1 2017-11-16 Martin Sebor <msebor@redhat.com>
3 PR tree-optimization/82588
4 PR tree-optimization/82583
5 * tree-vrp.c (check_array_ref): Handle flexible array members,
6 string literals, and inner indices.
7 (search_for_addr_array): Add detail to diagnostics.
9 2017-11-16 Nathan Sidwell <nathan@acm.org>
13 * tree.h (COPY_DECL_RTL): Rename parms for clarity.
14 (SET_DECL_ASSEMBLER_NAME): Forward to
15 overwrite_decl_assembler_name.
16 (COPY_DECL_ASSEMBLER_NAME): Rename parms for clarity.
17 (overwrite_decl_assembler_name): Declare.
18 * tree.c (overwrite_decl_assembler_name): New.
19 * langhooks-def.h (lhd_overwrite_decl_assembler_name): Declare.
20 (LANG_HOOKS_OVERWRITE_DECL_ASSEMBLER_NAME): Provide default.
21 (LANG_HOOKS_INITIALIZER): Add it.
22 * langhooks.h (struct lang_hooks): Add overwrite_decl_assembler_name.
23 * langhooks.c (lhd_set_decl_assembler_name): Use
24 SET_DECL_ASSEMBLER_NAME.
25 (lhd_overwrite_decl_assembler_name): Default implementation.
27 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
28 Jackson Woodruff <jackson.woodruff@arm.com>
30 PR tree-optimization/71026
31 * match.pd: Canonicalize constant multiplies in division.
33 2017-11-16 Wilco Dijkstra <wdijkstr@arm.com>
35 * opts.c (default_options_table): Add OPT_fomit_frame_pointer entry.
36 * common/config/alpha/alpha-common.c (TARGET_OPTION_OPTIMIZATION_TABLE):
37 Remove OPT_fomit_frame_pointer entry.
38 * common/config/arc/arc-common.c: Likewise.
39 * common/config/arm/arm-common.c: Likewise.
40 * common/config/avr/avr-common.c: Likewise.
41 * common/config/c6x/c6x-common.c: Likewise.
42 * common/config/cr16/cr16-common.c: Likewise.
43 * common/config/cris/cris-common.c: Likewise.
44 * common/config/epiphany/epiphany-common.c: Likewise.
45 * common/config/fr30/fr30-common.c: Likewise.
46 * common/config/frv/frv-common.c: Likewise.
47 * common/config/ia64/ia64-common.c: Likewise.
48 * common/config/iq2000/iq2000-common.c: Likewise.
49 * common/config/lm32/lm32-common.c: Likewise.
50 * common/config/m32r/m32r-common.c: Likewise.
51 * common/config/mcore/mcore-common.c: Likewise.
52 * common/config/microblaze/microblaze-common.c: Likewise.
53 * common/config/mips/mips-common.c: Likewise.
54 * common/config/mmix/mmix-common.c: Likewise.
55 * common/config/mn10300/mn10300-common.c: Likewise.
56 * common/config/nios2/nios2-common.c: Likewise.
57 * common/config/pa/pa-common.c: Likewise.
58 * common/config/pdp11/pdp11-common.c: Likewise.
59 * common/config/powerpcspe/powerpcspe-common.c: Likewise.
60 * common/config/riscv/riscv-common.c: Likewise.
61 * common/config/rs6000/rs6000-common.c: Likewise.
62 * common/config/rx/rx-common.c: Likewise.
63 * common/config/s390/s390-common.c: Likewise.
64 * common/config/sh/sh-common.c: Likewise.
65 * common/config/sparc/sparc-common.c: Likewise.
66 * common/config/tilegx/tilegx-common.c: Likewise.
67 * common/config/tilepro/tilepro-common.c: Likewise.
68 * common/config/v850/v850-common.c: Likewise.
69 * common/config/visium/visium-common.c: Likewise.
70 * common/config/xstormy16/xstormy16-common.c: Likewise.
71 * common/config/xtensa/xtensa-common.c: Likewise.
72 * invoke.texi (-fomit-frame-pointer): Update documentation.
74 2017-11-16 Jan Hubicka <hubicka@ucw.cz>
76 * tree-cfg.c (gimple_find_sub_bbs): Do not compute freq.
78 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
80 * cfg.c (scale_bbs_frequencies_int,
81 cale_bbs_frequencies_gcov_type): Remove.
82 * cfg.h (scale_bbs_frequencies_int,
83 cale_bbs_frequencies_gcov_type): Remove.
85 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
87 * tree-ssa-loop-manip.c
88 (scale_dominated_blocks_in_loop): Update to profile counts.
89 (tree_transform_and_unroll_loop): Likewise.
91 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
93 * tree-vect-loop-manip.c (vect_do_peeling): Do not use
94 scale_bbs_frequencies_int.
96 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
98 * final.c (compute_alignments): Use counts rather than frequencies.
100 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
102 * cfgloopanal.c: Include sreal.h
103 (average_num_loop_insns): Use counts and sreal for accounting.
105 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
107 * cfgloopmanip.c (duplicate_loop_to_header_edge): Cleanup profile
110 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
111 Kito Cheng <kito.cheng@gmail.com>
113 * config/nds32/constraints.md: Provide more constraints.
114 * config/nds32/nds32.h (enum reg_class, REG_CLASS_NAMES,
115 REG_CLASS_CONTENTS): Define R5_REG, R8_REG, and FRAME_POINTER_REG to
116 support constraints usage.
118 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
120 * config/nds32/constants.md (UNSPEC_VOLATILE_FUNC_RETURN): Remove.
122 2017-11-16 Chung-Ju Wu <jasonwucj@gmail.com>
123 Kito Cheng <kito.cheng@gmail.com>
125 * config/nds32/nds32.opt: Add mext-perf, mext-perf2, mext-string.
126 * config/nds32/nds32.opt: Refine the layout.
127 * config/nds32/nds32.c (TARGET_EXT_PERF, TARGET_EXT_PERF2,
128 TARGET_EXT_STRING): Support new options.
129 * config/nds32/nds32.h: Likewise.
