_in_expr): Likewise. Add A argument.
19371 (copy_deref_exprloc): Remove deref argument. Adjust for the
19372 original expression being DW_OP_GNU_variable_value with optionally
19373 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
19374 optionally after it.
19375 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
19376 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
19377 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
19378 (variable_value_hasher::hash, variable_value_hasher::equal): New
19380 (resolve_variable_value_in_expr, resolve_variable_value,
19381 resolve_variable_values, note_variable_value_in_expr,
19382 note_variable_value): New functions.
19383 (dwarf2out_early_finish): Call note_variable_value on all toplevel
19386 2017-02-24 Jakub Jelinek <jakub@redhat.com>
19389 * opts.h (handle_generated_option): Add GENERATED_P argument.
19390 * opts-common.c (handle_option): Adjust function comment.
19391 (handle_generated_option): Add GENERATED_P argument, pass it to
19393 (control_warning_option): Pass false to handle_generated_option
19395 * opts.c (maybe_default_option): Pass true to handle_generated_option
19397 * optc-gen.awk: Likewise.
19399 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19401 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
19402 a REG, look at the REG it is a SUBREG of.
19403 (splitter for cmpeqsi_t): Ditto.
19405 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19407 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
19408 the special USEs with the pattern of the insn, not the insn itself.
19410 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
19413 * doc/invoke.texi: Document -mload-store-pairs.
19415 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19416 Sandra Loosemore <sandra@codesourcery.com>
19418 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
19419 argument isn't a CONST_INT.
19420 (nios2_alternate_compare_const): Assert op is a CONST_INT.
19421 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
19422 (nios2_validate_compare): Bypass alternate compare logic if *op2
19423 is not a CONST_INT.
19424 (ldstwm_operation_p): Return false if first_base is not a REG or
19425 if first_offset is not a CONST_INT.
19427 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19429 * config/cris/cris.md: Use correct operand in a define_peephole2.
19431 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19433 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
19435 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
19437 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
19438 this_insn if it is an INSN or JUMP_INSN.
19439 (force_offsettable): Look at base, not at addr.
19440 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
19441 on things that aren't necessarily CONST_INTs.
19443 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
19445 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
19446 -mfpmath=sse is the default also for x86-32 targets with SSE2
19447 instruction set when @option{-ffast-math} is enabled
19449 2017-02-24 Jeff Law <law@redhat.com>
19451 PR rtl-optimizatoin/79286
19452 * ira.c (update_equiv_regs): Drop may_trap_p exception to
19455 2017-02-24 Richard Biener <rguenther@suse.de>
19457 PR tree-optimization/79389
19458 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
19461 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
19463 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
19464 function comment to reflect reality.
19465 (loop_exits_before_overflow): Fix typo in function description.
19467 2017-02-24 Richard Biener <rguenther@suse.de>
19469 PR tree-optimization/79389
19470 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
19471 properly that a threading opportunity exists. Detect conditional
19472 copy/constant propagation opportunities.
19474 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
19476 * config/visium/visium.md (type): Add trap.
19477 (b): New mode attribute.
19478 (*btst): Rename into...
19479 (*btst<mode>): ...this and adjust.
19480 (*cbranchsi4_btst_insn): Rename into...
19481 (*cbranch<mode>4_btst_insn): ...this and adjust.
19482 (trap): New define_insn.
19484 2017-02-23 Jakub Jelinek <jakub@redhat.com>
19486 PR tree-optimization/79389
19487 * ifcvt.c (struct noce_if_info): Add rev_cond field.
19488 (noce_reversed_cond_code): New function.
19489 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
19490 reversed_comparison_code. Formatting fix.
19491 (noce_try_store_flag): Test rev_cond != NULL in addition to
19492 reversed_comparison_code.
19493 (noce_try_store_flag_constants): Likewise.
19494 (noce_try_store_flag_mask): Likewise.
19495 (noce_try_addcc): Use rev_cond if non-NULL instead of
19496 reversed_comparison_code.
19497 (noce_try_cmove_arith): Likewise. Formatting fixes.
19498 (noce_try_minmax, noce_try_abs): Clear rev_cond.
19499 (noce_find_if_block): Initialize rev_cond.
19500 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
19501 instead of false as last argument never attempt to reverse it
19504 2017-02-23 Bin Cheng <bin.cheng@arm.com>
19506 PR tree-optimization/79663
19507 * tree-predcom.c (combine_chains): Process refs in reverse order
19508 only for ZERO length chains, and add explaining comment.
19510 2017-02-23 Jeff Law <law@redhat.com>
19512 PR tree-optimization/79578
19513 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
19514 in call to operand_equal_p.
19516 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
19519 * config/i386/cpuid.h: Fix another undefined behavior.
19521 2017-02-23 Richard Biener <rguenther@suse.de>
19523 PR tree-optimization/79683
19524 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
19525 vector types for data-refs.
19527 2017-02-23 Martin Liska <mliska@suse.cz>
19529 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
19531 2017-02-23 Jakub Jelinek <jakub@redhat.com>
19533 PR middle-end/79665
19534 * internal-fn.c (get_range_pos_neg): Moved to ...
19535 * tree.c (get_range_pos_neg): ... here. No longer static.
19536 * tree.h (get_range_pos_neg): New prototype.
19537 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
19538 are known to be in between 0 and signed maximum inclusive, try to
19539 expand both unsigned and signed divmod and use the cheaper one from
19542 2017-02-22 Jeff Law <law@redhat.com>
19544 PR tree-optimization/79578
19545 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
19546 to compare base operands.
19548 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
19551 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
19552 gpc_reg_operand instead of fpr_reg_operand.
19554 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
19556 * config/mips/mips.c (mips_return_in_memory): Force FP
19557 vector types to be returned in memory for o32 ABI.
19559 2017-02-22 Jakub Jelinek <jakub@redhat.com>
19561 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
19562 instead of DW_TAG_member for static data member declarations and don't
19563 set no_linkage_name for static inline data members.
19564 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
19567 2017-02-22 Martin Liska <mliska@suse.cz>
19569 * doc/invoke.texi: Replace inequality signs with square brackets
19572 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19574 PR tree-optimization/68644
19575 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
19577 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
19580 * lra-constraints.c (simplify_operand_subreg): Handle
19581 WORD_REGISTER_OPERATIONS targets.
19583 2017-02-22 Jakub Jelinek <jakub@redhat.com>
19586 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
19587 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
19588 elimination by swapping fld*.
19590 2017-02-22 Richard Biener <rguenther@suse.de>
19592 PR tree-optimization/79673
19593 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
19594 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
19595 irrelevant address-space qualifiers and avoiding a
19596 ADDR_SPACE_CONVERT_EXPR from fold_convert.
19598 2017-02-22 Richard Biener <rguenther@suse.de>
19600 PR tree-optimization/79666
19601 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
19602 to not symbolically negate if that may introduce undefined
19605 2017-02-22 Martin Liska <mliska@suse.cz>
19608 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
19609 * data-streamer-out.c (streamer_write_gcov_count_stream):
19611 * value-prof.c (stream_out_histogram_value): Make assert more
19612 precise based on type of counter.
19614 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
19617 * config/i386/i386.md (standard_x87sse_constant_load splitter):
19618 Use nonimmediate_operand instead of memory_operand for operand 1.
19619 (float-extend standard_x87sse_constant_load splitter): Ditto.
19621 2017-02-21 Jeff Law <law@redhat.com>
19623 PR tree-optimization/79621
19624 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
19625 blocks with edges to themselves.
19627 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19630 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
19631 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
19632 Use gimple_call_builtin_p.
19635 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
19636 on temporarily removed DEBUG_INSNs.
19638 PR tree-optimization/79649
19639 * tree-loop-distribution.c (classify_partition): Give up on
19640 non-generic address space loads/stores.
19642 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
19644 * doc/loop.texi (Loop manipulation): Remove nonexistent
19645 tree_ssa_loop_version from the documentation.
19646 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
19648 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19651 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
19652 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
19653 * config/rs6000/rs6000.c: Include except.h.
19654 (rs6000_expand_split_stack_prologue): Call
19655 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
19657 2017-02-21 Martin Jambor <mjambor@suse.cz>
19660 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
19661 have been analyzed.
19663 2017-02-21 Martin Jambor <mjambor@suse.cz>
19665 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
19666 for backward compatibility only.
19667 * doc/invoke.texi (Option Summary): Remove all references to
19668 -fipa-cp-alignment.
19670 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
19674 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19676 * lra-constraints.c (curr_insn_transform): Handle
19677 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19679 2017-02-21 Martin Liska <mliska@suse.cz>
19681 * config/i386/i386.opt: Replace -masm-dialect with -masm.
19683 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
19685 PR translation/79638
19686 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
19688 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
19691 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
19692 (arm_function_ok_for_sibcall): Return false for an indirect call by
19693 descriptor if all the argument registers are used.
19694 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
19695 alignment of the function.
19697 2017-02-21 Jakub Jelinek <jakub@redhat.com>
19699 PR tree-optimization/61441
19700 * simplify-rtx.c (simplify_const_unary_operation): For
19701 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
19702 the sNaN unmodified.
19704 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
19706 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
19707 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
19708 instead of SYSTEM_HEADER_DIR.
19710 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
19711 Martin Liška <mliska@suse.cz>
19713 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
19714 Fix typos and grammar, use active voice, and clarify.
19716 2017-02-20 Marek Polacek <polacek@redhat.com>
19718 PR middle-end/79537
19719 * gimplify.c (gimplify_expr): Handle unused *&&L;.
19722 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
19724 2017-02-20 Jakub Jelinek <jakub@redhat.com>
19727 * config/i386/i386.c (ix86_expand_builtin): Handle
19728 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
19729 ix86_builtins_isa[fcode].isa as a requirement of those
19730 flags and any other flag in the bitmask.
19731 (ix86_init_mmx_sse_builtins): Use 0 instead of
19732 ~OPTION_MASK_ISA_64BIT as mask.
19733 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
19734 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
19735 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
19736 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
19738 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19741 * lra-constraints.c (split_reg): Check requested split mode
19742 is supported by the register.
19744 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19746 * lra-constraints.c (simplify_operand_subreg): Remove early
19749 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19752 * lra-constraints.c (curr_insn_transform): Tighten condition
19753 for converting SUBREG reloads from OP_OUT to OP_INOUT.
19755 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
19758 * lra-constraints.c (curr_insn_transform): Handle
19759 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
19761 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
19764 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
19766 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
19768 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
19771 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
19774 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
19776 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
19779 2017-02-18 Jakub Jelinek <jakub@redhat.com>
19781 * final.c (last_columnnum, override_columnnum): New variables.
19782 (final_start_function): Set last_columnnum, pass it to begin_prologue
19783 hook and pass 0 to dwarf2out_begin_prologue.
19784 (final_scan_insn): Update override_columnnum. Pass last_columnnum
19785 to source_line debug hook.
19786 (notice_source_line): Compute last_columnnum and for debug_column_info
19787 return true on column changes.
19788 * debug.h (struct gcc_debug_hooks): Add column argument to
19789 source_line and begin_prologue hooks.
19790 (debug_nothing_int_charstar_int_bool): Remove prototype.
19791 (debug_nothing_int_int_charstar,
19792 debug_nothing_int_int_charstar_int_bool): New prototypes.
19793 (dwarf2out_begin_prologue): Add column argument.
19794 * debug.c (do_nothing_debug_hooks): Adjust source_line and
19795 begin_prologue hooks.
19796 (debug_nothing_int_charstar_int_bool): Remove.
19797 (debug_nothing_int_int_charstar,
19798 debug_nothing_int_int_charstar_int_bool): New functions.
19799 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
19800 through to dwarf2out_source_line.
19801 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
19802 (dwarf2out_source_line): Add column argument, emit it if requested.
19803 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
19805 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19806 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
19807 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
19808 through to dwarf2out_begin_prologue.
19809 (vmsdbgout_source_line): Add column argument, pass it through to
19810 dwarf2out_source_line.
19811 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
19812 dbxout_source_line caller.
19813 (dbxout_source_line): Add column argument.
19815 * common.opt (gno-column-info, gcolumn-info): New options.
19816 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
19817 (check_die): Also test for multiple DW_AT_decl_column attributes.
19818 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
19819 DW_AT_decl_column if requested.
19820 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
19822 (gen_variable_die): Likewise.
19823 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
19824 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
19827 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
19828 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
19829 (ix86_handle_option): Handle OPT_m3dnowa.
19830 * doc/invoke.texi (-m3dnowa): Document.
19831 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
19832 -m3dnowa instead of -m3dnow -march=athlon.
19835 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
19836 instead of gcc_assert for K, r and R code checks. Formatting fixes.
19838 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19841 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
19842 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
19843 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
19844 generator for vsx_xxpermdi_<mode>_be.
19845 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
19846 force big-endian semantics.
19847 (vsx_xxpermdi_<mode>_be): New define_expand with same
19848 implementation as previous version of vsx_xxpermdi_<mode>.
19850 2017-02-17 Jakub Jelinek <jakub@redhat.com>
19852 PR tree-optimization/79327
19853 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
19854 variable, its initialization and use.
19856 2017-02-17 Julia Koval <julia.koval@intel.com>
19858 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
19859 (OPTION_MASK_ISA_PKU_UNSET): New.
19860 (ix86_handle_option): Handle -mrdpid.
19861 * config/i386/cpuid.h (bit_RDPID): New.
19862 * config/i386/driver-i386.c (host_detect_local_cpu):
19863 Detect RDPID feature.
19864 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
19865 * config/i386/i386-c.c (ix86_target_macros_internal):
19867 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
19868 (ix86_valid_target_attribute_inner_p): Add "rdpid".
19869 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
19870 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
19871 * config/i386/i386.md (define_insn "rdpid"): New.
19872 * config/i386/i386.opt Add -mrdpid.
19873 * config/i386/immintrin.h (_rdpid_u32): New.
19875 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
19877 PR rtl-optimization/79541
19878 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
19879 instead of transforming it into USE.
19881 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
19883 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
19884 If HONOR_SNANS (SFmode) force the input to a register.
19885 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
19886 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
19887 an frsp or similar insn.
19889 2017-02-17 Martin Liska <mliska@suse.cz>
19891 PR rtl-optimization/79577
19892 * params.def (selsched-max-sched-times): Increase minimum to 1.
19894 2017-02-17 Martin Liska <mliska@suse.cz>
19896 PR rtl-optimization/79574
19897 * gcse.c (want_to_gcse_p): Prevent integer overflow.
19899 2017-02-17 Martin Liska <mliska@suse.cz>
19901 PR tree-optimization/79529
19902 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
19903 ssa_defined_default_def_p to handle cases which are implicitly
19905 * tree-ssa.c (ssa_defined_default_def_p): New function.
19906 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
19907 which are implicitly defined.
19908 * tree-ssa.h (ssa_defined_default_def_p): Declare.
19910 2017-02-17 Richard Biener <rguenther@suse.de>
19912 PR middle-end/79576
19913 * params.def (max-ssa-name-query-depth): Limit to 10.
19915 2017-02-17 Richard Biener <rguenther@suse.de>
19917 PR tree-optimization/79552
19918 * tree-ssa-structalias.c (visit_loadstore): Properly verify
19921 2017-02-17 Richard Biener <rguenther@suse.de>
19924 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
19926 2017-02-17 Marek Polacek <polacek@redhat.com>
19928 PR middle-end/79536
19929 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
19930 (fold_negate_expr): New wrapper.
19932 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
19934 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
19935 Correct terminology and de-emphasize pre-standard behavior.
19937 2017-02-16 Alan Modra <amodra@gmail.com>
19939 PR rtl-optimization/79286
19940 * ira.c (def_dominates_uses): New function.
19941 (update_equiv_regs): Don't create an equivalence for insns that
19942 may trap where the register def does not dominate the use.
19944 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
19946 PR rtl-optimization/78127
19947 * lra.c (lra): Call lra_eliminate before finish the loop after
19950 2017-02-16 Richard Biener <rguenther@suse.de>
19952 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
19954 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
19955 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
19956 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
19957 (isl_val_int_from_wi): New function.
19958 (extract_affine_gmp): Rename to ...
19959 (extract_affine_wi): ... this, take a widest_int.
19960 (extract_affine_int): Just wrap extract_affine_wi.
19961 (add_param_constraints): Use isl_val_int_from_wi.
19962 (add_loop_constraints): Likewise, and extract_affine_wi.
19964 2017-02-15 Jeff Law <law@redhat.com>
19966 PR middle-end/79521
19967 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
19968 ira_init_register_move_cost_if_necessary.
19970 2017-02-15 Martin Sebor <msebor@redhat.com>
19972 PR middle-end/32003
19973 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
19974 removed in a prior commit.
19976 2017-02-15 Bin Cheng <bin.cheng@arm.com>
19978 PR tree-optimization/79347
19979 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
19980 counters during peeling.
19982 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
19984 * Makefile.in (site.exp): Remove "set ISLVER".
19986 2017-02-15 Jakub Jelinek <jakub@redhat.com>
19989 * real.c (real_from_integer): Call real_convert even for decimal.
19991 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
19994 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
19996 2017-02-14 Andrew Pinski <apinski@cavium.com>
19998 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
19999 cores and change the partno/implementer to be correct.
20000 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
20001 the 'B" as the implementer.
20002 * config/aarch64/aarch64-tune.md: Regenerate.
20004 2017-02-14 Carl Love <cel@us.ibm.com>
20006 * config/rs6000/rs6000.c: Add case statement entry to make the
20007 xvcvuxdsp built-in argument unsigned.