130 * config/nds32/nds32.md: Likewise.
131 * config/nds32/nds32-predicates.c: Likewise.
132 * config/nds32/constraints.md: Likewise.
133 * common/config/nds32/nds32-common.c: Likewise.
135 2017-11-16 Julia Koval <julia.koval@intel.com>
138 * config/i386/gfniintrin.h: Add sse check.
139 * config/i386/i386.c (ix86_expand_builtin): Fix gfni check.
141 2017-11-16 Julia Koval <julia.koval@intel.com>
143 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VBMI2_SET,
144 OPTION_MASK_ISA_AVX512VBMI2_UNSET): New.
145 (ix86_handle_option): Handle -mavx512vbmi2.
146 * config/i386/cpuid.h: Add bit_AVX512VBMI2.
147 * config/i386/driver-i386.c (host_detect_local_cpu): Handle new bit.
148 * config/i386/i386-c.c (__AVX512VBMI2__): New.
149 * config/i386/i386.c (ix86_target_string): Handle -mavx512vbmi2.
150 (ix86_valid_target_attribute_inner_p): Ditto.
151 * config/i386/i386.h (TARGET_AVX512VBMI2, TARGET_AVX512VBMI2_P): New.
152 * config/i386/i386.opt (mavx512vbmi2): New option.
153 * doc/invoke.texi: Add new option.
155 [2017-11-16 Julia Koval <julia.koval@intel.com>
157 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8, _mm256_gf2p8mul_epi8,
158 _mm_mask_gf2p8mul_epi8, _mm_maskz_gf2p8mul_epi8,
159 _mm256_mask_gf2p8mul_epi8, _mm256_maskz_gf2p8mul_epi8,
160 _mm512_mask_gf2p8mul_epi8, _mm512_maskz_gf2p8mul_epi8,
161 _mm512_gf2p8mul_epi8): New intrinsics.
162 * config/i386/i386-builtin-types.def
163 (V64QI_FTYPE_V64QI_V64QI): New type.
164 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8mulb_v64qi,
165 __builtin_ia32_vgf2p8mulb_v64qi_mask, __builtin_ia32_vgf2p8mulb_v32qi,
166 __builtin_ia32_vgf2p8mulb_v32qi_mask, __builtin_ia32_vgf2p8mulb_v16qi,
167 __builtin_ia32_vgf2p8mulb_v16qi_mask): New builtins.
168 * config/i386/sse.md (vgf2p8mulb_*): New pattern.
169 * config/i386/i386.c (ix86_expand_args_builtin): Handle new type.
171 2017-11-15 Uros Bizjak <ubizjak@gmail.com>
173 * config/i386/i386.c (x86_print_call_or_nop): Emit 5 byte nop
174 explicitly as a stream of bytes.
176 2017-11-15 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
178 * config/rs6000/altivec.h (vec_xst_be): New #define.
179 * config/rs6000/altivec.md (altivec_vperm_<mode>_direct): Rename
180 and externalize from *altivec_vperm_<mode>_internal.
181 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI): Remove macro
183 (XL_BE_V8HI): Likewise.
184 (XL_BE_V4SI): Likewise.
185 (XL_BE_V4SI): Likewise.
186 (XL_BE_V2DI): Likewise.
187 (XL_BE_V4SF): Likewise.
188 (XL_BE_V2DF): Likewise.
189 (XST_BE): Add BU_VSX_OVERLOAD_X macro instantiation.
190 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Correct
191 all array entries with these keys: VSX_BUILTIN_VEC_XL,
192 VSX_BUILTIN_VEC_XL_BE, VSX_BUILTIN_VEC_XST. Add entries for key
193 VSX_BUILTIN_VEC_XST_BE.
194 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Remove.
195 (altivec_expand_builtin): Remove handling for VSX_BUILTIN_XL_BE_*
197 (altivec_init_builtins): Replace conditional calls to def_builtin
198 for __builtin_vsx_ld_elemrev_{v8hi,v16qi} and
199 __builtin_vsx_st_elemrev_{v8hi,v16qi} based on TARGET_P9_VECTOR
200 with unconditional calls. Remove calls to def_builtin for
201 __builtin_vsx_le_be_<mode>. Add a call to def_builtin for
202 __builtin_vec_xst_be.
203 * config/rs6000/vsx.md (vsx_ld_elemrev_v8hi): Convert define_insn
204 to define_expand, and add alternate RTL generation for P8.
205 (*vsx_ld_elemrev_v8hi_internal): New define_insn based on
207 (vsx_ld_elemrev_v16qi): Convert define_insn to define_expand, and
208 add alternate RTL generation for P8.
209 (*vsx_ld_elemrev_v16qi_internal): New define_insn based on
210 vsx_ld_elemrev_v16qi.
211 (vsx_st_elemrev_v8hi): Convert define_insn
212 to define_expand, and add alternate RTL generation for P8.
213 (*vsx_st_elemrev_v8hi_internal): New define_insn based on
215 (vsx_st_elemrev_v16qi): Convert define_insn to define_expand, and
216 add alternate RTL generation for P8.
217 (*vsx_st_elemrev_v16qi_internal): New define_insn based on
218 vsx_st_elemrev_v16qi.
220 2017-11-15 H.J. Lu <hongjiu.lu@intel.com>
223 * config/i386/i386.c (pass_insert_vzeroupper::gate): Remove
224 TARGET_AVX512ER check.
225 (ix86_option_override_internal): Set MASK_VZEROUPPER if
226 neither -mzeroupper nor -mno-zeroupper is used and
227 TARGET_EMIT_VZEROUPPER is set.
228 * config/i386/i386.h (TARGET_EMIT_VZEROUPPER): New.
229 * config/i386/x86-tune.def: Add X86_TUNE_EMIT_VZEROUPPER.
231 2017-11-15 Will Schmidt <will_schmidt@vnet.ibm.com>
233 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
234 folding of vector compares.
235 (fold_build_vec_cmp): New helper function.
236 (fold_compare_helper): New helper function.
237 (builtin_function_type): Add compare builtins to the list of functions
238 having unsigned arguments. Cosmetic updates to comment indentation.
239 * config/rs6000/vsx.md (vcmpneb, vcmpneh, vcmpnew): Update to specify
240 the not+eq combination.
242 2017-11-15 Bin Cheng <bin.cheng@arm.com>
244 PR tree-optimization/82726
245 PR tree-optimization/70754
246 * tree-predcom.c (order_drefs_by_pos): New function.
247 (combine_chains): Move code setting has_max_use_after to...
248 (try_combine_chains): ...here. New parameter. Sort combined chains
249 according to position information.
250 (tree_predictive_commoning_loop): Update call to above function.
251 (update_pos_for_combined_chains, pcom_stmt_dominates_stmt_p): New.
253 2017-11-15 Bin Cheng <bin.cheng@arm.com>
255 PR tree-optimization/82726
257 2017-01-23 Bin Cheng <bin.cheng@arm.com>
259 PR tree-optimization/70754
260 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
261 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
262 combined stmt before it if not NULL.
263 (combine_chains): Process refs reversely and compute dominance point
267 2017-02-23 Bin Cheng <bin.cheng@arm.com>
269 PR tree-optimization/79663
270 * tree-predcom.c (combine_chains): Process refs in reverse order
271 only for ZERO length chains, and add explaining comment.
273 2017-11-15 Tamar Christina <tamar.christina@arm.com>
275 * config/arm/arm-cpus.in (armv8_3, ARMv8_3a, armv8.3-a): New
276 * config/arm/arm-tables.opt (armv8.3-a): Regenerated.
277 * doc/invoke.texi (ARM Options): Add armv8.3-a.
279 2017-11-15 Tamar Christina <tamar.christina@arm.com>
281 * config/arm/arm.h (TARGET_DOTPROD): Add arm_arch8_2.
283 2017-11-15 Martin Liska <mliska@suse.cz>
285 * tree-cfg.c (pass_warn_function_return::execute):
286 Compare warn_return_type for greater than zero.
288 2017-11-15 Sebastian Peryt <sebastian.peryt@intel.com>
292 * config/i386/i386.c (pass_insert_vzeroupper): Modify gate condition
293 to return true on Xeon and not on Xeon Phi.
294 (ix86_check_avx256_register): Changed to ...
295 (ix86_check_avx_upper_register): ... this. Add extra check for
296 VALID_AVX512F_REG_OR_XI_MODE.
297 (ix86_avx_u128_mode_needed): Changed
298 ix86_check_avx256_register to ix86_check_avx_upper_register.
299 (ix86_check_avx256_stores): Changed to ...
300 (ix86_check_avx_upper_stores): ... this. Changed
301 ix86_check_avx256_register to ix86_check_avx_upper_register.
302 (ix86_avx_u128_mode_after): Changed
303 avx_reg256_found to avx_upper_reg_found. Changed
304 ix86_check_avx256_stores to ix86_check_avx_upper_stores.
305 (ix86_avx_u128_mode_entry): Changed
306 ix86_check_avx256_register to ix86_check_avx_upper_register.
307 (ix86_avx_u128_mode_exit): Ditto.
308 * config/i386/i386.h: (host_detect_local_cpu): New define.
310 2017-11-15 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
312 * config/arm/xgene1.md (xgene1): Split into automatons
313 xgene1_main, xgene1_decoder, xgene1_div, xgene1_simd.
314 (xgene1_f_load): Adjust reservations and/or types.
315 (xgene1_f_store): Likewise.
316 (xgene1_load_pair): Likewise.
317 (xgene1_store_pair): Likewise.
318 (xgene1_fp_load1): Likewise.
319 (xgene1_load1): Likewise.
320 (xgene1_store1): Likewise.
321 (xgene1_move): Likewise.
322 (xgene1_alu): Likewise.
323 (xgene1_simd): Likewise.
324 (xgene1_bfm): Likewise.
325 (xgene1_neon_load1): Likewise.
326 (xgene1_neon_store1): Likewise.
327 (xgene1_neon_logic): Likewise.
328 (xgene1_neon_st1): Likewise.
329 (xgene1_neon_ld1r): Likewise.
330 (xgene1_alu_cond): Added.
331 (xgene1_shift_reg): Likwise.
332 (xgene1_bfx): Likewise.
333 (xgene1_mul): Split into xgene1_mul32, xgene1_mul64.
335 2017-11-15 Jakub Jelinek <jakub@redhat.com>
338 * internal-fn.c: Include gimple-ssa.h, tree-phinodes.h and
340 (can_widen_mult_without_libcall): New function.
341 (expand_mul_overflow): If only checking unsigned mul overflow,
342 not result, and can do efficiently MULT_HIGHPART_EXPR, emit that.
343 Don't use WIDEN_MULT_EXPR if it would involve a libcall, unless
344 no other way works. Add MULT_HIGHPART_EXPR + MULT_EXPR support.
345 (expand_DIVMOD): Formatting fix.
346 * expmed.h (expand_mult): Add NO_LIBCALL argument.
347 * expmed.c (expand_mult): Likewise. Use OPTAB_WIDEN rather
348 than OPTAB_LIB_WIDEN if NO_LIBCALL is true, and allow it to fail.
350 PR tree-optimization/82977
351 * tree-ssa-strlen.c (strlen_optimize_stmt): Pass a reference to a copy
352 constructed temporary to strlen_to_stridx.put.
354 2017-11-15 Martin Liska <mliska@suse.cz>
356 * configure.ac: Remove -fkeep-inline-functions from coverage_flags.
357 * configure: Regenerate.
359 2017-11-15 Martin Liska <mliska@suse.cz>
362 * config/sh/sh-mem.cc: Use proper probability for
365 2017-11-14 Jeff Law <law@redhat.com>
367 * explow.c (anti_adjust_stack_and_probe_stack_clash): Avoid probing
368 the red zone for stack_clash_protection_final_dynamic_probe targets
369 when the total dynamic stack size is zero bytes.
371 * tree-ssa-threadupdate.c (thread_through_all_blocks): Thread
372 blocks is post order.