20008 * config/rs6000/vsx.md: Fix the source and return operand types so they
20009 match the instruction definitions from the ISA document. Fix typo
20010 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
20013 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
20016 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
20017 member early_clobber_alts.
20018 * lra-lives.c (reg_early_clobber_p): New.
20019 (process_bb_lives): Use it.
20020 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
20021 (debug_operand_data): Initialize early_clobber_alts.
20022 (setup_operand_alternative): Set up early_clobber_alts.
20023 (collect_non_operand_hard_regs): Ditto. Pass early clobber
20024 alternatives to new_insn_reg.
20025 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
20027 (lra_update_insn_regno_info): Pass the new arg.
20029 2017-02-14 Jakub Jelinek <jakub@redhat.com>
20031 PR middle-end/79505
20032 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
20033 (new_oacc_loop_raw): Don't clear already cleared fields.
20036 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
20037 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
20038 _mm512_prefetch_i64gather_ps): New inline functions and macros.
20040 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
20043 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
20045 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
20048 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
20049 the extra instruction to the right place to store 128-bit constant
20052 2017-02-14 Martin Sebor <msebor@redhat.com>
20054 PR middle-end/79448
20055 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
20056 warning for strings of unknown length.
20058 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
20060 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
20062 2017-02-14 Jeff Law <law@redhat.com>
20065 * ira-costs.c (scan_one_insn): Initialize register move costs
20066 for pseudos seen in USE/CLOBBER insns.
20068 PR tree-optimization/79095
20069 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
20070 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
20071 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
20072 if the operands are known to be not equal, then the resulting range
20074 (intersect_ranges): If the new range is ~[0,0] and the old range is
20075 wide, then prefer ~[0,0].
20076 * tree-vrp.c (overflow_comparison_p_1): New function.
20077 (overflow_comparison_p): New function.
20078 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
20079 if NAME is used in an overflow test.
20080 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
20081 overflow check that can be expressed as an equality test, then adjust
20082 ops to be that equality test.
20084 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20086 * config/s390/s390-builtin-types.def: Remove flags argument.
20087 * config/s390/s390.c (s390_init_builtins): Likewise.
20089 2017-02-14 Martin Liska <mliska@suse.cz>
20091 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
20092 vector. Fix trailing white spaces.
20094 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
20096 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
20099 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20101 PR rtl-optimization/68664
20102 * config/arm/arm.c (arm_sched_can_speculate_insn):
20103 New function. Declare prototype.
20104 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
20106 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20108 PR rtl-optimization/68664
20109 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
20111 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
20113 2017-02-14 Amit Pawar <amit.pawar@amd.com>
20115 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
20116 max skip bytes for function, loop and jump.
20118 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20120 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
20121 ABS_EXPR for gimple dump.
20123 2017-02-14 Jakub Jelinek <jakub@redhat.com>
20126 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
20128 PR tree-optimization/79408
20129 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
20130 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
20131 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
20132 also if rhs1 is INTEGER_CST.
20134 2017-02-14 Richard Biener <rguenther@suse.de>
20136 PR middle-end/79432
20137 * tree-into-ssa.c (insert_phi_nodes): When the function can
20138 have abnormal edges rewrite SSA names with broken use-def
20139 dominance out of SSA and register them for PHI insertion.
20141 2017-02-13 Martin Sebor <msebor@redhat.com>
20143 PR middle-end/79496
20144 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
20145 clearing info.nowrite flag when snprintf size argument is a range.
20147 2017-02-13 Jakub Jelinek <jakub@redhat.com>
20149 * cprop.c (cprop_jump): Add missing space in string literal.
20150 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
20151 (get_constraint_for_component_ref): Likewise.
20152 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
20153 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
20154 * lra-constraints.c (process_alt_operands): Likewise.
20155 * ipa-inline.c (inline_small_functions): Likewise.
20156 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
20157 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
20158 * trans-mem.c (diagnose_tm_1_op): Likewise.
20159 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
20160 (grid_parallel_clauses_gridifiable): Likewise.
20162 * config/nvptx/mkoffload.c (process): Add space in between
20165 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
20166 "MOD4_SSE_REGS" and "ALL_REGS".
20168 * spellcheck.c (test_data): Add , in between "foo" and "food".
20170 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20173 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
20174 boundary crossing check and subsequent code generation agree.
20176 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20178 * config/aarch64/aarch64.c (has_memory_op): Delete.
20179 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
20182 2017-02-13 Jakub Jelinek <jakub@redhat.com>
20184 PR rtl-optimization/79388
20185 PR rtl-optimization/79450
20186 * combine.c (distribute_notes): When removing TEM_INSN for which
20187 corresponding dest has last value recorded, invalidate that last
20190 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20192 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
20193 of explicit '@'. Add missing assembly comment marker on branch costs
20196 2017-02-13 Nathan Sidwell <nathan@acm.org>
20198 * gengtype-lex.l (<in_struct>): Add '/'.
20200 2017-02-13 Martin Liska <mliska@suse.cz>
20203 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
20205 2017-02-13 Richard Biener <rguenther@suse.de>
20207 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
20209 * configure: Re-generate.
20210 * config.in: Likewise.
20211 * graphite-dependences.c: Simplify as if
20212 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
20213 * graphite-isl-ast-to-gimple.c: Likewise.
20214 * graphite-optimize-isl.c: Likewise.
20215 * graphite-poly.c: Likewise.
20216 * graphite-sese-to-poly.c: Likewise.
20217 * graphite.h: Likewise.
20218 * toplev.c: Include isl/version.h and use isl_version () for
20219 printing the ISL version.
20220 * doc/install.texi: Update ISL requirement.
20222 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
20224 * doc/standards.texi (Standards): Update reference to
20227 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
20229 * doc/extend.texi (Named Address Spaces): sourceware.org now
20231 * doc/install.texi (Binaries): Ditto.
20234 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
20236 * doc/cpp.texi: Replace "stringify"/"stringification" with C
20237 standard terminology "stringize"/"stringizing" throughout.
20238 * doc/cppinternals.texi: Likewise.
20240 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
20242 * doc/extend.texi: Fix some spelling mistakes and typos.
20243 * doc/invoke.texi: Likewise.
20245 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
20248 * params.def (inline-min-speedup) Change from 10 to 8.
20250 2017-02-11 Jakub Jelinek <jakub@redhat.com>
20252 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
20255 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
20258 * ipa-inline-analysis.c (get_minimal_bb): New function.
20259 (record_modified): Use it.
20260 (remap_edge_change_prob): Handle also ancestor functions.
20262 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
20264 * doc/contrib.texi (Contributors): Remove broken link into
20265 the Mauve CVS repository.
20267 2017-02-11 Jakub Jelinek <jakub@redhat.com>
20269 PR middle-end/79454
20270 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
20271 result computation whenever lhs doesn't have vector mode, not
20272 just when it has BLKmode.
20274 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
20276 * doc/makefile.texi (profiledbootstrap): Refer to the
20277 installation instructions only in textual form.
20279 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20282 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
20284 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
20286 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
20287 (Specific): Update mingw-w64 reference.
20289 (Specific): Remove broken link to Renesas RX processor.
20291 2017-02-10 Richard Biener <rguenther@suse.de>
20293 * toplev.c (process_options): Do not mention obsolete graphite
20294 options when printing sorry message about missing graphite support.
20295 Mention -floop-nest-optimize.
20297 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
20299 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
20300 (vtst_p16): Likewise.
20301 (vtstq_p8): Likewise.
20302 (vtstq_p16): Likewise.
20304 (vtstq_p64): Likewise.
20305 * config/arm/arm_neon.h (vgetq_lane_p64): New.
20306 (vset_lane_p64): New.
20307 (vsetq_lane_p64): New.
20309 2017-02-10 Jakub Jelinek <jakub@redhat.com>
20311 PR tree-optimization/79411
20312 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
20313 stmt operands are SSA_NAMEs used in abnormal phis.
20314 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
20317 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
20320 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
20323 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
20325 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
20327 2017-02-09 Jakub Jelinek <jakub@redhat.com>
20329 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
20333 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
20334 not arbitrary TREE_CONSTANT.
20337 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
20338 "omp declare target link" attribute unless is_global_var.
20339 * omp-offload.c (find_link_var_op): Likewise.
20341 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
20342 Chung-Lin Tang <cltang@codesourcery.com>
20344 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
20346 (gimplify_adjust_omp_clauses): Don't delete TILE.
20347 (gimplify_omp_for): Deal with TILE.
20348 * internal-fn.c (expand_GOACC_TILE): New function.
20349 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
20351 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
20352 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
20354 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
20355 avoid DIV for outermost collapse var.
20356 (expand_oacc_for): Insert tile element loop as needed. Adjust.
20357 Remove out of date comments, fix whitespace.
20358 * omp-general.c (omp_extract_for_data): Deal with tiling.
20359 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
20360 adjust OLF_DIM_BASE value.
20361 (struct omp_for_data): Add tiling field.
20362 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
20363 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
20364 for auto loops. Remove default auto determining, moved to
20365 oacc_loop_fixed_partitions.
20366 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
20367 stmts, add e_mask field.
20368 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
20369 (oacc_thread_numbers): Use oacc_dim_call.
20370 (oacc_xform_tile): New.
20371 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
20372 (finish_oacc_loop): Adjust for ifns vector.
20373 (oacc_loop_discover_walk): Append loop abstraction sites to list,
20374 add case for GOACC_TILE fns.
20375 (oacc_loop_xform_loop): Delete.
20376 (oacc_loop_process): Iterate over call list directly, and add
20377 handling for GOACC_TILE fns.
20378 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
20380 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
20381 vector partitioning to outer loops. Assign 2 partitions to loops
20382 when available. Add TILE handling.
20383 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
20384 (execite_oacc_device_lower): Process GOACC_TILE fns,
20385 ignore unknown specs.
20386 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
20387 * tree.c (omp_clause_num_ops): Adjust TILE ops.
20388 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
20390 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
20392 * configure.ac (ACX_BUGURL): Update.
20393 * configure: Regenerate.
20395 2017-02-09 Richard Biener <rguenther@suse.de>
20397 PR tree-optimization/69823
20398 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
20399 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
20401 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
20403 * config/arc/arc-c.def: Add __NPS400__ definition.
20404 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
20405 (TARGET_NPS400): Define.
20407 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
20409 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
20411 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
20412 pointer, arch_info.
20413 (arc_cpu_types): Fill the arch_info field with a pointer into the
20414 arc_arch_types table.
20415 (arc_selected_cpu): Declare.
20416 * config/arc/arc.c (arc_selected_cpu): Make global.
20417 (arc_selected_arch): Delete.
20418 (arc_base_cpu): Delete.
20419 (arc_override_options): Remove references to deleted variables,
20420 update access to arch information.
20421 (ARC_OPT): Update access to arch information.
20422 (ARC_OPTX): Likewise.
20423 * config/arc/arc.h (arc_base_cpu): Remove declaration.
20424 (TARGET_ARC600): Update access to arch information.
20425 (TARGET_ARC601): Likewise.
20426 (TARGET_ARC700): Likewise.
20427 (TARGET_EM): Likewise.
20428 (TARGET_HS): Likewise.
20429 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
20432 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
20435 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
20436 condition/operands for integer GE/LE/GEU/LEU operations.
20438 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
20440 PR translation/79397
20441 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
20444 2017-02-08 Martin Jambor <mjambor@suse.cz>
20447 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
20448 whether allocation happened.
20449 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
20450 nothing was allocated.
20452 2017-02-08 Jakub Jelinek <jakub@redhat.com>
20454 PR tree-optimization/79408
20455 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
20456 constant, but SSA_NAME with a known integer range, use the minimum
20457 of that range instead of op1 to determine if modulo can be replaced
20458 with its first operand.
20460 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20462 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
20464 2017-02-08 Richard Biener <rguenther@suse.de>
20466 PR tree-optimization/71824
20467 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
20468 Check all loops contained in the merged region.
20470 2017-02-07 Andrew Pinski <apinski@cavium.com>
20472 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
20474 2017-02-07 Andrew Pinski <apinski@cavium.com>
20476 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
20477 (thunderxt88): Likewise.
20478 (thunderxt81): Disable LSE and change v8.1 to v8.
20479 (thunderxt83): Likewise.
20481 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20482 Richard Biener <rguenther@suse.de>
20484 PR middle-end/79399
20485 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
20486 type from int to size_t.
20487 * ira-costs.c (struct_costs_size): Change type from int to size_t.
20489 2017-02-07 Jakub Jelinek <jakub@redhat.com>
20491 PR rtl-optimization/79386
20492 * cprop.c (bypass_conditional_jumps): Initialize
20493 bypass_last_basic_block already before splitting bbs after
20494 unconditional traps...
20495 (bypass_conditional_jumps): ... rather than here.
20498 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
20499 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
20500 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
20501 fix -masm=intel patterns.
20503 2017-02-07 Richard Biener <rguenther@suse.de>
20505 PR tree-optimization/79256
20506 PR middle-end/79278
20507 * builtins.c (get_object_alignment_2): Use min_align_of_type
20508 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
20509 and ADJUST_FIELD_ALIGN.
20511 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
20513 * doc/tm.texi: Regenerate.
20514 * stor-layout.c (layout_decl): Adjust.
20515 (update_alignment_for_field): Likewise.
20516 (place_field): Likewise.
20517 (min_align_of_type): Likewise.
20518 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
20519 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
20520 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
20521 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
20522 * config/frv/frv.c (frv_adjust_field_align): Likewise.
20523 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
20524 * config/i386/i386.c (x86_field_alignment): Likewise.
20525 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
20526 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
20527 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
20528 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
20529 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
20530 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
20534 2017-01-30 Richard Biener <rguenther@suse.de>
20536 PR tree-optimization/79256
20537 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
20538 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
20541 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
20543 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
20544 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
20545 builtins to SImode and emit a zero-extend, if necessary.
20547 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20549 * docs/invoke.texi (RISC-V Options): Alphabetize.
20551 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20553 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
20556 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
20558 * config/riscv/riscv.c: New file.
20559 * common/config/riscv/riscv-common.c: Likewise.
20560 * config.gcc: Likewise.
20561 * config/riscv/constraints.md: Likewise.
20562 * config/riscv/elf.h: Likewise.
20563 * config/riscv/generic.md: Likewise.
20564 * config/riscv/linux.h: Likewise.
20565 * config/riscv/multilib-generator: Likewise.
20566 * config/riscv/peephole.md: Likewise.
20567 * config/riscv/pic.md: Likewise.
20568 * config/riscv/predicates.md: Likewise.
20569 * config/riscv/riscv-builtins.c: Likewise.
20570 * config/riscv/riscv-c.c: Likewise.
20571 * config/riscv/riscv-ftypes.def: Likewise.
20572 * config/riscv/riscv-modes.def: Likewise.
20573 * config/riscv/riscv-opts.h: Likewise.
20574 * config/riscv/riscv-protos.h: Likewise.
20575 * config/riscv/riscv.h: Likewise.
20576 * config/riscv/riscv.md: Likewise.
20577 * config/riscv/riscv.opt: Likewise.
20578 * config/riscv/sync.md: Likewise.
20579 * config/riscv/t-elf-multilib: Likewise.
20580 * config/riscv/t-linux: Likewise.
20581 * config/riscv/t-linux-multilib: Likewise.
20582 * config/riscv/t-riscv: Likewise.
20583 * configure.ac: Likewise.
20584 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
20585 Waterman as RISC-V maintainers.
20586 * doc/install.texi: Add RISC-V entries.
20587 * doc/invoke.texi: Add RISC-V options section.
20588 * doc/md.texi: Add RISC-V constraints section.
20589 * configure: Regenerated.
20591 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
20594 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
20595 false values to be constant vectors with all 0 or all 1 bits set.
20596 (vcondu<mode><mode>): Likewise.
20597 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
20599 (fpmask_comparison_operator): Update comment.
20600 (vecint_comparison_operator): New predicate.
20601 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
20602 vector conditionals when the true and false values are constant
20603 vectors with all 0 bits or all 1 bits set.
20605 2017-02-06 Martin Sebor <msebor@redhat.com>
20607 PR tree-optimization/79376
20608 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
20610 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
20612 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
20613 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
20614 to simplify split condition.
20616 2017-02-06 Jakub Jelinek <jakub@redhat.com>
20618 * omp-expand.c (oxpand_omp_atomic_fetch_op,
20619 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
20622 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
20624 PR rtl-optimization/68664
20625 * target.def (can_speculate_insn): New hook.
20626 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
20627 * doc/tm.texi: Regenerate.
20628 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
20629 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
20630 (rs6000_sched_can_speculate_insn): New function.
20632 2017-02-06 Jakub Jelinek <jakub@redhat.com>
20634 PR tree-optimization/79284
20635 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
20636 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
20637 vectorizable_mask_load_store, vectorizable_operation,
20638 vect_is_simple_cond, get_same_sized_vectype): Use it instead
20639 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
20640 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
20641 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
20642 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
20643 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
20644 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
20645 is_gimple_assign (stmt). Replace another such test with
20646 is_gimple_assign (stmt).
20648 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
20651 * config/avr/avr.c (rtl-iter.h): Include it.
20652 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
20653 (avr_legitimate_combined_insn): ...and implementation.
20655 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
20657 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
20658 * config/s390/s390.c (s390_const_operand_ok)
20659 (s390_canonicalize_comparison, s390_extract_part)
20660 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
20661 (s390_contiguous_bitmask_p, s390_rtx_costs)
20662 (legitimize_pic_address): Likewise.
20663 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
20664 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
20665 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
20666 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
20667 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
20669 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
20671 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
20672 REGNO($0) == REGNO($1).
20674 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20676 * config/s390/linux.h(SIZE_TYPE): Add comment.
20678 2017-02-06 Julian Brown <julian@codesourcery.com>
20679 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20680 Virendra Pathak <virendra.pathak@broadcom.com>
20682 * config/aarch64/aarch64-cores.def: Change the scheduler
20684 * config/aarch64/aarch64.md: Include thunderx2t99.md.
20685 * config/aarch64/thunderx2t99.md: New file.
20687 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20689 * doc/standards.texi (Go Language): Update link to language
20692 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20694 * tree-eh.c (lower_resx): Sanitize profile.
20695 (cleanup_empty_eh_move_lp): Likewise.
20697 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
20700 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
20702 * cfgloopmanip.h (loop_version): Update prototype.
20703 * modulo-sched.c (sms_schedule): Update call of loop_version.
20704 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
20705 * tree-parloops.c (gen_parallel_loop): Likewise.
20706 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
20707 * tree-ssa-loop-split.c (split_loop): Likewise.
20708 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
20709 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
20711 2017-02-05 Martin Liska <mliska@suse.cz>
20714 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
20716 (print_operand_address): Initialize a struct to zero.
20718 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20720 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
20721 garbage collector only in textual form.
20723 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
20725 * doc/extend.texi (x86 specific memory model extensions for
20726 transactional memory): Simplify a phrase.
20728 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
20731 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
20732 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
20733 (atomic_storedi_1): Likewise.
20735 2017-02-04 Jakub Jelinek <jakub@redhat.com>
20737 PR tree-optimization/79338
20738 * tree-parloops.c (gather_scalar_reductions): Don't call
20739 vect_analyze_loop_form for loop->inner before destroying loop's
20742 2017-02-03 Martin Sebor <msebor@redhat.com>
20744 PR tree-optimization/79327
20745 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
20746 when precision has resulted in leading zeros.
20747 (format_integer): Adjust the likely counter to assume an unknown
20748 argument that may be zero is non-zero.
20750 2017-02-03 Jason Merrill <jason@redhat.com>
20753 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
20754 avoid copying non-taken branch.
20756 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20758 PR tree-optimization/79340
20759 * tree-vect-loop.c (vectorizable_reduction): Release
20760 vec_defs elements after safe_splicing them into other vectors.
20763 PR tree-optimization/79327
20764 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
20765 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
20767 (format_integer): Use wide_int_to_tree instead of build_int_cst
20768 + to_?hwi. If argmin is NULL, just set argmin and argmax to
20769 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
20770 of shortest and longest sequence.
20772 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
20774 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
20775 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
20777 2017-02-03 Walter Lee <walt@tilera.com>
20780 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
20781 after initial stackframe link reg save.
20782 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
20784 2017-02-03 Jakub Jelinek <jakub@redhat.com>
20787 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
20788 wu for stxssp alternative.
20790 2017-02-03 Martin Sebor <msebor@redhat.com>
20792 PR tree-optimization/79352
20793 * gimple-fold.c (get_range_strlen): Add argument.
20794 (get_range_strlen): Change return type to bool.
20795 (get_maxval_strlen): Pass in a dummy argument.
20796 * gimple-fold.h (get_range_strlen): Change return type to bool.
20797 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
20798 * tree.h (array_at_struct_end_p): Add argument.
20799 * tree.c (array_at_struct_end_p): Handle it.
20801 2017-02-03 Martin Liska <mliska@suse.cz>
20804 * multiple_target.c (create_dispatcher_calls): Redirect edge
20805 from a caller of a dispatcher.
20806 (expand_target_clones): Make the clones local.
20807 (ipa_target_clone): Do both target clones and resolvers.
20808 (ipa_dispatcher_calls): Remove the pass.
20809 (pass_dispatcher_calls::gate): Likewise.
20810 (make_pass_dispatcher_calls): Likewise.
20811 * passes.def (pass_target_clone): Put as very first IPA early
20814 2017-02-03 Martin Liska <mliska@suse.cz>
20816 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
20817 in case of a function with ifunc attribute.
20819 2017-02-03 Martin Liska <mliska@suse.cz>
20821 * cgraph.c (cgraph_node::dump): Dump function version info.
20822 * symtab.c (symtab_node::dump_base): Add missing new line.
20824 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20826 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
20827 (ifcombine_ifandif): Use it.
20829 2017-02-03 Martin Liska <mliska@suse.cz>
20831 * doc/invoke.texi: Document default value for
20832 use-after-scope-direct-emission-threshold.
20834 2017-02-03 Martin Liska <mliska@suse.cz>
20836 PR tree-optimization/79339
20837 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
20838 (format_floating): Likewise.
20840 2017-02-03 Martin Liska <mliska@suse.cz>
20843 * ipa-prop.c (ipa_node_params_t::insert): Remove current
20845 (ipa_node_params_t::remove): Likewise.
20846 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
20847 initialization from removed ipa_node_params_t::insert.
20848 (ipa_node_params::~ipa_node_params): Move from removed
20849 ipa_node_params_t::release.
20850 * symbol-summary.h (symbol_summary::m_released): New member.
20851 Do not release a summary twice. Do not allow to call finalizer
20852 for types of a summary that live in GGC memory.
20854 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20856 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
20859 2017-02-02 Martin Sebor <msebor@redhat.com>
20861 PR middle-end/79275
20862 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
20863 (format_string): Tighten up the range of output for non-constant
20864 strings and correct the expected range for wide non-constant strings.
20866 2017-02-02 Martin Sebor <msebor@redhat.com>
20868 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
20870 PR middle-end/32003
20871 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
20873 (-fdump-tree-@var): Add to index and document how to come up
20874 with pass-specific option and dump file names.
20875 (-fdump-passes): Clarify where to look for output.
20877 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
20879 PR middle-end/77445
20880 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
20881 statistics of the analyzed path; allow threading for speed when
20882 any of BBs along the path are optimized for speed.
20884 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
20886 PR middle-end/78468
20887 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
20888 settings of the virtual registers.
20891 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
20893 * explow.c (get_dynamic_stack_size): Take known alignment of stack
20894 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
20897 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20899 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
20900 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
20902 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20904 * config/s390/s390.md: Add missing comments with the expanded
20906 * config/s390/vector.md: Likewise.
20907 * config/s390/vx-builtins.md: Likewise.
20909 2017-02-02 Jakub Jelinek <jakub@redhat.com>
20912 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
20913 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
20914 conditions on a single line.
20916 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20918 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
20919 __S390_VX__ to __VX__.
20921 2017-02-01 Andrew Pinski <apinski@cavium.com>
20923 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
20924 stmt_info to record_stmt_cost.
20925 (vect_get_known_peeling_cost): Pass stmt_info if known to
20927 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
20928 cpu_vector_cost field into
20929 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
20930 field into vec_int_stmt_cost and vec_fp_stmt_cost.
20931 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
20932 splitting of scalar_stmt_cost and vec_stmt_cost.
20933 (thunderx_vector_cost): Likewise.
20934 (cortexa57_vector_cost): LIkewise.
20935 (exynosm1_vector_cost): Likewise.
20936 (xgene1_vector_cost): Likewise.
20937 (thunderx2t99_vector_cost): Improve after the splitting of the two
20939 (aarch64_builtin_vectorization_cost): Update for the splitting of
20940 scalar_stmt_cost and vec_stmt_cost.
20942 2017-02-01 Torvald Riegel <triegel@redhat.com>
20943 Richard Henderson <rth@redhat.com>
20945 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
20946 conditional on existance of a fast atomic load.
20947 * optabs-query.c (can_atomic_load_p): New function.
20948 * optabs-query.h (can_atomic_load_p): Declare it.
20949 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
20950 no fast atomic load is available for the particular size of access.
20951 (expand_atomic_compare_and_swap): Likewise.
20952 (expand_atomic_load): Likewise.
20953 (expand_atomic_store): Likewise.
20954 (expand_atomic_fetch_op): Likewise.
20955 * testsuite/lib/target-supports.exp
20956 (check_effective_target_sync_int_128): Remove x86 because it provides
20957 no fast atomic load.
20958 (check_effective_target_sync_int_128_runtime): Likewise.
20960 2017-02-01 Richard Biener <rguenther@suse.de>
20962 * graphite.c: Include tree-vectorizer.h for find_loop_location.
20963 (graphite_transform_loops): Provide opt-info for optimized nests.
20964 * tree-parloop.c (parallelize_loops): Provide opt-info for
20965 parallelized loops.
20967 2017-02-01 Richard Biener <rguenther@suse.de>
20969 PR middle-end/79315
20970 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
20971 was not set before.
20973 2017-02-01 Richard Biener <rguenther@suse.de>
20975 PR tree-optimization/71824
20976 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
20977 Verify the loops are valid in the merged SESE region.
20978 (scop_detection::can_represent_loop_1): Check analyzing the
20979 evolution of the number of iterations in the region succeeds.
20981 2017-01-31 Ian Lance Taylor <iant@golang.org>
20983 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
20984 REG_ARGS_SIZE note to 32-bit push insns and call insn.
20986 2017-01-31 David Malcolm <dmalcolm@redhat.com>
20988 PR preprocessor/79210
20989 * input.c (get_substring_ranges_for_loc): Replace line_width
20990 assertion with error-handling.
20992 2017-01-31 Richard Biener <rguenther@suse.de>
20994 PR tree-optimization/77318
20995 * graphite-sese-to-poly.c (extract_affine): Fix assert.
20996 (create_pw_aff_from_tree): Take loop parameter.
20997 (add_condition_to_pbb): Pass loop of the condition to
20998 create_pw_aff_from_tree.
21000 2017-01-31 Jakub Jelinek <jakub@redhat.com>
21002 * config/s390/s390.c (s390_asan_shadow_offset): New function.
21003 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
21005 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
21009 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
21011 (convert_int_to_float128): Likewise.
21012 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
21013 (convert_int_to_float128): Likewise.
21014 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
21015 (UNSPEC_IEEE128_CONVERT): Likewise.
21016 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
21017 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
21018 Use local variables for IBM extended format.
21019 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
21020 (fix_trunc<mode>si2_fprs): Likewise.
21021 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
21022 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
21023 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
21024 to know that we can now have integers of all sizes in vector
21026 (fix<uns>_<mode>di2_hw): Likewise.
21027 (float<uns>_<mode>si2_hw): Likewise.
21028 (fix_<mode>si2_hw): Likewise.
21029 (fixuns_<mode>si2_hw): Likewise.
21030 (float<uns>_<mode>di2_hw): Likewise.
21031 (float_<mode>di2_hw): Likewise.
21032 (float_<mode>si2_hw): Likewise.
21033 (floatuns_<mode>di2_hw): Likewise.
21034 (floatuns_<mode>si2_hw): Likewise.
21035 (xscvqp<su>wz_<mode>): Delete, no longer used.
21036 (xscvqp<su>dz_<mode>): Likewise.
21037 (xscv<su>dqp_<mode>): Likewise.
21038 (ieee128_mfvsrd_64bit): Likewise.
21039 (ieee128_mfvsrd_32bit): Likewise.
21040 (ieee128_mfvsrwz): Likewise.
21041 (ieee128_mtvsrw): Likewise.
21042 (ieee128_mtvsrd_64bit): Likewise.
21043 (ieee128_mtvsrd_32bit): Likewise.
21045 2017-01-31 Martin Liska <mliska@suse.cz>
21048 * ipa-prop.c (ipa_free_all_node_params): Call release method
21049 instead of ~sumbol_summary to not to trigger double times
21052 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
21054 PR tree-optimization/71691
21055 * bitmap.h (class auto_bitmap): New.
21056 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
21057 is_maybe_undefined instead of ssa_undefined_value_p.
21059 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21061 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
21062 __S390_ARCH_LEVEL__ to __ARCH__.
21064 2017-01-31 Jakub Jelinek <jakub@redhat.com>
21066 PR tree-optimization/79267
21067 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
21068 if should_remove_lhs_p is true.
21070 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
21073 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
21074 (add_alignment_attribute): New.
21075 (base_type_die): Add alignment attribute.
21076 (subrange_type_die): Likewise.
21077 (modified_type_die): Likewise.
21078 (gen_array_type_die): Likewise.
21079 (gen_descr_array_type_die: Likewise.
21080 (gen_enumeration_type_die): Likewise.
21081 (gen_subprogram_die): Likewise.
21082 (gen_variable_die): Likewise.
21083 (gen_field_die): Likewise.
21084 (gen_ptr_to_mbr_type_die): Likewise.
21085 (gen_struct_or_union_type_die): Likewise.
21086 (gen_subroutine_type_die): Likewise.
21087 (gen_typedef_die): Likewise.
21088 (base_type_cmp): Compare alignment attribute.
21090 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
21093 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
21094 (setb_unsigned) New pattern for setb with CCUNS.
21095 * config/rs6000/rs6000.c (expand_block_compare): Use a different
21096 subfc./subfe sequence to avoid overflow problems. Generate a
21097 shorter sequence with cmpld/setb for power9.
21098 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
21099 for generating subfc. instruction.
21100 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
21101 now uses this instruction.
21103 2017-01-30 Ian Lance Taylor <iant@google.com>
21106 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
21107 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
21109 2017-01-30 Martin Sebor <msebor@redhat.com>
21111 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
21112 Move constant to the right of a relational operator.
21113 (get_mpfr_format_length, format_character, format_string): Ditto.
21114 (should_warn_p, maybe_warn): Same.
21116 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
21118 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
21121 * asan.c (get_translation_unit_decl): Remove function.
21122 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
21124 2017-01-30 Martin Liska <mliska@suse.cz>
21126 PR gcov-profile/79259
21127 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
21128 -fprofile-generate.
21130 2017-01-30 Martin Liska <mliska@suse.cz>
21133 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
21134 Initialize variables with NULL value.
21136 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
21139 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
21141 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
21143 2017-01-30 Richard Biener <rguenther@suse.de>
21145 PR tree-optimization/79276
21146 * tree-vrp.c (process_assert_insertions): Properly adjust common
21147 when removing a duplicate.
21149 * gcc.dg/torture/pr79276.c: New testcase.
21151 2017-01-30 Richard Biener <rguenther@suse.de>
21153 PR tree-optimization/79256
21154 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
21155 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
21157 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
21159 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
21162 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
21163 ("*r<noxa>sbg_<mode>_sll_bitmask")
21164 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
21165 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
21166 Use contiguous_bitmask_nowrap_operand.
21168 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21171 * config/rs6000/altivec.h (vec_xl): Revise #define.
21172 (vec_xst): Likewise.
21174 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
21176 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
21178 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
21180 PR rtl-optimization/79194
21181 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
21182 traps before call to bypass_conditional_jumps.
21184 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
21186 PR tree-optimization/71374
21187 * lra-constraints.c (check_conflict_input_operands): New.
21188 (match_reload): Use it.
21190 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
21193 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
21194 account to calculate conflict_set.
21196 2017-01-27 Bin Cheng <bin.cheng@arm.com>
21198 PR rtl-optimization/78559
21199 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
21200 other_insn in combine.
21202 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21204 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
21205 uint16_type_node for BT_UINT16.
21207 2017-01-27 David Malcolm <dmalcolm@redhat.com>
21209 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
21210 "RTL Tests" to menu.
21211 (GIMPLE Tests): New node.
21212 (RTL Tests): New node.
21214 2017-01-27 Richard Biener <rguenther@suse.de>
21216 PR tree-optimization/79245
21217 * tree-loop-distribution.c (distribute_loop): Apply cost
21218 modeling also to detected patterns.
21220 2017-01-27 Richard Biener <rguenther@suse.de>
21222 PR tree-optimization/71433
21223 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
21224 (compare_assert_loc): New function.
21225 (process_assert_insertions): Sort and optimize assert locations
21226 to remove duplicates and push down identical assertions on
21227 edges to their destination block.
21229 2017-01-27 Richard Biener <rguenther@suse.de>
21231 PR tree-optimization/79244
21232 * tree-vrp.c (remove_range_assertions): Forcefully propagate
21233 out SSA names even if abnormal.
21235 2017-01-27 Jakub Jelinek <jakub@redhat.com>
21237 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
21238 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
21239 instead of MPFR_RNDN.
21241 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
21244 * arm.c (arm_option_override): Don't call build_target_option_node
21245 until after doing all option overrides.
21246 (arm_valid_target_attribute_tree): Likewise.
21248 2017-01-27 Martin Liska <mliska@suse.cz>
21250 * doc/invoke.texi (-fprofile-arcs): Document profiling support
21251 for {cd}tors and C++ {cd}tors.
21253 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
21255 * config/s390/s390.md ("*setmem_long_and")
21256 ("*setmem_long_and_31z"): Use zero_extend instead of and.
21258 2017-01-26 Martin Sebor <msebor@redhat.com>
21260 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
21263 2017-01-26 Martin Sebor <msebor@redhat.com>
21265 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
21266 HAVE_DFmode before using XFmode or DFmode.
21267 (parse_directive): Avoid using the z length modifier to avoid
21268 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
21270 PR middle-end/78703
21271 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
21272 to accept adjustment as an array.
21273 (get_int_range): New function.
21274 (struct directive): Make width and prec arrays.