374 2017-11-15 Alexandre Oliva <aoliva@redhat.com>
376 * dumpfile.h (TDF_COMPARE_DEBUG): New.
377 * final.c (rest_of_clean_state): Set it for the
378 -fcompare-debug dump.
379 * tree-pretty-print.c (dump_generic_node): Omit OBJ_TYPE_REF
380 class when TDF_COMPARE_DEBUG is set.
382 * dwarf2out.c (gen_producer_string): Discard
385 2017-11-15 Joseph Myers <joseph@codesourcery.com>
388 * doc/extend.texi (Other Builtins): Document __builtin_tgmath.
389 * ginclude/tgmath.h (__tg_cplx, __tg_ldbl, __tg_dbl, __tg_choose)
390 (__tg_choose_2, __tg_choose_3, __TGMATH_REAL_1_2)
391 (__TGMATH_REAL_2_3): Remove macros.
392 (__TGMATH_CPLX, __TGMATH_CPLX_2, __TGMATH_REAL, __TGMATH_REAL_2)
393 (__TGMATH_REAL_3, __TGMATH_CPLX_ONLY): Define using
395 (frexp, ldexp, nexttoward, scalbn, scalbln): Define using
397 (remquo): Define using __TGMATH_REAL_3.
399 2017-11-14 Jeff Law <law@redhat.com>
401 * vr-values.c: New file with contents extracted from tree-vrp.c.
402 * Makefile.in (OBJS): Add vr-values.o
403 * tree-vrp.h (set_value_range_to_nonnull): Prototype.
404 (set_value_range, set_and_canonicalize_value_range): Likewise.
405 (vrp_bitmap_equal_p, range_is_nonnull): Likewise.
406 (value_range_constant_singleton, symbolic_range_p): Likewise.
407 (compare_values, compare_values_warnv, vrp_val_is_min): Likewise.
408 (vrp_val_is_max, copy_value_range, set_value_range_to_value): Likewise.
409 (extract_range_from_binary_expr_1, vrp_val_min, vrp_val_max): Likewise.
410 (set_value_range_to_null, range_int_cst_p, opreand_less_p): Likewise.
411 (find_case_label_range, find_case_label_index): Likewise.
412 (zero_nonzero_bits_from_vr, overflow_comparison_p): Likewise.
413 (range_int_cst_singleton_p, value_inside_range): Likewise.
414 (get_single_symbol): Likewise.
415 (switch_update): Move structure definition here.
416 (to_remove_edges, to_update_switch_stmts): Provide externs.
417 * tree-vrp.c: Move all methods for vr-values class to vr-values.c
418 (vrp_val_max, vrp_val_min, vrp_val_is_max): Make externally visible.
419 (vrp_val_is_min, set_value_range): Likewise.
420 (set_and_canonicalize_value_range, copy_value_range): Likewise.
421 (set_value_range_to_value, set_value_range_to_nonnull): Likewise.
422 (set_value_range_to_null, vrp_bitmap_equal_p): Likewise.
423 (range_is_nonnull, range_int_cst_p): Likewwise.
424 (range_int_cst_singleton_p, symbolic_range_p): Likewise.
425 (get_single_symbol, operand_less_p): Likewise
426 (compare_values_warnv, compare_values): Likewise.
427 (value_inside_range, value_range_constant_singleton): Likewise.
428 (zero_nonzero_bitgs_from_vr): Likewise.
429 (extract_range_from_binary_expr_1): Likewise.
430 (overflow_comparison_p): Likewise.
431 (to_remove_edges, to_update_switch_stmts): Likewise.
432 (find_case_label-index, find_case_label_range): Likewise.
433 (switch_update, set_value_range_to_nonnegative): Remove.
434 (set_value_range_to_truthvalue): Likewise.
435 (symbolic_range_based_on_p, gimple_assign_nonzero_p): Likewise.
436 (gimple_stmt_nonzero_p, compare_ranges): Likewise.
437 (compare_range_with_value, vrp_valueize, vrp_valueize_1): Likewise.
438 (find_case_label_ranges, test_for_singularity): Likewise.
439 (range_fits_type_p, simplify_conversion_using_ranges): LIkewise.
440 (x_vr_values): Move to its remaining use site.
442 2017-11-10 Jeff Law <law@redhat.com>
444 * vr-values.h (VR_INITIALIZER): Move #define here.
445 * gimple-ssa-evrp.c: New file with contents extracted from tree-vrp.c
446 * Makefile.in (OBJS): Add tree-evrp.o
447 * tree-vrp.h (assert_info): Move structure definition here.
448 (set_value_range_to_varying): Prototype.
449 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
450 (infer_value_range, register_edge_assert_for): Likewise.
451 (stmt_interesting_for_vrp): Likewise.
452 * tree-vrp.c: Move all methods for evrp class into gimple-ssa-evrp.c.
453 (set_value_range_to_varying): No longer static.
454 (vrp_operand_equal_p, range_includes_zero_p): Likewise.
455 (infer_value_range, register_edge_assert_for): Likewise.
457 2017-11-14 Michael Meissner <meissner@linux.vnet.ibm.com>
459 * config/rs6000/rs6000.md (bswapdi2): On 32-bit ISA 3.0, don't
460 generate the XXBRD instruction.
462 * config/rs6000/rs6000-c.c (is_float128_p): New helper function.
463 (rs6000_builtin_type_compatible): Treat _Float128 and long double
464 as being compatible if -mabi=ieeelongdouble.
465 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_HW_1): New macros
466 to setup float128 built-ins with hardware support.
467 (BU_FLOAT128_HW_2): Likewise.
468 (BU_FLOAT128_HW_3): Likewise.
469 (BU_FLOAT128_HW_VSX_1): Likewise.
470 (BU_FLOAT128_HW_VSX_2): Likewise.
471 (scalar_extract_expq): Change float128 built-in functions to
472 accommodate having both KFmode and TFmode functions. Use the
473 KFmode variant as the default.
474 (scalar_extract_sigq): Likewise.
475 (scalar_test_neg_qp): Likewise.
476 (scalar_insert_exp_q): Likewise.