21275 (directive::set_width, directive::set_precision): Call get_int_range.
21276 (format_integer, format_floating): Handle width and precision ranges.
21277 (format_string, parse_directive): Same.
21279 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21282 * dwarf2out.c (generate_skeleton_bottom_up): For children with
21283 comdat_type_p set, just clone them, but keep the children in the
21287 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
21288 which have direct callers with -fvar-tracking-assignments enabled
21290 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
21291 inside of type units.
21293 2017-01-26 Martin Sebor <msebor@redhat.com>
21295 PR middle-end/78703
21296 * gimple-ssa-sprintf.c (struct result_range): Add likely and
21298 (struct format_result): Replace number_chars, number_chars_min,
21299 and number_chars_max with a single member of struct result_range.
21301 (format_result::operator+=): Adjust.
21302 (struct fmtresult): Remove bounded. Handle likely and unlikely
21304 (fmtresult::adjust_for_width_or_precision): New function.
21305 (fmtresult:type_max_digits): New function.
21306 (bytes_remaining): Handle likely and unlikely counters.
21307 (min_bytes_remaining): Remove.
21308 (format_percent): Simplify.
21309 (format_integer, format_floating): Set likely and unlikely counters.
21310 (get_string_length, format_character, format_string): Same.
21311 (format_plain, should_warn_p): New function.
21312 (maybe_warn): Call should_warn_p. Update diagnostic messages
21313 and handle those for all directives, including plain strings.
21314 (format_directive): Handle likely and unlikely counters.
21315 Remove unnecessary quoting from diagnostics. Add an informational
21317 (add_bytes): Remove.
21318 (pass_sprintf_length::compute_format_length): Simplify.
21319 (try_substitute_return_value): Handle likely and unlikely counters.
21321 2017-01-26 Carl Love <cel@us.ibm.com>
21323 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
21324 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
21326 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
21329 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
21330 endianess for subregs into account.
21331 * lra-constraints.c (lra_constraints): Do risky transformations
21332 always on the first iteration.
21333 * lra-lives.c (check_pseudos_live_through_calls): Add arg
21334 last_call_used_reg_set.
21335 (process_bb_lives): Define and use last_call_used_reg_set.
21336 * lra.c (lra): Always continue after lra_constraints on the first
21339 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
21341 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
21343 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
21345 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21347 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
21348 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
21349 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
21350 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
21351 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
21352 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
21353 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
21354 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
21355 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
21357 2017-01-26 Marek Polacek <polacek@redhat.com>
21360 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
21361 for the third operand.
21363 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21365 PR middle-end/79236
21366 * omp-low.c (struct omp_context): Add simt_stmt field.
21367 (scan_omp_for): Return omp_context *.
21368 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
21369 context to the _simt_ SIMD stmt.
21370 (lower_omp_for): For combined SIMD with sibling _simt_
21371 SIMD, make sure to use the same decls in _looptemp_
21372 clauses as in the sibling.
21374 2017-01-26 David Sherwood <david.sherwood@arm.com>
21376 PR middle-end/79212
21377 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
21380 2017-01-26 Jakub Jelinek <jakub@redhat.com>
21383 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
21384 emit fld b; fld a; if possible.
21386 * brig-builtins.def: Update copyright years.
21387 * config/arm/arm_acle_builtins.def: Update copyright years.
21389 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
21392 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
21393 constraint instead of o for the stxsd instruction.
21395 2017-01-25 Carl Love <cel@us.ibm.com>
21397 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
21398 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
21400 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
21402 * doc/invoke.texi (C++ Dialect Options): Fix typo.
21404 2017-01-25 Richard Biener <rguenther@suse.de>
21406 PR tree-optimization/69264
21407 * target.def (vector_alignment_reachable): Improve documentation.
21408 * doc/tm.texi: Regenerate.
21409 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
21411 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
21412 earlier changes with respect to TYPE_USER_ALIGN.
21413 (vector_alignment_reachable_p): Likewise. Improve dumping.
21415 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21418 * config/arm/arm.md (xordi3): Force constant operand into a register
21421 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21423 * doc/invoke.texi (-fstore-merging): Correct default optimization
21424 levels at which it is enabled.
21425 (-O): Move -fstore-merging from list to...
21428 2017-01-25 Richard Biener <rguenther@suse.de>
21431 * omp-expand.c: Include debug.h.
21432 (expand_omp_taskreg): Make sure to generate early debug before
21433 outlining anything from a function.
21434 (expand_omp_target): Likewise.
21435 (grid_expand_target_grid_body): Likewise.
21437 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
21440 * asan.c (get_translation_unit_decl): New function.
21441 (asan_add_global): Extract modules file name from globals
21442 TRANSLATION_UNIT_DECL name.
21444 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
21447 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
21448 for long calls with APCS frame and VFP.
21450 2017-01-24 David Malcolm <dmalcolm@redhat.com>
21452 * cfg.c (original_copy_tables_initialized_p): New function.
21453 * cfg.h (original_copy_tables_initialized_p): New decl.
21454 * cfgrtl.c (relink_block_chain): Guard the call to
21455 free_original_copy_tables with a call to
21456 original_copy_tables_initialized_p.
21457 * cgraph.h (symtab_node::native_rtl_p): New decl.
21458 * cgraphunit.c (symtab_node::native_rtl_p): New function.
21459 (symtab_node::needed_p): Don't assert for early assembly output
21460 for __RTL functions.
21461 (cgraph_node::finalize_function): Set "force_output" for __RTL
21463 (cgraph_node::analyze): Bail out early for __RTL functions.
21464 (analyze_functions): Update assertion to support __RTL functions.
21465 (cgraph_node::expand): Bail out early for __RTL functions.
21466 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
21468 * function.h (struct function): Update comment for field
21470 * gimple-expr.c: Include "tree-pass.h".
21471 (gimple_has_body_p): Return false for __RTL functions.
21472 * Makefile.in (OBJS): Add run-rtl-passes.o.
21473 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
21475 (gcc::pass_manager::get_clean_slate): New accessor.
21476 * passes.c: Include "insn-addr.h".
21477 (should_skip_pass_p): Add logging. Update logic for running
21478 "expand" to be compatible with both __GIMPLE and __RTL. Guard
21479 property-provider override so it is only done for gimple passes.
21481 (skip_pass): New function.
21482 (execute_one_pass): Call skip_pass when skipping passes.
21483 * read-md.c (md_reader::read_char): Support filtering
21484 the input to a subset of line numbers.
21485 (md_reader::md_reader): Initialize fields
21486 m_first_line and m_last_line.
21487 (md_reader::read_file_fragment): New function.
21488 * read-md.h (md_reader::read_file_fragment): New decl.
21489 (md_reader::m_first_line): New field.
21490 (md_reader::m_last_line): New field.
21491 * read-rtl-function.c (function_reader::create_function): Only
21492 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
21493 curr_properties. Set DECL_INITIAL to a dummy block.
21494 (read_rtl_function_body_from_file_range): New function.
21495 * read-rtl-function.h (read_rtl_function_body_from_file_range):
21497 * run-rtl-passes.c: New file.
21498 * run-rtl-passes.h: New file.
21500 2017-01-24 Jeff Law <law@redhat.com>
21502 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
21505 2017-01-24 Bin Cheng <bin.cheng@arm.com>
21507 PR tree-optimization/79159
21508 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
21509 (record_nonwrapping_iv): Improve boundary using above function if no
21510 value range information.
21512 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
21513 Martin Jambor <mjambor@suse.cz>
21515 * brig-builtins.def: New file.
21516 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
21517 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
21518 (DEF_HSAIL_SAT_BUILTIN): Likewise.
21519 (DEF_HSAIL_INTR_BUILTIN): Likewise.
21520 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
21521 * builtin-types.def (BT_INT8): New.
21522 (BT_INT16): Likewise.
21523 (BT_UINT8): Likewise.
21524 (BT_UINT16): Likewise.
21525 (BT_FN_ULONG): Likewise.
21526 (BT_FN_UINT_INT): Likewise.
21527 (BT_FN_UINT_ULONG): Likewise.
21528 (BT_FN_UINT_LONG): Likewise.
21529 (BT_FN_UINT_PTR): Likewise.
21530 (BT_FN_ULONG_PTR): Likewise.
21531 (BT_FN_INT8_FLOAT): Likewise.
21532 (BT_FN_INT16_FLOAT): Likewise.
21533 (BT_FN_UINT32_FLOAT): Likewise.
21534 (BT_FN_UINT16_FLOAT): Likewise.
21535 (BT_FN_UINT8_FLOAT): Likewise.
21536 (BT_FN_UINT64_FLOAT): Likewise.
21537 (BT_FN_UINT16_UINT32): Likewise.
21538 (BT_FN_UINT32_UINT16): Likewise.
21539 (BT_FN_UINT16_UINT16_UINT16): Likewise.
21540 (BT_FN_INT_PTR_INT): Likewise.
21541 (BT_FN_UINT_PTR_UINT): Likewise.
21542 (BT_FN_LONG_PTR_LONG): Likewise.
21543 (BT_FN_ULONG_PTR_ULONG): Likewise.
21544 (BT_FN_VOID_UINT64_UINT64): Likewise.
21545 (BT_FN_UINT8_UINT8_UINT8): Likewise.
21546 (BT_FN_INT8_INT8_INT8): Likewise.
21547 (BT_FN_INT16_INT16_INT16): Likewise.
21548 (BT_FN_INT_INT_INT): Likewise.
21549 (BT_FN_UINT_FLOAT_UINT): Likewise.
21550 (BT_FN_FLOAT_UINT_UINT): Likewise.
21551 (BT_FN_ULONG_UINT_UINT): Likewise.
21552 (BT_FN_ULONG_UINT_PTR): Likewise.
21553 (BT_FN_ULONG_ULONG_ULONG): Likewise.
21554 (BT_FN_UINT_UINT_UINT): Likewise.
21555 (BT_FN_VOID_UINT_PTR): Likewise.
21556 (BT_FN_UINT_UINT_PTR: Likewise.
21557 (BT_FN_UINT32_UINT64_PTR): Likewise.
21558 (BT_FN_INT_INT_UINT_UINT): Likewise.
21559 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
21560 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
21561 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
21562 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
21563 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
21564 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
21565 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
21566 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
21567 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
21568 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
21569 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
21570 * doc/frontends.texi: List BRIG FE.
21571 * doc/install.texi (Testing): Add BRIG tesring requirements.
21572 * doc/invoke.texi (Overall Options): Mention BRIG.
21573 * doc/standards.texi (Standards): Doucment BRIG HSA version.
21575 2017-01-24 Richard Biener <rguenther@suse.de>
21577 PR translation/79208
21578 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
21580 2017-01-24 Martin Jambor <mjambor@suse.cz>
21583 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
21584 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
21585 and known_contexts.
21587 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
21589 PR middle-end/79123
21590 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
21591 casts from signed to unsigned really don't have a range.
21593 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
21595 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
21596 GMP_RNDx for compatiblity.
21598 2017-01-24 Martin Liska <mliska@suse.cz>
21601 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
21602 that would prevent us to call alloca with -1 as argument.
21604 2017-01-24 Jakub Jelinek <jakub@redhat.com>
21606 * dwarf2out.c (output_compilation_unit_header, output_file_names):
21607 Avoid -Wformat-security warning.
21609 2017-01-23 Andrew Pinski <apinski@cavium.com>
21611 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
21614 2017-01-23 Martin Sebor <msebor@redhat.com>
21616 PR middle-end/78703
21617 * gimple-ssa-sprintf.c (warn_level): New global.
21618 (format_integer): Use it here and throughout the rest of the file.
21619 Use the same switch to compute sign as base.
21620 (maybe_warn): New function.
21621 (format_directive): Factor out warnings into maybe_warn.
21622 Add debugging output. Use warn_level.
21623 (add_bytes): Use warn_level.
21624 (pass_sprintf_length::compute_format_length): Add debugging output.
21625 (try_substitute_return_value): Same.
21626 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
21628 PR middle-end/78703
21629 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
21630 (struct fmtresult, format_integer, format_floating): Adjust.
21631 (fmtresult::fmtresult): Set max correctly in two argument ctor.
21632 (get_string_length, format_string,format_directive): Same.
21633 (pass_sprintf_length::compute_format_length): Same.
21634 (try_substitute_return_value): Simplify slightly.
21636 PR middle-end/78703
21637 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
21638 (fmtresult::operator+=): Outlined.
21639 (struct fmtresult): Add ctors.
21640 (struct conversion_spec): Rename...
21641 (struct directive): ...to this. Add and remove data members.
21642 (directive::set_width, directive::set_precision): New functions.
21643 (format_percent): Use fmtresult ctor.
21644 (get_width_and_precision): Remove.
21645 (format_integer): Make naming changes. Avoid computing width and
21647 (format_floating): Same. Adjust indentation.
21648 (format_character, format_none): New functions.
21649 (format_string): Moved character handling to format_character.
21650 (format_directive): Remove arguments, change return type.
21651 (parse_directive): New function.
21652 (pass_sprintf_length::compute_format_length): Move directive
21653 parsing to parse_directive.
21655 2017-01-23 Jakub Jelinek <jakub@redhat.com>
21657 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
21658 (assign_assembler_name_if_needed): ... this.
21659 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
21660 (assign_assembler_name_if_needed): ... this.
21661 (free_lang_data_in_cgraph): Adjust callers.
21662 * cgraphunit.c (cgraph_node::analyze): Likewise.
21663 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
21666 2017-01-23 Richard Biener <rguenther@suse.de>
21668 PR tree-optimization/79088
21669 PR tree-optimization/79188
21670 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
21671 resetting loop bounds after last path deletion. Reset loop
21672 bounds of the target loop, make code match the comments.
21673 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
21674 Make sure loops need no fixups.
21676 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
21678 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
21679 exponent support with double type for first argument.
21680 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
21681 type returned by __builtin_vec_extract_sig,
21682 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
21683 functions from "vector int" to "vector unsigned int" or from
21684 "vector long long int" to "vector unsigned long long int".
21685 Changed type returned by __builtin_vec_extract_exp,
21686 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
21687 functions from "vector int" to "vector unsigned int" or from
21688 "vector long long int" to "vector unsigned long long int".
21689 Changed return type of __builtin_vec_test_data_class,
21690 __builtin_vec_test_data_class_sp, and
21691 __builtin_vec_test_data_class_dp from "vector int" to
21692 "vector bool int" or from "vector long long int" to "vector bool
21693 long long int" and changed second argument type from "unsigned
21694 int" to "int". Added new overloaded function forms "vector float
21695 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
21696 "vector float __builtin_vec_insert_exp_sp (vector float, vector
21697 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
21698 double, vector unsigned long long int)" and "vector double
21699 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
21700 long int)". Changed return type of
21701 __builtin_scalar_test_data_class and
21702 __builtin_scalar_test_data_class_sp and
21703 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
21704 int" and changed second argument from "unsigned int" to "int".
21705 Changed type returned by __builtin_scalar_test_neg,
21706 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
21707 from "int" to "bool int". Added new overloaded function form
21708 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
21709 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
21710 exponent double-precision with floating point first argument.
21711 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
21712 documentation of scalar_test_data_class, scalar_test_neg,
21713 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
21714 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
21715 vec_test_data_class built-in functions to reflect refinements in
21716 their type signatures.
21718 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
21720 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
21722 (aarch64_elf_asm_destructor): Likewise.
21724 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
21726 PR rtl-optimization/78634
21727 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
21728 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
21729 * ifcvt.c (noce_try_cmove): Add missing cost check.
21731 PR rtl-optimization/71724
21732 * combine.c (if_then_else_cond): Look for situations where it is
21733 beneficial to undo the work of one of the recursive calls.
21735 2017-01-23 Bin Cheng <bin.cheng@arm.com>
21737 PR tree-optimization/70754
21738 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
21739 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
21740 combined stmt before it if not NULL.
21741 (combine_chains): Process refs reversely and compute dominance point
21744 2017-01-23 Martin Liska <mliska@suse.cz>
21746 PR tree-optimization/79196
21747 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
21748 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
21750 (strlen_optimize_stmt): Call the renamed function.
21752 2017-01-23 Michael Matz <matz@suse.de>
21754 PR tree-optimization/78384
21755 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
21757 2017-01-23 Richard Biener <rguenther@suse.de>
21759 PR tree-optimization/79186
21760 * tree-vrp.c (register_new_assert_for): Make sure we've seen
21761 both incoming edges before moving an assert.
21763 2017-01-23 Martin Jambor <mjambor@suse.cz>
21765 * ipa-prop.c (load_from_param_1): Removed.
21766 (load_from_unmodified_param): Bits from load_from_param_1 put back
21768 (load_from_param): Removed.
21769 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
21770 with stmt. Reverted back to use of load_from_unmodified_param.
21772 2017-01-23 Martin Jambor <mjambor@suse.cz>
21775 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
21776 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
21777 field a pointer to garbage collected vector, mark lattices and
21778 ipcp_orig_node with GTY((skip)).
21779 (ipa_get_param_count): Adjust to descriptors being a pointer.
21780 (ipa_get_param): Likewise.
21781 (ipa_get_type): Likewise.
21782 (ipa_get_param_move_cost): Likewise.
21783 (ipa_set_param_used): Likewise.
21784 (ipa_get_controlled_uses): Likewise.
21785 (ipa_set_controlled_uses): Likewise.
21786 (ipa_is_param_used): Likewise.