477 (scalar_insert_exp_qp): Likewise.
478 (scalar_test_data_class_qp): Likewise.
479 (sqrtf128_round_to_odd): Delete processing the round to odd
480 built-in functions as special built-in functions, and define them
481 as float128 built-ins. Use the KFmode variant as the default.
482 (truncf128_round_to_odd): Likewise.
483 (addf128_round_to_odd): Likewise.
484 (subf128_round_to_odd): Likewise.
485 (mulf128_round_to_odd): Likewise.
486 (divf128_round_to_odd): Likewise.
487 (fmaf128_round_to_odd): Likewise.
488 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
489 support for KFmode and TFmode xststdcqp calls.
490 (rs6000_expand_builtin): If long double is IEEE 128-bit floating
491 point, switch the built-in handlers for the get/set float128
492 exponent, get float128 mantissa, float128 test built-ins, and the
493 float128 round to odd built-in functions. Eliminate creating the
494 float128 round to odd built-in functions as special built-ins.
495 (rs6000_init_builtins): Eliminate special creation of the float128
496 round to odd built-in functions.
497 * config/rs6000/vsx.md (xsxexpqp_<mode>): Change float128 built-in
498 function insns to support both TFmode and KFmode variants.
499 (xsxsigqp_<mode>): Likewise.
500 (xsiexpqpf_<mode>): Likewise.
501 (xsiexpqp_<mode>): Likewise.
502 (xststdcqp_<mode>): Likewise.
503 (xststdcnegqp_<mode>): Likewise.
504 (xststdcqp_<mode>): Likewise.
506 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
508 * tree-ssa-threadupdate.c (compute_path_counts): Remove
509 unused path_in_freq_ptr parameter.
510 (ssa_fix_duplicate_block_edges): Do not pass around path_in_freq
512 2017-11-14 Jan Hubicka <hubicka@ucw.cz>
514 * ipa-inline.c (edge_badness): Dump sreal frequency.
515 (compute_inlined_call_time): Match natural implementaiton ...
516 * ipa-fnsummary.c (estimate_edge_size_and_time): ... here; remove
517 forgotten division by CGRAPH_FREQ_BASE.
519 2017-11-14 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
521 * config.gcc (*-*-solaris2*): Enable default_use_cxa_atexit since
522 Solaris 11. Update comment.
523 * configure.ac (gcc_cv_ld_pid): Adapt comment for Solaris 12
525 * config/sol2.h (STARTFILE_SPEC): Likewise.
526 * configure: Regenerate.
528 2017-11-14 Carl Love <cel@us.ibm.com>
530 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Remove
531 le_ and be_ prefixes to swap* variables. Remove
532 if (VECTOR_ELT_ORDER_BIG) statement. Remove E_V16QImode case
535 2017-11-14 Jason Merrill <jason@redhat.com>
537 Support GTY((cache)) on hash_map.
538 * hash-traits.h (ggc_remove): Add ggc_maybe_mx member function.
539 (ggc_cache_remove): Override it instead of ggc_mx.
540 * hash-table.h (gt_ggc_mx): Call it instead of ggc_mx.
541 (gt_cleare_cache): Call ggc_mx instead of gt_ggc_mx.
542 * hash-map-traits.h (simple_hashmap_traits): Add maybe_mx member.
543 (simple_cache_map_traits): Override maybe_mx.
544 * hash-map.h (hash_entry): Add ggc_maybe_mx and keep_cache_entry.
545 (hash_map): Friend gt_cleare_cache.
546 (gt_cleare_cache): New.
547 * tree.h (tree_cache_traits): New hash_map traits class.
548 (tree_cache_map): New typedef.
550 2017-11-14 Richard Biener <rguenther@suse.de>
552 * tree-cfgcleanup.c (cleanup_control_expr_graph): Remove first_p
553 paramter and handling.
554 (cleanup_control_flow_bb): Likewise.
555 (cleanup_control_flow_pre): New helper performing a DFS walk
556 to call cleanup_control_flow_bb in PRE order.
557 (cleanup_tree_cfg_1): Do the first phase of cleanup_control_flow_bb
558 via cleanup_control_flow_pre.
560 2017-11-14 James Greenhalgh <james.greenhalgh@arm.com>
562 * config/aarch64/aarch64-simd.md
563 (aarch64_simd_bsl<mode>_internal): Remove DImode.
564 (*aarch64_simd_bsl<mode>_alt): Likewise.
565 (aarch64_simd_bsldi_internal): New.
566 (aarch64_simd_bsldi_alt): Likewise.
568 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
570 * tracer.c (better_p): Do not compare frequencies.
571 * reg-stack.c (better_edge): Likewise.
572 * shrink-wrap.c (try_shrink_wrapping): Do not convert to gcov counts
575 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
577 * auto-profile.c (afdo_annotate_cfg): Use update_max_bb_count.
578 * cgraphunit.c (cgraph_node::expand_thunk): Use update_max_bb_count.
579 * ipa-utils.c (ipa_merge_profiles): Use update_max_bb_count.
580 * lto-streamer-in.c (input_function): Use update_max_bb_count.
581 * omp-expand.c (expand_omp_taskreg): Use update_max_bb_count.
582 * predict.c (maybe_hot_frequency_p): Inline to ...
583 (maybe_hot_count_p): ... here; rewrite to counts.
584 (counts_to_freqs): Rename to ...
585 (update_max_bb_count): ... this one.
586 (expensive_function_p): Use counts.
587 (estimate_bb_frequencies): Update.
588 (rebuild_frequencies): Update.
589 * predict.h (counts_to_freqs): Rename to ...
590 (update_max_bb_count): ... this one.
591 * profile.c (compute_branch_probabilities): Add debug info
592 * tree-inline.c (expand_call_inline): Update debug info.
593 (optimize_inline_calls): Use update_max_bb_count..
594 (tree_function_versioning): Use update_max_bb_count..
595 * value-prof.c (gimple_value_profile_transformations):
596 Do not use update_max_bb_count.
598 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
600 * ipa-inline.c (compute_uninlined_call_time, compute_inlined_call_time):
601 always use frequencies.
603 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
605 * bb-reorder.c: Remove frequencies from comments.
606 (better_edge_p): Use profile counts.
607 (find_traces): Dump profile counts.
608 (rotate_loop): Use profile counts.
609 (find_traces_1_round): Likewise.
610 (connect_better_edge_p): Use counts instead of probabilities for
612 (copy_bb_p): Drop early check for non-0 frequency.
613 (sanitize_hot_paths): Update comments.
615 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
617 * ipa-split.c (struct split_point): Add count.
618 (consider_split): Do not compute incoming frequency; compute incoming
619 count and store it to split_point.
620 (split_function): Set count of the call to split part correctly.
622 2017-11-13 Carl Love <cel@us.ibm.com>
624 * config/rs6000/altivec.md (altivec_vsumsws_be): Add define_expand.
626 2017-11-13 Tom Tromey <tom@tromey.com>
628 * doc/cpp.texi (Variadic Macros): Document __VA_OPT__.
630 2017-11-13 Carl Love <cel@us.ibm.com>
632 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins):
633 Add support for builtins:
634 unsigned int vec_first_{,miss}_match_{,or_eos}index,
635 vector {un,}signed {char,int,short},
636 vector {un,}signed {char,int,short}) arguments.
637 * config/rs6000/rs6000-builtin.def (VFIRSTMATCHINDEX,
638 VFIRSTMATCHOREOSINDEX, VFIRSTMISMATCHINDEX, VFIRSTMISMATCHOREOSINDEX):
639 Add BU_P9V_AV_2 expansions for the builtins.
640 * config/rs6000/altivec.h (vec_first_match_index,
641 vec_first_mismatch_index, vec_first_match_or_eos_index,
642 vec_first_mismatch_or_eos_index): Add #defines for the builtins.
643 * config/rs6000/rs6000-protos.h (bytes_in_mode): Add
644 new extern declaration.
645 * config/rs6000/rs6000.c (bytes_in_mode): Add new function.
646 * config/rs6000/vsx.md (first_match_index_<mode>,
647 first_match_or_eos_index_<mode>, first_mismatch_index_<mode>,
648 first_mismatch_or_eos_index_<mode>): Add define expand.
649 (vctzlsbb_<mode>): Add mode field to define_insn for vctzlsbb.
650 * doc/extend.texi: Update the built-in documenation file for the new
653 2017-11-13 Michael Meissner <meissner@linux.vnet.ibm.com>
655 * match.pd: Convert fminf<N>, fminf<N>x, fmax<N>, and fmax<N>x
656 into the min/max operations for _Float<N> and _Float<N>X types.
658 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
661 * dwarf2out.c (do_eh_frame): New static variable.
662 (dwarf2out_begin_prologue): Set it.
663 (dwarf2out_frame_finish): Test it instead of dwarf2out_do_eh_frame.
665 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
667 * tree-ssa-coalesce.c (coalesce_cost): Fix formating.
669 * tree-ssa-sink.c (select_best_block): Do not use frequencies.
671 2017-11-13 Eric Botcazou <ebotcazou@adacore.com>
674 * debug.h (dwarf2out_do_eh_frame): Declare.
675 * dwarf2cfi.c (dwarf2out_do_eh_frame): New predicate.
676 (dwarf2out_do_frame): Use it.
677 (dwarf2out_do_cfi_asm): Likewise.
678 * dwarf2out.c (dwarf2out_frame_finish): Likewise.
679 (dwarf2out_assembly_start): Likewise.
680 (dwarf2out_begin_prologue): Fix comment.
681 * toplev.c (compile_file): Always call dwarf2out_frame_finish
682 if the target needs either debug or unwind DWARF2 info.
683 * lto-opts.c (lto_write_options): Do not save -fexceptions,
684 -fnon-call-exceptions, -ffp-contract, -fmath-errno, -fsigned-zeros,
685 -ftrapping-math, -ftrapv and -fwrapv.
687 2017-11-13 Jan Hubicka <hubicka@ucw.cz>
689 * cgraph.c (cgraph_edge::sreal_frequency): New function.
690 * cgraph.h (cgraph_edge::sreal_frequency): Declare.
691 * ipa-fnsummary.c (dump_ipa_call_summary): Use sreal_frequency.
692 (estimate_edge_size_and_time): Likewise.
693 (ipa_merge_fn_summary_after_inlining): Likewise.
694 * ipa-inline.c (cgraph_freq_base_rec): Remove.
695 (compute_uninlined_call_time): Use sreal_frequency.
696 (compute_inlined_call_time): Likewise.
697 (ipa_inline): Do not initialize cgraph_freq_base_rec.
698 * profile-count.c: Include sreal.h.
699 (profile_count::to_sreal_scale): New.
700 * profile-count.h: Forward declare sreal.
701 (profile_count::to_sreal_scale): Declare.
703 2017-11-13 Nathan Sidwell <nathan@acm.org>
705 * diagnostic.c (maybe_line_and_column): New.
706 (diagnostic_get_location_text): Use it.
707 (diagnostic_report_current_module): Likewise.
708 (test_diagnostic_get_location_text): Add tests.
710 2017-11-13 Luis Machado <luis.machado@linaro.org>
712 * doc/md.texi (Specifying processor pipeline description): Fix
713 incorrect latency for the div instruction example.
715 2017-11-13 Jakub Jelinek <jakub@redhat.com>
717 PR tree-optimization/78821
718 * gimple-ssa-store-merging.c (compatible_load_p): Don't require
719 that bit_not_p is the same.
720 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
721 (split_group): Count precisely bit_not_p bits in each statement.
722 (invert_op): New function.
723 (imm_store_chain_info::output_merged_store): Use invert_op to
724 emit BIT_XOR_EXPR with a xor_mask instead of BIT_NOT_EXPR if some
725 but not all orig_stores have BIT_NOT_EXPR in the corresponding spots.
727 2017-11-13 Martin Liska <mliska@suse.cz>
729 * gcov.c (struct coverage_info): Remove typedef of coverage_t.
730 (struct source_info): Likewise.