21787 (ipa_node_params_t): Move into garbage collector. New methods insert
21789 (ipa_node_params_sum): Annotate wth GTY(()).
21790 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
21792 (ipa_load_from_parm_agg): Adjust declaration.
21793 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
21794 * ipa-profile.c (ipa_profile): Likewise.
21795 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
21796 (ipa_populate_param_decls): Make descriptors parameter garbage
21798 (ipa_dump_param): Adjust to descriptors being a pointer.
21799 (ipa_alloc_node_params): Likewise.
21800 (ipa_initialize_node_params): Likewise.
21801 (load_from_param_1): Make descriptors parameter garbage collected.
21802 (load_from_unmodified_param): Likewise.
21803 (load_from_param): Likewise.
21804 (ipa_load_from_parm_agg): Likewise.
21805 (ipa_node_params::~ipa_node_params): Removed.
21806 (ipa_free_all_node_params): Remove call to delete operator.
21807 (ipa_node_params_t::insert): New.
21808 (ipa_node_params_t::remove): Likewise.
21809 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
21810 copy known_csts and known_contexts vectors.
21811 (ipa_read_node_info): Adjust to descriptors being a pointer.
21812 (ipcp_modif_dom_walker): Make m_descriptors field garbage
21814 (ipcp_transform_function): Make descriptors variable garbage
21817 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
21819 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
21820 * config/i386/avx512dqintrin.h: Ditto.
21821 * config/i386/avx512fintrin.h: Ditto.
21822 * config/i386/i386.c: Handle new builtins.
21823 * config/i386/i386-builtin.def: Add new builtins.
21824 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
21825 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
21827 2017-01-23 Jakub Jelinek <jakub@redhat.com>
21828 Martin Liska <mliska@suse.cz>
21830 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
21831 * asan.c (asan_expand_poison_ifn): Support stores and use
21832 appropriate ASAN report function.
21833 * internal-fn.c (expand_ASAN_POISON_USE): New function.
21834 * internal-fn.def (ASAN_POISON_USE): Declare.
21835 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
21836 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
21837 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
21838 ASAN_POISON calls w/o LHS.
21839 * tree-ssa.c (execute_update_addresses_taken): Create clobber
21840 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
21841 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
21842 * gimplify.c (asan_poison_variables): Add attribute
21843 use_after_scope_memory to variables that really needs to live
21845 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
21846 having the attribute.
21848 2017-01-23 Martin Liska <mliska@suse.cz>
21850 * asan.c (create_asan_shadow_var): New function.
21851 (asan_expand_poison_ifn): Likewise.
21852 * asan.h (asan_expand_poison_ifn): New declaration.
21853 * internal-fn.c (expand_ASAN_POISON): Likewise.
21854 * internal-fn.def (ASAN_POISON): New builtin.
21855 * sanopt.c (pass_sanopt::execute): Expand
21856 asan_expand_poison_ifn.
21857 * tree-inline.c (copy_decl_for_dup_finish): Make function
21859 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
21860 * tree-ssa.c (is_asan_mark_p): New function.
21861 (execute_update_addresses_taken): Rewrite local variables
21862 (identified just by use-after-scope as addressable) into SSA.
21864 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21866 * doc/install.texi (Specific): opensource.apple.com uses https
21867 now. Remove trailing slash.
21869 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
21871 * README.Portability: Remove note on an Irix compatibility issue.
21873 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
21875 * gcov.c (INCLUDE_ALGORITHM): Define.
21876 (INCLUDE_VECTOR): Define.
21877 No longer include <vector> and <algorithm> directly.
21879 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
21881 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
21883 * doc/invoke.texi (Code Gen Options): Ditto.
21885 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
21888 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
21890 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
21892 rtl-optimization/79125
21893 * cprop.c (local_cprop_pass): Handle cases where we make an
21894 unconditional trap.
21896 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
21900 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
21901 read from, for big endian.
21903 2017-01-20 Jiong Wang <jiong.wang@arm.com>
21905 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
21906 register pauth builtins for LP64 only.
21908 2017-01-20 Marek Polacek <polacek@redhat.com>
21911 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
21914 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21916 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
21918 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
21919 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
21920 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
21922 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21925 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
21926 in big-endian mode when they are not a single duplicated value.
21928 2017-01-20 Richard Biener <rguenther@suse.de>
21930 * BASE-VER: Bump to 7.0.1.
21932 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
21934 * omp-low.c (omplow_simd_context): New struct. Use it...
21935 (lower_rec_simd_input_clauses): ...here and...
21936 (lower_rec_input_clauses): ...here to hold common data. Adjust all
21937 references to idx, lane, max_vf, is_simt.
21939 2017-01-20 Graham Markall <graham.markall@embecosm.com>
21941 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
21944 2017-01-20 Martin Jambor <mjambor@suse.cz>
21946 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
21947 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
21949 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
21950 (GTFILES): Rename hsa.c to hsa-common.c.
21951 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
21952 * hsa-dump.c: Likewise.
21953 * hsa-gen.c: Likewise.
21954 * hsa-regalloc.c: Likewise.
21955 * ipa-hsa.c: Likewise.
21956 * omp-expand.c: Likewise.
21957 * omp-low.c: Likewise.
21958 * toplev.c: Likewise.
21960 2017-01-20 Marek Polacek <polacek@redhat.com>
21963 * doc/invoke.texi: Document -Wduplicated-branches.
21964 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
21965 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
21966 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
21967 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
21968 return 0 only when not OEP_LEXICOGRAPHIC.
21969 (fold_build_cleanup_point_expr): Use the expression
21970 location when building CLEANUP_POINT_EXPR.
21971 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
21972 * tree.c (add_expr): Handle error_mark_node.
21974 2017-01-20 Martin Liska <mliska@suse.cz>
21977 * tree-profile.c (init_ic_make_global_vars): Do not call
21979 (gimple_init_gcov_profiler): Likewise.
21981 2017-01-20 Martin Liska <mliska@suse.cz>
21984 * cgraph.h (maybe_create_reference): Remove argument and
21986 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
21988 * ipa-cp.c (create_specialized_node): Likewise.
21989 * symtab.c (symtab_node::maybe_create_reference): Handle
21990 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
21992 2017-01-20 Martin Liska <mliska@suse.cz>
21994 * read-rtl-function.c (function_reader::create_function): Use
21995 build_decl instread of build_decl_stat.
21997 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
21999 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
22000 * config/i386/avx512dqintrin.h: Ditto.
22001 * config/i386/avx512fintrin.h: Ditto.
22002 * config/i386/i386-builtin-types.def: Add new types.
22003 * config/i386/i386.c: Handle new types.
22004 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
22005 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
22006 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
22007 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
22008 (__builtin_ia32_kshiftridi): New.
22009 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
22011 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
22015 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
22016 define to rs6000_init_stack_protect_guard.
22017 (rs6000_init_stack_protect_guard): New function.
22019 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
22020 Yunqiang Su <yunqiang.su@imgtec.com>
22022 * config.gcc (supported_defaults): Add madd4.
22023 (with_madd4): Add validation.
22024 (all_defaults): Add madd4.
22025 * config/mips/mips.opt (mmadd4): New option.
22026 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
22028 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
22030 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
22031 (ISA_HAS_FUSED_MADD4): Likewise.
22032 * doc/invoke.texi (-mmadd4): Document the new option.
22033 * doc/install.texi (--with-madd4): Document the new option.
22035 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22037 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
22038 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
22039 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
22040 (aarch64_init_pauth_hint_builtins): New.
22041 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
22042 (aarch64_expand_builtin): Expand new builtins.
22044 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22046 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
22047 * combine-stack-adj.c (no_unhandled_cfa): Handle
22048 REG_CFA_TOGGLE_RA_MANGLE.
22049 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
22050 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
22051 info for return address signing.
22052 (aarch64_expand_epilogue): Likewise.
22054 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22056 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
22057 * config/aarch64/aarch64-protos.h
22058 (aarch64_return_address_signing_enabled): New declaration.
22059 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
22061 (aarch64_expand_prologue): Sign return address before it's pushed onto
22063 (aarch64_expand_epilogue): Authenticate return address fetched from
22065 (aarch64_override_options): Sanity check for ILP32 and ISA level.
22066 (aarch64_attributes): New function attributes for "sign-return-address".
22067 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
22068 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
22069 ("*do_return"): Generate combined instructions according to key index.
22070 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
22071 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
22073 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
22074 * config/aarch64/aarch64.opt (msign-return-address=): New.
22075 * doc/extend.texi (AArch64 Function Attributes): Documents
22076 "sign-return-address=".
22077 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
22079 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
22081 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
22082 overall option summary.
22084 2017-01-19 Jiong Wang <jiong.wang@arm.com>
22086 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
22087 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
22088 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
22089 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
22091 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
22093 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
22094 -mpower9-minmax by default for -mcpu=power9.
22095 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
22096 128-bit floating point.
22098 2017-01-20 Alan Modra <amodra@gmail.com>
22100 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
22101 optimizing for size.
22103 2017-01-20 Alan Modra <amodra@gmail.com>
22106 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
22107 for strcmp and strncmp from corresponding builtin decl.
22109 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
22111 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
22112 instead of i386/rtems-64.h.
22113 * config/i386/rtems-64.h: Remove.
22115 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
22119 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
22121 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
22123 2017-01-19 Tamar Christina <tamar.christina@arm.com>
22125 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
22126 Change int to HOST_WIDE_INT.
22127 * config/aarch64/aarch64-protos.h
22128 (aarch64_simd_gen_const_vector_dup): Likewise.
22129 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
22131 2017-01-19 David Malcolm <dmalcolm@redhat.com>
22133 * langhooks-def.h (lhd_type_for_size): New decl.
22134 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
22135 * langhooks.c (lhd_type_for_size): New function, taken from
22138 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
22140 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
22141 define_bypass for CR latency.
22142 (power9-cracked-alu): Update bypass latency and remove power9-branch.
22143 (power9-alu2): Add define_bypass for CR latency.
22145 (power9-mul): Update insn latency.
22146 (power9-mul-compare): Update insn latency, bypass latency and remove
22149 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22151 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
22153 * config/aarch64/aarch64.md
22154 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
22155 aarch64_nopcrelative_literal_loads.
22156 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
22158 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
22160 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
22161 TARGET_LOONGSON_3A.
22162 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
22164 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
22167 * config.gcc (supported_defaults): Add lxc1-sxc1.
22168 (with_lxc1_sxc1): Add validation.
22169 (all_defaults): Add lxc1-sxc1.
22170 * config/mips/mips.opt (mlxc1-sxc1): New option.
22171 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
22173 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
22174 __mips_no_lxc1_sxc1.
22175 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
22176 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
22177 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
22179 2017-01-19 Richard Biener <rguenther@suse.de>
22181 PR tree-optimization/72488
22182 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
22183 sure to restore SSA info.
22184 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
22186 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
22188 PR rtl-optimization/79121
22189 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
22190 of the inner type when shifting an extended value.
22192 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
22195 * symtab.c (symtab_node::equal_address_to): Fix comparing of
22196 interposable aliases.
22198 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
22201 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
22202 Use the evmergelohi instruction.
22203 (mov_si<mode>_e500_subreg4_2_le): Likewise.
22204 (mov_sitf_e500_subreg8_2_be): Likewise.
22205 (mov_sitf_e500_subreg12_2_le): Likewise.
22206 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
22207 (mov_si<mode>_e500_subreg4_2_be): Likewise.
22208 (mov_sitf_e500_subreg8_2_le): Likewise.
22209 (mov_sitf_e500_subreg12_2_be): Likewise.
22211 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22213 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
22214 attribute from vecsimple to vecperm.
22215 (altivec_vbpermq2): Likewise.
22217 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22220 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
22222 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22223 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
22224 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
22225 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
22226 case where N arg is SIZE_MAX.
22227 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
22228 (cmpstrsi): Add pattern.
22230 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22232 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22233 __builtin_vec_revb builtins.
22234 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
22235 built-in functions to support generation of the ISA 3.0 XXBR<x>
22236 vector byte reverse instructions.
22237 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
22238 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
22239 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
22240 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
22241 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
22242 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
22243 (P9V_BUILTIN_VEC_REVB): Likewise.
22244 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
22245 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
22246 (p9_xxbrq_v16qi): Likewise.
22247 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
22248 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
22249 (p9_xxbrh_v8hi): Likewise.
22250 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
22251 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
22252 vec_revb built-in functions.
22254 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
22256 PR rtl-optimization/78952
22257 * config/i386/i386.md (any_extract): New code iterator.
22258 (*insvqi_2): Use any_extract for source operand.
22259 (*insvqi_3): Use any_shiftrt for source operand.
22261 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
22263 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
22265 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
22267 2017-01-18 Matthias Klose <doko@ubuntu.com>
22269 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
22271 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22273 * config/rs6000/altivec.h (vec_bperm): Change #define.
22274 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
22275 (altivec_vbpermq2): New define_insn.
22276 (altivec_vbpermd): Likewise.
22277 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
22278 function interface.
22279 (VBPERMD): Likewise.
22280 (VBPERM): New polymorphic function interface.
22281 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
22282 Add entries for P9V_BUILTIN_VEC_VBPERM.
22283 * doc/extend.texi: Add interfaces for vec_bperm.
22285 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22287 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
22288 first letter of error messages.
22289 (s390_resolve_overloaded_builtin): Likewise.
22290 * config/s390/s390.c (s390_expand_builtin): Likewise.
22291 (s390_invalid_arg_for_unprototyped_fn): Likewise.
22292 (s390_valid_target_attribute_inner_p): Likewise.
22293 * config/s390/s390.md ("tabort"): Likewise.
22295 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
22297 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
22298 (ISA_AVOID_DIV_HILO): New macro.
22299 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
22300 (ISA_HAS_DDIV): Likewise.
22302 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22304 * doc/invoke.texi (fabi-version): Correct number of occurrences.
22306 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22308 * doc/invoke.texi (fabi-version): Spelling fix.
22310 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22313 * doc/invoke.texi (fabi-version): Mention mangling fix for
22316 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
22319 * doc/invoke.texi (fabi-version): Document discriminator mangling.
22321 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
22324 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
22325 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
22327 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
22329 (stack_protect_test): Ditto.
22330 * config/rs6000/rs6000.opt (mstack-protector-guard=,
22331 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
22333 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
22334 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
22335 -mstack-protector-guard-offset=.
22336 (RS/6000 and PowerPC Options): Ditto.
22338 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
22340 * config/i386/i386.h (MASK_CLASS_P): New define.
22341 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
22342 there are no registers from different register sets also when
22343 mask registers are used. Update function comment.
22344 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
22345 to (*k/*r) and (*k/*km) alternatives.
22347 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
22349 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
22350 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
22351 (EH_RETURN_HANDLER_RTX): New define.
22352 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
22353 Force frame pointer in EH return functions.
22354 (aarch64_expand_epilogue): Add barrier for eh_return.
22355 (aarch64_final_eh_return_addr): Remove.
22356 (aarch64_eh_return_handler_rtx): New function.
22357 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
22359 (aarch64_eh_return_handler_rtx): New prototype.
22361 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22363 * config/rs6000/altivec.h (vec_rlmi): New #define.
22364 (vec_vrlnm): Likewise.
22365 (vec_rlnm): Likewise.
22366 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
22367 (UNSPEC_VRLNM): Likewise.
22368 (VIlong): New mode iterator.
22369 (altivec_vrl<VI_char>mi): New define_insn.
22370 (altivec_vrl<VI_char>nm): Likewise.
22371 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
22373 (VRLDNM): Likewise.
22374 (RLNM): New polymorphic function entry.
22375 (VRLWMI): New monomorphic function entry.
22376 (VRLDMI): Likewise.
22377 (RLMI): New polymorphic function entry.
22378 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
22379 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
22380 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
22383 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22386 * dwarf2out.c (field_byte_offset): Restore the
22387 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
22388 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
22389 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
22392 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
22395 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
22397 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22400 * dwarf2out.c (add_data_member_location_attribute): For constant
22401 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
22402 instead of DW_AT_data_member_location, DW_AT_bit_offset and
22403 DW_AT_byte_size attributes.
22405 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
22407 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
22408 after forcing to constant memory when the code model is medium.
22410 2017-01-17 Julia Koval <julia.koval@intel.com>
22413 * config/i386/avx512fintrin.h
22414 (_mm512_i32gather_ps): Change __addr type to void const*.
22415 (_mm512_mask_i32gather_ps): Ditto.
22416 (_mm512_i32gather_pd): Ditto.
22417 (_mm512_mask_i32gather_pd): Ditto.
22418 (_mm512_i64gather_ps): Ditto.
22419 (_mm512_mask_i64gather_ps): Ditto.
22420 (_mm512_i64gather_pd): Ditto.
22421 (_mm512_mask_i64gather_pd): Ditto.
22422 (_mm512_i32gather_epi32): Ditto.
22423 (_mm512_mask_i32gather_epi32): Ditto.
22424 (_mm512_i32gather_epi64): Ditto.
22425 (_mm512_mask_i32gather_epi64): Ditto.
22426 (_mm512_i64gather_epi32): Ditto.
22427 (_mm512_mask_i64gather_epi32): Ditto.
22428 (_mm512_i64gather_epi64): Ditto.
22429 (_mm512_mask_i64gather_epi64): Ditto.
22430 (_mm512_i32scatter_ps): Change __addr type to void*.
22431 (_mm512_mask_i32scatter_ps): Ditto.
22432 (_mm512_i32scatter_pd): Ditto.