731 (add_branch_counts): Likewise.
732 (add_line_counts): Likewise.
733 (function_summary): Likewise.
734 (output_intermediate_line): Likewise.
735 (generate_results): Likewise.
737 2017-11-13 Martin Liska <mliska@suse.cz>
739 * gcov.c (struct block_info): Remove typedef for block_t.
740 (struct line_info): Likewise.
741 (line_info::has_block): Likewise.
742 (EXIT_BLOCK): Likewise.
745 (get_cycles_count): Likewise.
746 (process_file): Likewise.
747 (read_graph_file): Likewise.
748 (solve_flow_graph): Likewise.
749 (find_exception_blocks): Likewise.
750 (add_line_counts): Likewise.
751 (accumulate_line_info): Likewise.
752 (output_line_details): Likewise.
754 2017-11-13 Martin Liska <mliska@suse.cz>
756 * gcov.c (struct arc_info): Remove typedef for arc_t.
757 (struct line_info): Likewise.
758 (add_branch_counts): Likewise.
759 (output_branch_count): Likewise.
760 (function_info::~function_info): Likewise.
762 (output_intermediate_line): Likewise.
763 (read_graph_file): Likewise.
764 (solve_flow_graph): Likewise.
765 (find_exception_blocks): Likewise.
766 (add_line_counts): Likewise.
767 (accumulate_line_info): Likewise.
768 (output_line_details): Likewise.
769 (output_function_details): Likewise.
771 2017-11-13 Martin Liska <mliska@suse.cz>
773 * gcov.c (struct function_info): Remove typedef for function_t.
774 (struct source_info): Likewise.
775 (source_info::get_functions_at_location): Likewise.
776 (solve_flow_graph): Likewise.
777 (find_exception_blocks): Likewise.
778 (add_line_counts): Likewise.
779 (output_intermediate_file): Likewise.
780 (process_file): Likewise.
781 (generate_results): Likewise.
782 (release_structures): Likewise.
783 (read_graph_file): Likewise.
784 (read_count_file): Likewise.
785 (accumulate_line_counts): Likewise.
786 (output_lines): Likewise.
788 2017-11-13 Martin Liska <mliska@suse.cz>
790 * gcov.c (function_info::function_info): Remove num_counts
791 and add vector<gcov_type>.
792 (function_info::~function_info): Use the vector.
793 (process_file): Likewise.
794 (read_graph_file): Likewise.
795 (read_count_file): Likewise.
796 (solve_flow_graph): Likewise.
798 2017-11-13 Martin Liska <mliska@suse.cz>
800 * gcov.c (function_info::is_artificial): New function.
801 (process_file): Erase all artificial early.
802 (generate_results): Skip as all artificial are already
805 2017-11-13 Martin Liska <mliska@suse.cz>
807 * gcov.c (read_graph_file): Store to global vector of functions.
808 (read_count_file): Iterate the vector.
809 (process_file): Likewise.
810 (generate_results): Likewise.
811 (release_structures): Likewise.
813 2017-11-13 Jakub Jelinek <jakub@redhat.com>
815 PR tree-optimization/82954
816 * gimple-ssa-store-merging.c
817 (imm_store_chain_info::coalesce_immediate_stores): If
818 !infof->ops[N].base_addr, split group if info->ops[N].base_addr.
820 2017-11-13 Richard Sandiford <richard.sandiford@linaro.org>
822 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
823 Upddate call to ENDIAN_LANE_N.
824 (aarch64_<sur>dot_lane<vsi2qi>): Use aarch64_endian_lane_rtx.
825 (aarch64_<sur>dot_laneq<vsi2qi>): Likewise.
826 (*aarch64_simd_vec_copy_lane<mode>): Update calls to ENDIAN_LANE_N
827 and use aarch64_endian_lane_rtx.
828 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Likewise.
830 2017-11-12 Tom de Vries <tom@codesourcery.com>
832 * config/riscv/riscv.h (ASM_OUTPUT_LABELREF): Wrap in do {} while (0).
834 2017-11-12 Tom de Vries <tom@codesourcery.com>
836 * config/elfos.h (ASM_OUTPUT_ASCII): Remove semicolon after macro body.
838 2017-11-12 Tom de Vries <tom@codesourcery.com>
840 * config/cr16/cr16.h (ASM_OUTPUT_LABELREF): Remove semicolon after macro
842 * config/powerpcspe/xcoff.h (ASM_OUTPUT_LABELREF): Same.
843 * config/rs6000/xcoff.h (ASM_OUTPUT_LABELREF): Same.
844 * defaults.h (ASM_OUTPUT_LABELREF): Same.
846 2017-11-11 Martin Sebor <msebor@redhat.com>
849 * doc/extend.texi (attribute nonstring): Remove spurious argument.
852 * prefic.c (translate_name): Replace strncpy with memcpy to
853 avoid -Wstringop-truncation.
855 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
857 * tree-ssa-loop-im.c (execute_sm_if_changed): Do not compute freq_sum.
859 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
861 * predict.c (maybe_hot_frequency_p): Do not use cfun.
863 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
865 * tree-ssa-tail-merge.c (replace_block_by): Fix and re-enable profile
868 2017-11-10 Michael Meissner <meissner@linux.vnet.ibm.com>
870 * config/rs6000/rs6000.md (bswaphi2_reg): On ISA 3.0 systems,
871 enable generating XXBRH if the value is in a vector register.
872 (bswapsi2_reg): On ISA 3.0 systems, enable generating XXBRW if the
873 value is in a vector register.
874 (bswapdi2_reg): On ISA 3.0 systems, always use XXBRD to do
875 register to register bswap64's instead of doing the GPR sequence
876 used on previous machines.
877 (bswapdi2_xxbrd): New insn.
878 (bswapdi2_reg): Disallow on ISA 3.0.
879 (register to register bswap64 splitter): Do not split the insn on
880 ISA 3.0 systems that use XXBRD.
882 2017-11-10 Martin Sebor <msebor@redhat.com>
885 * config/darwin-c.c (framework_construct_pathname): Replace strncpy
887 (find_subframework_file): Same.