22433 (_mm512_mask_i32scatter_pd): Ditto.
22434 (_mm512_i64scatter_ps): Ditto.
22435 (_mm512_mask_i64scatter_ps): Ditto.
22436 (_mm512_i64scatter_pd): Ditto.
22437 (_mm512_mask_i64scatter_pd): Ditto.
22438 (_mm512_i32scatter_epi32): Ditto.
22439 (_mm512_mask_i32scatter_epi32): Ditto.
22440 (_mm512_i32scatter_epi64): Ditto.
22441 (_mm512_mask_i32scatter_epi64): Ditto.
22442 (_mm512_i64scatter_epi32): Ditto.
22443 (_mm512_mask_i64scatter_epi32): Ditto.
22444 (_mm512_i64scatter_epi64): Ditto.
22445 (_mm512_mask_i64scatter_epi64): Ditto.
22446 * config/i386/avx512pfintrin.h
22447 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
22448 (_mm512_mask_prefetch_i32gather_ps): Ditto.
22449 (_mm512_mask_prefetch_i64gather_pd): Ditto.
22450 (_mm512_mask_prefetch_i64gather_ps): Ditto.
22451 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
22452 (_mm512_prefetch_i32scatter_ps): Ditto.
22453 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
22454 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
22455 (_mm512_prefetch_i64scatter_pd): Ditto.
22456 (_mm512_prefetch_i64scatter_ps): Ditto.
22457 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
22458 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
22459 * config/i386/avx512vlintrin.h
22460 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
22461 (_mm_mmask_i32gather_ps): Ditto.
22462 (_mm256_mmask_i32gather_pd): Ditto.
22463 (_mm_mmask_i32gather_pd): Ditto.
22464 (_mm256_mmask_i64gather_ps): Ditto.
22465 (_mm_mmask_i64gather_ps): Ditto.
22466 (_mm256_mmask_i64gather_pd): Ditto.
22467 (_mm_mmask_i64gather_pd): Ditto.
22468 (_mm256_mmask_i32gather_epi32): Ditto.
22469 (_mm_mmask_i32gather_epi32): Ditto.
22470 (_mm256_mmask_i32gather_epi64): Ditto.
22471 (_mm_mmask_i32gather_epi64): Ditto.
22472 (_mm256_mmask_i64gather_epi32): Ditto.
22473 (_mm_mmask_i64gather_epi32): Ditto.
22474 (_mm256_mmask_i64gather_epi64): Ditto.
22475 (_mm_mmask_i64gather_epi64): Ditto.
22476 (_mm256_i32scatter_ps): Change __addr type to void*.
22477 (_mm256_mask_i32scatter_ps): Ditto.
22478 (_mm_i32scatter_ps): Ditto.
22479 (_mm_mask_i32scatter_ps): Ditto.
22480 (_mm256_i32scatter_pd): Ditto.
22481 (_mm256_mask_i32scatter_pd): Ditto.
22482 (_mm_i32scatter_pd): Ditto.
22483 (_mm_mask_i32scatter_pd): Ditto.
22484 (_mm256_i64scatter_ps): Ditto.
22485 (_mm256_mask_i64scatter_ps): Ditto.
22486 (_mm_i64scatter_ps): Ditto.
22487 (_mm_mask_i64scatter_ps): Ditto.
22488 (_mm256_i64scatter_pd): Ditto.
22489 (_mm256_mask_i64scatter_pd): Ditto.
22490 (_mm_i64scatter_pd): Ditto.
22491 (_mm_mask_i64scatter_pd): Ditto.
22492 (_mm256_i32scatter_epi32): Ditto.
22493 (_mm256_mask_i32scatter_epi32): Ditto.
22494 (_mm_i32scatter_epi32): Ditto.
22495 (_mm_mask_i32scatter_epi32): Ditto.
22496 (_mm256_i32scatter_epi64): Ditto.
22497 (_mm256_mask_i32scatter_epi64): Ditto.
22498 (_mm_i32scatter_epi64): Ditto.
22499 (_mm_mask_i32scatter_epi64): Ditto.
22500 (_mm256_i64scatter_epi32): Ditto.
22501 (_mm256_mask_i64scatter_epi32): Ditto.
22502 (_mm_i64scatter_epi32): Ditto.
22503 (_mm_mask_i64scatter_epi32): Ditto.
22504 (_mm256_i64scatter_epi64): Ditto.
22505 (_mm256_mask_i64scatter_epi64): Ditto.
22506 (_mm_i64scatter_epi64): Ditto.
22507 (_mm_mask_i64scatter_epi64): Ditto.
22508 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
22509 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
22510 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
22511 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
22512 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
22513 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
22514 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
22515 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
22516 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
22517 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
22518 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
22519 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
22520 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
22521 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
22522 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
22523 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
22524 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
22525 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
22526 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
22527 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
22528 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
22529 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
22530 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
22531 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
22532 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
22533 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
22534 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
22535 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
22536 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
22537 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
22538 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
22539 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
22540 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
22541 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
22542 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
22543 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
22544 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
22545 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
22546 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
22547 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
22548 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
22549 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
22550 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
22551 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
22552 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
22553 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
22554 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
22555 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
22556 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
22557 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
22558 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
22559 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
22560 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
22561 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
22562 definitions accordingly.
22564 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
22565 Kuan-Lin Chen <kuanlinchentw@gmail.com>
22568 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
22571 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
22574 * ira-conflicts.c (ira_build_conflicts): Update total conflict
22575 hard regs for inner regno.
22577 2017-01-17 Martin Liska <mliska@suse.cz>
22580 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
22581 assumption and add comment.
22583 2017-01-17 Nathan Sidwell <nathan@acm.org>
22585 * ipa-visibility.c (localize_node): New function, broken out of ...
22586 (function_and_variable_visibility): ... here. Call it.
22588 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
22590 PR middle-end/77445
22591 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
22592 correctly set frequency of oudgoing edge.
22593 (duplicate_thread_path): Fix profile updating.
22595 2017-01-17 Jakub Jelinek <jakub@redhat.com>
22598 * configure.ac: Add GCC_BASE_VER.
22599 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
22600 version from BASE-VER file.
22601 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
22602 (gcc.o): Depend on $(BASEVER).
22603 * common.opt (dumpfullversion): New option.
22604 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
22605 * doc/invoke.texi: Document -dumpfullversion.
22606 * doc/install.texi: Document --with-gcc-major-version-only.
22607 * configure: Regenerated.
22609 2017-01-17 Richard Biener <rguenther@suse.de>
22611 PR tree-optimization/71433
22612 * tree-vrp.c (register_new_assert_for): Merge same asserts
22613 on all incoming edges.
22614 (process_assert_insertions_for): Handle insertions at the
22617 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
22619 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
22620 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
22622 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
22625 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
22628 2017-01-17 Alan Modra <amodra@gmail.com>
22631 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
22632 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
22633 symbolic stack limit when pic.
22635 2017-01-16 Martin Sebor <msebor@redhat.com>
22637 PR tree-optimization/78608
22638 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
22640 2017-01-16 Jeff Law <law@redhat.com>
22643 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
22644 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
22645 for several include directories that may be relative to sysroot.
22646 * config/i386/x-mingw32 (gplus_includedir): Define.
22647 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
22648 (native_system_includedir): Likewise.
22649 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
22650 override if TARGET_SYSTEM_ROOT is defined.
22651 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
22653 PR tree-optimization/79090
22654 PR tree-optimization/33562
22655 PR tree-optimization/61912
22656 PR tree-optimization/77485
22657 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
22658 and computed trims into the dump file.
22660 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
22662 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
22664 2017-01-16 Jakub Jelinek <jakub@redhat.com>
22667 * gimplify.c (gimplify_init_constructor): If want_value and
22668 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
22672 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
22673 sequence. Formatting fixes.
22674 (doloop_optimize): Formatting fixes.
22677 * gcc.c (debug_level_greater_than_spec_func): New function.
22678 (static_spec_functions): Add debug-level-gt spec function.
22679 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
22681 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22682 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
22683 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
22684 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
22685 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
22686 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
22688 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
22690 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
22691 QImode fixups to general and mask registers only.
22693 2017-01-16 Carl Love <cel@us.ibm.com>
22695 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
22696 for built-in functions
22697 vector signed char vec_nabs (vector signed char)
22698 vector signed short vec_nabs (vector signed short)
22699 vector signed int vec_nabs (vector signed int)
22700 vector signed long long vec_nabs (vector signed long long)
22701 vector float vec_nabs (vector float)
22702 vector double vec_nabs (vector double)
22703 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
22705 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
22706 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
22707 * doc/extend.texi: Update the documentation file for the new built-in
22710 2017-01-16 Martin Sebor <msebor@redhat.com>
22712 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
22715 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22717 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
22718 UNSPEC_VSX__XXSPLTD to require special splat handling.
22720 2017-01-16 David Malcolm <dmalcolm@redhat.com>
22723 * system.h: Poison strndup.
22725 2017-01-16 Alan Modra <amodra@gmail.com>
22728 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
22731 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
22733 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
22735 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
22737 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
22738 call recog here. Assert that INSN_CODE (insn) is non-negative.
22740 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22743 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
22745 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
22746 in the currently scheduled RTL fragment.
22748 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
22750 PR rtl-optimization/78751
22751 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
22754 2017-01-14 Jeff Law <law@redhat.com>
22756 PR tree-optimization/79090
22757 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
22758 variable length stores.
22759 (compute_trims): Delete dead assignment to *trim_tail.
22760 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
22763 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
22765 PR rtl-optimization/78626
22766 PR rtl-optimization/78727
22767 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
22768 of a block, and split such blocks after everything else is finished.
22770 2017-01-14 Alan Modra <amodra@gmail.com>
22773 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
22774 target legitimate_combined_insn.
22775 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
22776 (rs6000_legitimate_combined_insn): New function.
22777 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
22779 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
22780 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
22781 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
22783 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
22785 * doc/frontends.texi (G++ and GCC): Remove references to Java.
22787 2017-01-13 Jeff Law <law@redhat.com>
22789 PR tree-optimization/33562
22790 PR tree-optimization/61912
22791 PR tree-optimization/77485
22792 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
22794 (delete_dead_assignment): Likewise.
22795 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
22796 statement to delete_dead_call and delete_dead_assignment.
22798 2017-01-13 David Malcolm <dmalcolm@redhat.com>
22801 * substring-locations.c (format_warning_va): Strengthen case 1 so
22802 that both endpoints of the substring must be within the format
22803 range for just the substring to be printed.
22805 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
22807 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
22808 * config/i386/i386.c (ix86_target_string): Add missing options
22809 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
22810 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
22811 flags_other and ix86_target_other to flags2_other. Display unknown
22813 (ix86_valid_target_attribute_inner_p): Add missing options and
22814 reorder options by implied ISAs, as in ix86_target_string.
22816 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22818 * hash-table.h (hash_table::too_empty_p): New function.
22819 (hash_table::expand): Use it.
22820 (hash_table::traverse): Likewise.
22821 (hash_table::empty_slot): Use sizeof (value_type) instead of
22822 sizeof (PTR) to convert bytes to elements. Shrink the table
22823 if the current size is excessive for the current number of
22826 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
22828 * ira-costs.c (record_reg_classes): Break from the inner loop
22829 early once alt_fail is known to be true. Update outer loop
22830 handling accordingly.
22832 2017-01-13 Jeff Law <law@redhat.com>
22834 * tree-ssa-dse.c (decrement_count): New function.
22835 (increment_start_addr, maybe_trim_memstar_call): Likewise.
22836 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
22837 when we know the partially dead statement is a mem* function.
22839 PR tree-optimization/61912
22840 PR tree-optimization/77485
22841 * tree-ssa-dse.c: Include expr.h.
22842 (maybe_trim_constructor_store): New function.
22843 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
22845 PR tree-optimization/33562
22846 PR tree-optimization/61912
22847 PR tree-optimization/77485
22848 * doc/invoke.texi: Document new dse-max-object-size param.
22849 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
22850 * tree-ssa-dse.c: Include params.h.
22851 (dse_store_status): New enum.
22852 (initialize_ao_ref_for_dse): New, partially extracted from
22854 (valid_ao_ref_for_dse, normalize_ref): New.
22855 (setup_live_bytes_from_ref, compute_trims): Likewise.
22856 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
22857 (maybe_trim_partially_dead_store): Likewise.
22858 (maybe_trim_complex_store): Likewise.
22859 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
22860 Track what bytes live from the original store. Return tri-state
22861 for dead, partially dead or live.
22862 (dse_dom_walker): Add constructor, destructor and new private members.
22863 (delete_dead_call, delete_dead_assignment): New extracted from
22865 (dse_optimize_stmt): Make a member of dse_dom_walker.
22866 Use initialize_ao_ref_for_dse.
22868 PR tree-optimization/33562
22869 PR tree-optimization/61912
22870 PR tree-optimization/77485
22871 * sbitmap.h (bitmap_count_bits): Prototype.
22872 (bitmap_clear_range, bitmap_set_range): Likewise.
22873 * sbitmap.c (bitmap_clear_range): New function.
22874 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
22876 2017-01-13 Martin Liska <mliska@suse.cz>
22879 * function.c (set_cfun): Add new argument force.
22880 * function.h (set_cfun): Likewise.
22881 * ipa-inline-transform.c (inline_call): Use the function when
22882 strict alising from is dropped for function we inline to.
22884 2017-01-13 Richard Biener <rguenther@suse.de>
22886 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
22887 for dumping GIMPLE INTEGER_CSTs.
22889 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22891 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
22892 to 201112L since C++17.
22894 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
22897 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
22898 if -fsanitize=kernel-address is present.
22900 2017-01-13 Richard Biener <rguenther@suse.de>
22902 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
22903 as _Literal ( type ) number in case usual suffixes do not
22904 preserve all information.
22906 2017-01-13 Richard Biener <rguenther@suse.de>
22908 PR tree-optimization/77283
22909 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
22910 and ssa-iterators.h.
22911 (is_feasible_trace): Implement a cost model based on joiner
22914 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
22917 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
22918 char or short to __float128/_Float128 directly.
22920 2017-01-12 Martin Sebor <msebor@redhat.com>
22922 to -Wformat-overflow.
22923 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
22924 (min_bytes_remaining): Same.
22925 (get_string_length): Same.
22926 (format_string): Same.
22927 (format_directive): Same.
22929 (pass_sprintf_length::handle_gimple_call): Same.
22931 2017-01-12 Jakub Jelinek <jakub@redhat.com>
22933 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
22934 info.nowrite calls with no lhs that can't throw. Return bool
22935 whether gsi_remove has been called or not.
22936 (pass_sprintf_length::handle_gimple_call): Return bool whether
22937 try_substitute_return_value called gsi_remove. Formatting fix.
22938 (pass_sprintf_length::execute): Don't use gsi_remove if
22939 handle_gimple_call returned true.
22942 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
22943 be removed due to side-effects, don't remove following barrier nor
22944 turn the successor edge into fallthru edge.
22946 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22949 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
22950 element-reversing loads and stores as not swappable.
22952 2017-01-12 Nathan Sidwell <nathan@acm.org>
22953 Nicolai Stange <nicstange@gmail.com>
22955 * combine.c (try_combine): Don't ignore result of overlap checking
22956 loop. Combine overlap & asm check into single loop.
22958 2017-01-12 Richard Biener <rguenther@suse.de>
22960 * tree-pretty-print.c (dump_generic_node): Provide -gimple
22961 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
22963 2017-01-12 Richard Biener <rguenther@suse.de>
22965 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
22966 and TS_TARGET_OPTION directly derive from TS_BASE.
22967 * tree-core.h (tree_optimization_option): Derive from tree_base.
22968 (tree_target_option): Likewise.
22970 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
22972 * config/i386/i386.c (memory_address_length): Increase len
22973 only when rip_relative_addr_p returns false.
22975 2017-01-11 Julia Koval <julia.koval@intel.com>
22977 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
22978 (OPTION_MASK_ISA_SGX_SET): New.
22979 (ix86_handle_option): Handle OPT_msgx.
22980 * config.gcc: Added sgxintrin.h.
22981 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
22982 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
22983 * config/i386/i386.c (ix86_target_string): Add -msgx.
22985 (ix86_option_override_internal): Handle new options.
22986 (ix86_valid_target_attribute_inner_p): Add sgx.
22987 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
22988 * config/i386/i386.opt: Add msgx.
22989 * config/i386/sgxintrin.h: New file.
22990 * config/i386/x86intrin.h: Add sgxintrin.h.
22992 2017-01-11 Jakub Jelinek <jakub@redhat.com>
22995 * fold-const.c (maybe_nonzero_address): Return 1 for function
22997 (tree_single_nonzero_warnv_p): Don't handle function local objects
23001 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
23004 2017-01-11 David Malcolm <dmalcolm@redhat.com>
23007 * opts.c: Include "spellcheck.h"
23008 (struct string_fragment): New struct.
23009 (struct edit_distance_traits<const string_fragment &>): New
23011 (get_closest_sanitizer_option): New function.
23012 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
23014 2017-01-11 Jakub Jelinek <jakub@redhat.com>
23016 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
23018 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
23019 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
23020 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
23021 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
23022 for initial die_offset if dwarf_split_debug_info.
23023 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
23024 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
23026 (output_skeleton_debug_sections): Formatting fix. Use
23027 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
23028 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
23030 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
23032 * config/arm/cortex-a53.md: Add bypasses for
23033 cortex_a53_r2f_cvt.
23034 (cortex_a53_r2f): Only use for transfers.
23035 (cortex_a53_f2r): Likewise.
23036 (cortex_a53_r2f_cvt): Add reservation for conversions.
23037 (cortex_a53_f2r_cvt): Likewise.
23039 2017-01-11 Tamar Christina <tamar.christina@arm.com>
23041 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
23042 to all inlined functions, change static to extern.
23044 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
23047 * config/arm/arm.c (legitimize_pic_address): Handle reference to
23049 (arm_assemble_integer): Likewise.
23051 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
23053 * config.gcc: Use new awk script to check CPU, FPU and architecture
23054 parameters for --with-... options.
23055 * config/arm/parsecpu.awk: New file
23056 * config/arm/arm-cpus.in: New file.
23057 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
23059 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
23061 * config/arm/t-arm: Update dependency rules.
23062 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
23063 of processing .def files.
23064 * config/arm/genopt.sh: Deleted.
23065 * config/arm/gentune.sh: Deleted.
23066 * config/arm/arm-cores.def: Deleted.
23067 * config/arm/arm-arches.def: Deleted.
23068 * config/arm/arm-fpus.def: Deleted.
23069 * config/arm/arm-tune.md: Regenerated.
23070 * config/arm/arm-tables.opt: Regenerated.
23071 * config/arm/arm-cpu.h: New generated file.
23072 * config/arm/arm-cpu-data.h: New generated file.
23073 * config/arm/arm-cpu-cdata.h: New generated file.
23075 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
23078 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
23080 (input_varpool_node): Unpack dynamically_initialized bit.
23082 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
23084 PR rtl-optimization/79032
23085 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
23086 the alignment of the adjusted memory reference against that of MODE,
23087 instead of the alignment of the original memory reference.
23089 2017-01-11 Martin Jambor <mjambor@suse.cz>
23091 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
23093 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
23094 decorated functions.
23096 2017-01-11 Richard Biener <rguenther@suse.de>
23098 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
23099 set range/nonnull info for PHI results. Do not set it on
23100 stmts marked for removal.
23102 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
23104 * expr.c (store_field): In the bitfield case, fetch the return value
23105 from the registers before applying a single big-endian adjustment.
23106 Always do a final load for a BLKmode value not larger than a word.
23108 2017-01-10 David Malcolm <dmalcolm@redhat.com>
23111 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
23112 that we correctly handle column numbers greater than
23113 LINE_MAP_MAX_COLUMN_NUMBER.
23115 2017-01-10 Martin Sebor <msebor@redhat.com>
23117 PR middle-end/78245
23118 * gimple-ssa-sprintf.c (get_destination_size): Call
23119 {init,fini}object_sizes.
23120 * tree-object-size.c (addr_object_size): Adjust.
23121 (pass_through_call): Adjust.
23122 (pass_object_sizes::execute): Adjust.
23123 * tree-object-size.h (fini_object_sizes): Declare.
23125 2017-01-10 Martin Sebor <msebor@redhat.com>
23127 PR tree-optimization/78775
23128 * builtins.c (get_size_range): Move...
23129 * calls.c: ...to here.
23130 (alloc_max_size): Accept zero argument.
23131 (operand_signed_p): Remove.
23132 (maybe_warn_alloc_args_overflow): Call get_size_range.
23133 * calls.h (get_size_range): Declare.
23135 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
23137 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
23138 from TI's devices.csv file as of September 2016.
23139 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
23141 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
23143 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
23144 * doc/invoke.texi: Likewise.
23145 * doc/md.texi: Likewise.
23146 * doc/objc.texi: Likewise.
23148 2017-01-10 Joshua Conner <joshconner@google.com>
23150 * config/arm/fuchsia-elf.h: New file.
23151 * config/fuchsia.h: New file.
23152 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
23153 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
23155 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
23157 2016-01-10 Richard Biener <rguenther@suse.de>
23159 PR tree-optimization/79034
23160 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
23161 Propagate out degenerate PHIs in the joiner.
23163 2017-01-10 Martin Liska <mliska@suse.cz>
23165 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
23166 (sort_congruence_classes_by_decl_uid): Likewise.
23167 (sort_congruence_class_groups_by_decl_uid): Likewise.
23168 (sem_item_optimizer::merge_classes): Sort class, groups in these
23169 classes and members in the groups by DECL_UID of declarations.
23170 This would make merge operations stable.
23172 2017-01-10 Martin Liska <mliska@suse.cz>
23174 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
23175 usage of m_classes_vec.
23176 (sem_item_optimizer::~sem_item_optimizer): Likewise.
23177 (sem_item_optimizer::get_group_by_hash): Likewise.
23178 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
23179 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
23180 (sem_item_optimizer::verify_classes): Likewise.
23181 (sem_item_optimizer::process_cong_reduction): Likewise.
23182 (sem_item_optimizer::dump_cong_classes): Likewise.
23183 (sem_item_optimizer::merge_classes): Likewise.
23184 * ipa-icf.h (congruence_class_hash): Rename from
23185 congruence_class_group_hash. Remove declaration of m_classes_vec.
23187 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
23189 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
23190 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
23191 * config.gcc: Add avx512vpopcntdqintrin.h.
23192 * config/i386/avx512vpopcntdqintrin.h: New.
23193 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
23194 * config/i386/i386-builtin-types.def: Add new types.
23195 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
23196 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
23197 __builtin_ia32_vpopcountq_v8di_mask): New.
23198 * config/i386/i386-c.c (ix86_target_macros_internal): Define
23199 __AVX512VPOPCNTDQ__.
23200 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
23201 (PTA_AVX512VPOPCNTDQ): Define.
23202 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
23203 TARGET_AVX512VPOPCNTDQ_P): Define.
23204 * config/i386/i386.opt: Add mavx512vpopcntdq.
23205 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
23206 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
23208 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23210 PR middle-end/77484
23211 * predict.def (PRED_CALL): Set to 67.
23213 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
23215 * expr.c (store_field): In the bitfield case, if the value comes from
23216 a function call and is of an aggregate type returned in registers, do
23217 not modify the field mode; extract the value in all cases if the mode
23218 is BLKmode and the size is not larger than a word.
23220 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
23223 * config/i386/cpuid.h: Fix undefined behavior.
23225 2017-01-04 Jeff Law <law@redhat.com>
23227 PR tree-optimization/79007
23228 PR tree-optimization/67955
23229 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
23230 conservative for pt.null when flag_non_call_exceptions is on.
23232 2017-01-09 Jakub Jelinek <jakub@redhat.com>
23234 PR translation/79019
23235 PR translation/79020
23236 * params.def (PARAM_INLINE_MIN_SPEEDUP,
23237 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
23238 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
23240 * config/avr/avr.opt (maccumulate-args): Likewise.
23241 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
23242 * common.opt (freport-bug): Likewise.
23243 * cif-code.def (CIF_FINAL_ERROR): Likewise.
23244 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
23245 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
23246 translatable string.
23247 * config/i386/i386.c (function_value_32): Likewise.
23248 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
23249 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
23251 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
23252 * common/config/msp430/msp430-common.c (msp430_handle_option):
23254 * symtab.c (symtab_node::verify_base): Likewise.
23255 * opts.c (set_debug_level): Likewise.
23256 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
23257 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
23258 missing whitespace to translatable strings.
23259 * config/avr/avr.md (bswapsi2): Fix typo in comment.
23260 * config/sh/superh.h: Likewise.
23261 * config/i386/xopintrin.h: Likewise.
23262 * config/i386/znver1.md: Likewise.
23263 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
23264 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
23265 * double-int.h (struct double_int): Likewise.
23266 * double-int.c (div_and_round_double): Likewise.
23267 * wide-int.cc: Likewise.
23268 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
23269 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
23270 * cfgcleanup.c (crossjumps_occured): Renamed to ...
23271 (crossjumps_occurred): ... this.
23272 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
23275 PR tree-optimization/78899
23276 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
23277 returning bool return struct loop *, NULL for failure and the new
23279 (versionable_outer_loop_p): Don't version outer loop if it has
23280 dont_vectorized bit set.
23281 (tree_if_conversion): When versioning outer loop, ensure
23282 tree_if_conversion is performed also on the inner loop of the
23283 non-vectorizable outer loop copy.
23284 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
23285 LOOP_VECTORIZED in inner loop of the scalar outer loop and
23286 prevent vectorization of it.
23287 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
23288 the outer loop vectorization of the non-scalar version is attempted
23289 before vectorization of the inner loop in scalar version. If
23290 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
23291 vectorization of its inner loop.
23292 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
23293 has 2 inner loops, rename also on edges from bb whose single pred
23294 is outer_loop->header. Fix typo in function comment.
23296 2017-01-09 Martin Sebor <msebor@redhat.com>
23299 * asan.c (asan_emit_stack_protection): Increase local buffer size
23300 to avoid snprintf truncation warning.
23302 2017-01-09 Andrew Pinski <apinski@cavium.com>
23304 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
23305 to reference thunderx2t99 for the tuning structure
23306 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
23308 (thunderx2t99_extra_costs): This.
23309 * config/aarch64/aarch64-tune.md: Regenerate.
23310 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
23311 (vulcan_addrcost_table): This.
23312 (vulcan_regmove_cost): Rename to ...
23313 (thunderx2t99_regmove_cost): This.
23314 (vulcan_vector_cost): Rename to ...
23315 (thunderx2t99_vector_cost): this.
23316 (vulcan_branch_cost): Rename to ...
23317 (thunderx2t99_branch_cost): This.
23318 (vulcan_tunings): Rename to ...
23319 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
23320 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
23322 2017-01-09 Martin Jambor <mjambor@suse.cz>
23326 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
23327 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
23328 (propagate_vr_accross_jump_function): Use the above function for all
23329 value range computations for pass-through jump functions and type
23330 converasion from explicit value range values.
23331 (ipcp_propagate_stage): Do not attempt to deduce types of formal
23332 parameters from TYPE_ARG_TYPES.
23333 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
23334 (ipa_write_node_info): Stream type of the actual argument.
23335 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
23337 2017-01-09 Martin Liska <mliska@suse.cz>
23340 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
23341 (lookup_compiler): Do not show error message with have_E.
23343 2017-01-09 Jakub Jelinek <jakub@redhat.com>
23345 PR tree-optimization/78938
23346 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
23347 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
23348 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
23349 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
23352 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23354 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
23357 2017-01-09 Richard Biener <rguenther@suse.de>
23359 PR tree-optimization/78997
23360 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
23361 name condition properly.
23363 2017-01-09 Richard Biener <rguenther@suse.de>
23366 * dwarf2out.c (is_cxx): New overload with context.
23367 (is_naming_typedef_decl): Use it.
23369 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23371 * invoke.texi (Option Summary): Correct spacing in option lists
23372 and add line breaks to fix over-long lines.
23374 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23376 PR middle-end/17660
23378 * extend.texi (Common Variable Attributes): Add xref to GCC
23379 Internals manual to explain mode attribute keywords.
23381 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
23384 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
23385 and Preprocessor Options.
23386 (Options for Linking): Document -pthread here....
23387 (RS/6000 and PowerPC Options): ...not here.
23388 (Solaris 2 Options): ...or here.
23389 * doc/cppopts.texi: Document -pthread.
23391 2017-01-08 Martin Sebor <msebor@redhat.com>
23393 PR middle-end/77708
23394 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
23395 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
23396 New member functions.
23397 (format_directive): Used them.
23399 (pass_sprintf_length::handle_gimple_call): Same.
23400 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
23401 to avoid truncation for any argument.
23402 (extract_affine_mul): Same.
23403 * tree.c (get_file_function_name): Same.
23405 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23407 PR middle-end/77484
23408 * predict.def (PRED_INDIR_CALL): Set to 86.
23410 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23412 PR preprocessor/54124
23413 * doc/cppopts.texi: Reformat -d subtable to list the full name
23414 of the options. Add cross-reference to the docs for the general
23415 compiler -d options.
23416 * doc/invoke.texi (Developer Options): Add cross-reference to the
23417 preprocessor-specific -d option documentation.
23419 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23421 PR preprocessor/13498
23422 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
23423 redudant material, and reflect new command-line options.
23424 (System Headers): Likewise.
23426 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
23428 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
23429 -isystem, and -idirafter. Copy-edit.
23430 * doc/cppopts.texi: Copy-edit. Remove contradiction about
23431 default for -ftrack-macro-expansion. Delete obsolete and
23432 badly-formatted implementation details about -fdebug-cpp output.
23433 * doc/cppwarnopts.texi: Copy-edit.
23435 2017-01-07 David Malcolm <dmalcolm@redhat.com>
23438 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
23439 that the transition from a max line width >= 1<<10 to narrower
23440 lines works correctly.
23442 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
23444 * doc/options.texi (PerFunction): New.
23445 * opt-functions.awk (switch_flags): Map both Optimization and
23446 PerFunction to CL_OPTIMIZATION.
23447 * opth-gen.awk: Test for PerFunction flag along with
23449 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
23450 it only when the latter is present. Skip those that don't in
23451 the hash function generator.
23452 * common.opt (fvar-tracking): Mark as PerFunction instead of
23454 (fvar-tracking-assignments): Likewise.
23455 (fvar-tracking-assignments-toggle): Likewise.
23456 (fvar-tracking-uninit): Likewise.
23458 2017-01-07 Jakub Jelinek <jakub@redhat.com>
23460 PR translation/79018
23461 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
23464 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
23467 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
23468 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
23469 TARGET_LONG_JUMP_TABLE_OFFSETS.
23470 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
23471 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
23472 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
23473 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
23474 * config/m68k/m68k.md (tablejump expander): Likewise.
23475 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
23476 TARGET_LONG_JUMP_TABLE_OFFSETS.
23477 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
23478 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
23480 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
23481 David Holsgrove <david.holsgrove@xilinx.com>
23483 * common/config/microblaze/microblaze-common.c
23484 (TARGET_EXCEPT_UNWIND_INFO): Remove.
23485 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
23487 * config/microblaze/microblaze.c (microblaze_must_save_register)
23488 (microblaze_expand_epilogue, microblaze_return_addr): Handle
23490 (microblaze_eh_return): New function.
23491 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
23492 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
23493 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
23494 * config/microblaze/microblaze.md (eh_return): New pattern.
23496 2017-01-06 Jakub Jelinek <jakub@redhat.com>
23498 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
23499 GCC_DIAGNOSTIC_STRINGIFY): Define.
23501 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
23503 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23505 * config/arm/arm.md (<mcrr>): New.
23507 * config/arm/arm.c (arm_arch5te): New.
23508 (arm_option_override): Set arm_arch5te.
23509 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
23511 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
23512 (arm_mcrr_qualifiers): ... this. New.
23513 (MRRC_QUALIFIERS): Define to...
23514 (arm_mrrc_qualifiers): ... this. New.
23515 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
23517 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
23518 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
23519 (MRRCI, mrrc, MRRC): New.
23520 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
23521 VUNSPEC_MRRC2): New.
23523 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23525 * config/arm/arm.md (<mcr>): New.
23527 * config/arm/arm.c (arm_coproc_builtin_available): Add
23528 support for mcr, mrc, mcr2 and mrc2.
23529 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
23530 (arm_mcr_qualifiers): ... this. New.
23531 (MRC_QUALIFIERS): Define to ...
23532 (arm_mrc_qualifiers): ... this. New.
23533 (MCR_QUALIFIERS): Define to ...
23534 (arm_mcr_qualifiers): ... this. New.
23535 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
23537 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
23538 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
23539 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
23540 VUNSPEC_MRC2): New.
23542 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23544 * config/arm/arm.md (*ldc): New.
23548 * config/arm/arm.c (arm_coproc_builtin_available): Add
23549 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
23550 (arm_coproc_ldc_stc_legitimate_address): New.
23551 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
23552 'qualifier_const_pointer'.
23553 (LDC_QUALIFIERS): Define to...
23554 (arm_ldc_qualifiers): ... this. New.
23555 (STC_QUALIFIERS): Define to...
23556 (arm_stc_qualifiers): ... this. New.
23557 * config/arm/arm-protos.h
23558 (arm_coproc_ldc_stc_legitimate_address): New.
23559 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
23560 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
23561 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
23562 stc2, stcl, stc2l): New.
23563 * config/arm/constraints.md (Uz): New.
23564 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
23565 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
23566 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
23567 VUNSPEC_STC2L): New.
23569 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23571 * config/arm/arm.md (<cdp>): New.
23572 * config/arm/arm.c (neon_const_bounds): Rename this ...
23573 (arm_const_bounds): ... this.
23574 (arm_coproc_builtin_available): New.
23575 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
23576 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
23577 (CDP_QUALIFIERS): Define to...
23578 (arm_cdp_qualifiers): ... this. New.
23580 (arm_expand_builtin_args): Add case for 6 arguments.
23581 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
23582 (arm_const_bounds): ... this.
23583 (arm_coproc_builtin_available): New.
23584 * config/arm/arm_acle.h (__arm_cdp): New.
23586 * config/arm/arm_acle_builtins.def (cdp): New.
23588 * config/arm/iterators.md (CDPI,CDP,cdp): New.
23589 * config/arm/neon.md: Rename all 'neon_const_bounds' to
23590 'arm_const_bounds'.
23591 * config/arm/types.md (coproc): New.
23592 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
23593 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
23594 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
23595 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
23597 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23599 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
23600 (UBINOP_QUALIFIERS): New.
23602 (acle_builtin_data): New. Change comment.
23603 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
23604 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
23605 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
23606 arm_acle_builtins.def.
23607 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
23608 (arm_init_acle_builtins): New.
23609 (CRC32_BUILTIN): Remove.
23610 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
23611 crc32cb, crc32ch and crc32cw.
23612 (arm_init_crc32_builtins): Remove.
23613 (arm_init_builtins): Use arm_init_acle_builtins rather
23614 than arm_init_crc32_builtins.
23615 (arm_expand_acle_builtin): New.
23616 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
23617 * config/arm/arm_acle_builtins.def: New.
23619 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
23621 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
23622 (arm_builtin_datum): ... this.
23623 (arm_init_neon_builtin): Rename to ...
23624 (arm_init_builtin): ... this. Add a new parameters PREFIX
23625 and USE_SIG_IN_NAME.
23626 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
23627 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
23628 'arm_builtin_datum'.
23629 (arm_init_vfp_builtins): Likewise.
23630 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
23631 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
23632 (arm_expand_neon_args): Rename to ...
23633 (arm_expand_builtin_args): ... this. Rename builtin_arg
23634 enum values and differentiate between ARG_BUILTIN_MEMORY
23635 and ARG_BUILTIN_NEON_MEMORY.
23636 (arm_expand_neon_builtin_1): Rename to ...
23637 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
23638 values, arm_expand_builtin_args and add bool parameter NEON.
23639 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
23640 (arm_expand_vfp_builtin): Likewise.
23641 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
23643 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
23645 PR middle-end/77484
23646 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
23647 * predict.c (tree_estimate_probability_bb): Reverse direction of
23648 polymorphic call predictor.
23650 2017-01-06 David Malcolm <dmalcolm@redhat.com>
23652 * passes.c (execute_one_pass): Split out pass-skipping logic into...
23653 (determine_pass_name_match): ...this new function and...
23654 (should_skip_pass_p): ...this new function.
23656 2017-01-06 Nathan Sidwell <nathan@acm.org>
23658 * ipa-visibility.c (function_and_variable_visibility): Reformat
23659 comments and long lines. Remove extrneous if.
23660 * symtab.c (symtab_node::make_decl_local): Fix code format.
23661 (symtab_node::set_section_for_node): Fix comment typo.
23663 2017-01-06 Martin Liska <mliska@suse.cz>
23666 * lra-constraints.c: Rename invariant to lra_invariant.
23667 * predict.c (set_even_probabilities): Initialize e to NULL.
23669 2017-01-05 Martin Sebor <msebor@redhat.com>
23671 PR tree-optimization/78910
23672 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
23673 (format_integer): Correct off-by-one error in the handling
23674 of precision with negative numbers in signed conversions..
23676 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
23678 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
23680 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23682 PR tree-optimization/71016
23683 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
23684 factor_out_conditional_conversion. Formatting fix.
23685 (factor_out_conditional_conversion): Add cond_stmt argument.
23686 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
23687 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
23690 2017-01-05 David Malcolm <dmalcolm@redhat.com>
23692 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
23693 read-rtl-function.o, and selftest-rtl.o.
23694 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
23695 (selftest::aarch64_test_loading_full_dump): New function.
23696 (selftest::aarch64_run_selftests): New function.
23697 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
23698 selftest::aarch64_run_selftests.
23699 * config/i386/i386.c
23700 (selftest::ix86_test_loading_dump_fragment_1): New function.
23701 (selftest::ix86_test_loading_call_insn): New function.
23702 (selftest::ix86_test_loading_full_dump): New function.
23703 (selftest::ix86_test_loading_unspec): New function.
23704 (selftest::ix86_run_selftests): Call the new functions.
23705 * emit-rtl.c (maybe_set_max_label_num): New function.
23706 * emit-rtl.h (maybe_set_max_label_num): New decl.
23707 * function.c (instantiate_decls): Guard call to
23708 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
23709 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
23711 * gensupport.c (gen_reader::gen_reader): Pass "false"
23712 for new "compact" param of rtx_reader.
23713 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
23714 rather than an empty string for NULL strings.
23715 * read-md.c: Potentially include config.h rather than bconfig.h.
23716 Wrap include of errors.h with #ifdef GENERATOR_FILE.
23717 (have_error): New global, copied from errors.c.
23718 (md_reader::read_name): Rename to...
23719 (md_reader::read_name_1): ...this, adding "out_loc" param,
23720 and converting "missing name or number" to returning false, rather
23722 (md_reader::read_name): Reimplement in terms of read_name_1.
23723 (md_reader::read_name_or_nil): New function.
23724 (md_reader::read_string): Handle "(nil)" by returning NULL.
23725 (md_reader::md_reader): Add new param "compact".
23726 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
23727 (md_reader::read_file): New method.
23728 * read-md.h (md_reader::md_reader): Add new param "compact".
23729 (md_reader::read_file): New method.
23730 (md_reader::is_compact): New accessor.
23731 (md_reader::read_name): Convert return type from void to file_location.
23732 (md_reader::read_name_or_nil): New decl.
23733 (md_reader::read_name_1): New decl.
23734 (md_reader::m_compact): New field.
23735 (noop_reader::noop_reader): Pass "false" for new "compact" param
23737 (rtx_reader::rtx_reader): Add new "compact" param.
23738 (rtx_reader::read_rtx_operand): Make virtual and convert return
23739 type from void to rtx.
23740 (rtx_reader::read_until): New decl.
23741 (rtx_reader::handle_any_trailing_information): New virtual function.
23742 (rtx_reader::postprocess): New virtual function.
23743 (rtx_reader::finalize_string): New virtual function.
23744 (rtx_reader::m_in_call_function_usage): New field.
23745 (rtx_reader::m_reuse_rtx_by_id): New field.
23746 * read-rtl-function.c: New file.
23747 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
23748 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
23749 (selftest::verify_three_block_rtl_cfg): New decl.
23750 * read-rtl-function.h: New file.
23751 * read-rtl.c: Potentially include config.h rather than bconfig.h.
23752 For host, include function.h, memmodel.h, and emit-rtl.h.
23753 (one_time_initialization): New function.
23754 (struct compact_insn_name): New struct.
23755 (compact_insn_names): New array.
23756 (find_code): Handle insn codes in compact dumps.
23757 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
23758 (bind_subst_iter_and_attr): Likewise.
23759 (add_condition_to_string): Likewise.
23760 (add_condition_to_rtx): Likewise.
23761 (apply_attribute_uses): Likewise.
23762 (add_current_iterators): Likewise.
23763 (apply_iterators): Likewise.
23764 (initialize_iterators): Guard usage of apply_subst_iterator with
23765 #ifdef GENERATOR_FILE.
23766 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
23767 (md_reader::read_mapping): Likewise.
23768 (add_define_attr_for_define_subst): Likewise.
23769 (add_define_subst_attr): Likewise.
23770 (read_subst_mapping): Likewise.
23771 (check_code_iterator): Likewise.
23772 (rtx_reader::read_rtx): Likewise. Move one-time initialization
23774 (one_time_initialization): New function.
23775 (rtx_reader::read_until): New method.
23776 (read_flags): New function.
23777 (parse_reg_note_name): New function.
23778 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
23779 Handle reuse_rtx ids.
23780 Wrap iterator lookup within #ifdef GENERATOR_FILE.
23781 Add parsing support for RTL dumps, mirroring the special-cases in
23782 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
23783 values, and calling handle_any_trailing_information.
23784 (rtx_reader::read_rtx_operand): Convert return type from void
23785 to rtx, returning return_rtx. Handle case 'e'. Call
23786 finalize_string on XSTR and XTMPL fields.
23787 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
23788 "(nil)" values were omitted. Call the postprocess vfunc on the
23790 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
23791 class ctor. Initialize m_in_call_function_usage. Call
23792 one_time_initialization.
23793 * rtl-tests.c (selftest::test_uncond_jump): Call
23794 set_new_first_and_last_insn.
23795 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
23796 * selftest-rtl.c: New file.
23797 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
23798 (selftest::get_insn_by_uid): New decl.
23799 * selftest-run-tests.c (selftest::run_tests): Call
23800 read_rtl_function_c_tests.
23801 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
23802 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
23805 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
23807 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
23808 operands in a special way. Assert that pos+len <= mode precision.
23810 2017-01-05 Jakub Jelinek <jakub@redhat.com>
23812 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
23813 3 argument Alias with unlimited for the negative form.
23814 (fno-vect-cost-model): Removed.
23816 2017-01-05 Martin Liska <mliska@suse.cz>
23818 * hsa-gen.c (gen_hsa_divmod): New function.
23819 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
23821 2017-01-05 Martin Liska <mliska@suse.cz>
23824 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
23827 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23829 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
23830 small constant length operands.
23832 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23834 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
23835 between loop iterations.
23837 2017-01-05 Martin Liska <mliska@suse.cz>
23840 * gimplify.c (gimplify_decl_expr): Compare to
23841 asan_poisoned_variables instread of checking flags.
23842 (gimplify_target_expr): Likewise.
23843 (gimplify_expr): Likewise.
23844 (gimplify_function_tree): Conditionally initialize
23845 asan_poisoned_variables.
23847 2017-01-04 Jeff Law <law@redhat.com>
23849 PR tree-optimizatin/78812
23850 * rtl.h (contains_mem_rtx_p): Prototype.
23851 * ifcvt.c (containts_mem_rtx_p): Move from here to...
23852 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
23853 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
23854 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
23855 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
23857 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23859 * input.c (assert_char_at_range): Default-initialize actual_range.
23861 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23863 * df-scan.c (df_ref_create_structure): Make regno unsigned,
23864 to match the caller.
23866 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23868 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
23869 insns after final jump in test to emit dummy move.
23871 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23873 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
23874 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
23876 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
23878 * multiple_target.c (create_dispatcher_calls): Init e_next.
23879 * tree-ssa-loop-split.c (split_loop): Init border.
23880 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
23883 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
23888 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
23889 (altivec_register_operand): Do not return true if the operand
23890 contains a SUBREG mixing SImode and SFmode.
23891 (vsx_register_operand): Likewise.
23892 (vsx_reg_sfsubreg_ok): New predicate.
23893 (vfloat_operand): Do not return true if the operand contains a
23894 SUBREG mixing SImode and SFmode.
23895 (vint_operand): Likewise.
23896 (vlogical_operand): Likewise.
23897 (gpc_reg_operand): Likewise.
23898 (int_reg_operand): Likewise.
23899 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
23900 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
23901 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
23903 (rs6000_emit_move_si_sf_subreg): New helper function.
23904 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
23905 fixup SUBREGs involving SImode and SFmode.
23906 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
23907 numbers for the new peephole2 optimization.
23908 (peephole2 for SFmode unions): New peephole2 to optimize cases in
23909 the GLIBC math library that do AND/IOR/XOR operations on single
23910 precision floating point.
23911 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
23912 target macros to say whether we need to avoid SUBREGs mixing
23914 (TARGET_ALLOW_SF_SUBREG): Likewise.
23915 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
23916 (UNSPEC_SI_FROM_SF): Likewise.
23917 (iorxor): Change spacing.
23918 (and_ior_xor): New iterator for AND, IOR, and XOR.
23919 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
23920 (movdi_from_sf_zero_ext): Likewise.
23921 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
23922 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
23923 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
23924 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
23925 (fms<mode>4): Likewise.
23926 (fnma<mode>4): Likewise.
23927 (fnms<mode>4): Likewise.
23928 (nfma<mode>4): Likewise.
23929 (nfms<mode>4): Likewise.
23931 2017-01-04 Marek Polacek <polacek@redhat.com>
23934 * doc/invoke.texi: Document -Wpointer-compare.
23936 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23938 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
23941 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
23942 descriptions for -gdwarf-5 and emit them as uleb128 instead of
23945 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
23948 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
23949 documentation of the powerpc_popcntb_ok attribute.
23950 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23951 code to issue warning messages if a requested CPU configuration is
23952 not supported by the binary (assembler and loader) toolchain.
23953 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
23954 made to define a built-in function that has been disabled.
23955 (paired_init_builtins): Add assertion to prevent ICE if attempt is
23956 made to define a built-in function that has been disabled.
23957 (altivec_init_builtins): Add comment explaining why definition
23958 of the DST built-in functions is not preceded by an assertion
23959 check. Add assertions to prevent ICE if attempts are made to
23960 define an altivec predicate or an abs* built-in function that has
23962 (htm_init_builtins): Add comment explaining why definition of the
23963 htm built-in functions is not preceded by an assertion check.
23965 2017-01-04 Jeff Law <law@redhat.com>
23967 PR tree-optimizatin/67955
23968 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
23969 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
23970 the points-to solution does not include pt_null. Use DECL_PT_UID
23973 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
23975 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
23976 Use gen_int_mode instead of gen_lopwart for const_int operands.
23978 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23980 PR tree-optimization/71563
23981 * match.pd: Simplify X << Y into X if Y is known to be 0 or
23982 out of range value - has low bits known to be zero.
23984 2017-01-04 Alan Modra <amodra@gmail.com>
23986 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
23987 * configure: Regenerate.
23988 * config.in: Regenerate.
23990 2017-01-04 Jakub Jelinek <jakub@redhat.com>
23993 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
23994 a substring of the message, but strcmp with the whole message. Ifdef
23995 ENABLE_NLS, translate the message first using dgettext.
23997 2017-01-03 Jeff Law <law@redhat.com>
23999 PR tree-optimizatin/78856
24000 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
24001 (mark_threaded_blocks): Remove code to truncate thread paths that
24002 cross multiple loop headers. Instead invalidate the cached loop
24003 iteration information and handle case of a thread path walking
24004 into an irreducible region.
24006 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
24009 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
24010 assertions. Add support for doing the signbit if the IEEE 128-bit
24011 floating point value is in a GPR.
24012 * config/rs6000/rs6000.md (Fsignbit): Delete.
24013 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
24014 Update the length attribute if the value is in a GPR.
24015 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
24016 the sign or zero extension instruction, since the value is always 0/1.
24017 (signbit<mode>2_dm2): Delete using <Fsignbit>.
24020 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
24021 extracting SImode to a GPR register so that we can generate a
24022 store, limit the vector to be in a traditional Altivec register
24023 for the vextuwrx instruction.
24025 2017-01-03 Ian Lance Taylor <iant@google.com>
24027 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
24029 2017-01-03 Martin Sebor <msebor@redhat.com>
24031 PR tree-optimization/78696
24032 * gimple-ssa-sprintf.c (format_floating): Correct handling of
24033 precision. Use MPFR for %f for greater fidelity. Correct handling
24035 (pass_sprintf_length::compute_format_length): Set width and precision
24036 specified by asrerisk to void_node for vararg functions.
24037 (try_substitute_return_value): Adjust dump output.
24039 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
24041 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
24043 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
24045 * doc/invoke.texi (SPARC options): Document -mlra as the default.
24046 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
24047 -mlra/-mno-lra was passed to the compiler.
24049 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
24051 PR rtl-optimization/65618
24052 * emit-rtl.c (try_split): Move initialization of "before" and
24053 "after" to just before the call to emit_insn_after_setloc.
24055 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
24057 * doc/md.texi (Standard Names): Remove reference to Java frontend.
24059 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
24061 * dwarf2out.c (gen_enumeration_type_die): When
24062 -gno-strict-dwarf, add a DW_AT_encoding attribute.
24064 2017-01-03 Jakub Jelinek <jakub@redhat.com>
24066 PR tree-optimization/78965
24067 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
24068 Change first argument from const call_info & to call_info &. For %n
24069 set info.nowrite to false.
24071 PR middle-end/78901
24072 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
24073 possibly throwing calls.
24075 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
24076 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
24077 and fns handling, rather than in a separate case SSA_NAME.
24079 2017-01-02 Jeff Law <law@redhat.com>
24081 * config/darwin-driver.c (darwin_driver_init): Const-correctness
24082 fixes for first_period and second_period variables.
24084 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
24087 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
24088 (*insvqi_1): New insn pattern.
24089 (*insvqi_1_mem_rex64): Ditto.
24090 (*insvqi_2): Ditto.
24091 (*insvqi_3): Rename from *insvqi.
24093 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
24095 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
24097 * doc/cfg.texi (Edges): Remove reference to Java.
24098 (Maintaining the CFG): Ditto.
24100 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24102 PR middle-end/77674
24103 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
24104 transparent aliases.
24106 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
24108 PR middle-end/77484
24109 * predict.def (PRED_CALL): Update hitrate.
24110 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
24111 * predict.c (tree_estimate_probability_bb): Split CALL predictor
24112 into direct/indirect/polymorphic variants.
24114 2017-01-01 Jakub Jelinek <jakub@redhat.com>
24116 Update copyright years.
24118 * gcc.c (process_command): Update copyright notice dates.
24119 * gcov-dump.c (print_version): Ditto.
24120 * gcov.c (print_version): Ditto.
24121 * gcov-tool.c (print_version): Ditto.
24122 * gengtype.c (create_file): Ditto.
24123 * doc/cpp.texi: Bump @copying's copyright year.
24124 * doc/cppinternals.texi: Ditto.
24125 * doc/gcc.texi: Ditto.
24126 * doc/gccint.texi: Ditto.
24127 * doc/gcov.texi: Ditto.
24128 * doc/install.texi: Ditto.
24129 * doc/invoke.texi: Ditto.
24131 Copyright (C) 2017 Free Software Foundation, Inc.
24133 Copying and distribution of this file, with or without modification,
24134 are permitted in any medium without royalty provided the copyright
24135 notice and this notice are preserved.