889 2017-11-10 Jan Hubicka <hubicka@ucw.cz>
891 * auto-profile.c (afdo_indirect_call): Drop frequency.
892 * cgraph.c (symbol_table::create_edge): Drop frequency argument.
893 (cgraph_node::create_edge): Drop frequency argument.
894 (cgraph_node::create_indirect_edge): Drop frequency argument.
895 (cgraph_edge::make_speculative): Drop frequency arguments.
896 (cgraph_edge::resolve_speculation): Do not update frequencies
897 (cgraph_edge::dump_edge_flags): Do not dump frequency.
898 (cgraph_node::dump): Check consistency in IPA mode.
899 (cgraph_edge::maybe_hot_p): Use IPA counter.
900 (cgraph_edge::verify_count_and_frequency): Rename to ...
901 (cgraph_edge::verify_count): ... this one; drop frequency checking.
902 (cgraph_node::verify_node): Update.
903 * cgraph.h (struct cgraph_edge): Drop frequency.
904 (cgraph_edge::frequency): New function.
905 * cgraphbuild.c (pass_build_cgraph_edges::execute): Donot pass
907 (cgraph_edge::rebuild_edges): Likewise.
908 * cgraphclones.c (cgraph_edge::clone): Scale only counts.
909 (duplicate_thunk_for_node): Do not pass frequency.
910 (cgraph_node::create_clone): Scale only counts.
911 (cgraph_node::create_virtual_clone): Do not pass frequency.
912 (cgraph_node::create_edge_including_clones): Do not pass frequency.
913 (cgraph_node::create_version_clone): Do not pass frequency.
914 * cgraphunit.c (cgraph_node::analyze): Do not pass frequency.
915 (cgraph_node::expand_thunk): Do not pass frequency.
916 (cgraph_node::create_wrapper): Do not pass frequency.
917 * gimple-iterator.c (update_call_edge_frequencies): Do not pass
919 * gimple-streamer-in.c (input_bb): Scale only IPA counts.
920 * ipa-chkp.c (chkp_produce_thunks): Do not pass frequency.
921 * ipa-cp.c (ipcp_lattice::print): Use frequency function.
922 (gather_caller_stats): Use frequency function.
923 (ipcp_cloning_candidate_p): Use frequency function.
924 (ipcp_propagate_stage): Use frequency function.
925 (get_info_about_necessary_edges): Use frequency function.
926 (update_profiling_info): Update only IPA profile.
927 (update_specialized_profile): Use frequency functoin.
928 (perhaps_add_new_callers): Update only IPA profile.
929 * ipa-devirt.c (ipa_devirt): Use IPA profile.
930 * ipa-fnsummary.c (redirect_to_unreachable): Do not set frequrency.
931 (dump_ipa_call_summary): Use frequency function.
932 (estimate_edge_size_and_time): Use frequency function.
933 (ipa_merge_fn_summary_after_inlining): Use frequency function.
934 * ipa-inline-analysis.c (do_estimate_edge_time): Use IPA profile.
935 * ipa-inline-transform.c (update_noncloned_frequencies): Rename to ..
936 (update_noncloned_counts): ... ths one; scale counts only.
937 (clone_inlined_nodes): Do not scale frequency.
938 (inline_call): Do not pass frequency.
939 * ipa-inline.c (compute_uninlined_call_time): Use IPA profile.
940 (compute_inlined_call_time): Use IPA profile.
941 (want_inline_small_function_p): Use IPA profile.
942 (want_inline_self_recursive_call_p): Use IPA profile.
943 (edge_badness): Use IPA profile.
944 (lookup_recursive_calls): Use IPA profile.
945 (recursive_inlining): Do not pass frequency.
946 (resolve_noninline_speculation): Do not update frequency.
947 (inline_small_functions): Collect max of IPA profile.
948 (dump_overall_stats): Dump IPA porfile.
949 (dump_inline_stats): Dump IPA porfile.
950 (ipa_inline): Collect IPA stats.
951 * ipa-inline.h (clone_inlined_nodes): Update prototype.
952 * ipa-profile.c (ipa_propagate_frequency_1): Use frequency function.
953 (ipa_propagate_frequency): Use frequency function.
954 (ipa_profile): Cleanup.
955 * ipa-prop.c (ipa_make_edge_direct_to_target): Do not pass frequency
956 * ipa-utils.c (ipa_merge_profiles): Merge all profiles.
957 * lto-cgraph.c (lto_output_edge): Do not stream frequency.
958 (input_node): Do not stream frequency.
959 (input_edge): Do not stream frequency.
960 (merge_profile_summaries): Scale only IPA profiles.
961 * omp-simd-clone.c (simd_clone_adjust): Do not pass frequency.
962 * predict.c (drop_profile): Do not recompute frequency.
963 * trans-mem.c (ipa_tm_insert_irr_call): Do not pass frequency.
964 (ipa_tm_insert_gettmclone_call): Do not pass frequency.
965 * tree-cfg.c (execute_fixup_cfg): Drop profile to global0 if needed.
966 * tree-chkp.c (chkp_copy_bounds_for_assign): Do not pass frequency.
967 * tree-emutls.c (gen_emutls_addr): Do not pass frequency.
968 * tree-inline.c (copy_bb): Do not scale frequency.
969 (expand_call_inline): Do not scale frequency.
970 (tree_function_versioning): Do not scale frequency.
971 * ubsan.c (ubsan_create_edge): Do not pass frequency.
973 2017-11-10 Julia Koval <julia.koval@intel.com>
975 * config/i386/gfniintrin.h (_mm_gf2p8affine_epi64_epi8)
976 (_mm256_gf2p8affine_epi64_epi8, _mm_mask_gf2p8affine_epi64_epi8)
977 (_mm_maskz_gf2p8affine_epi64_epi8, _mm256_mask_gf2p8affine_epi64_epi8)
978 (_mm256_maskz_gf2p8affine_epi64_epi8)
979 (_mm512_mask_gf2p8affine_epi64_epi8, _mm512_gf2p8affine_epi64_epi8)
980 (_mm512_maskz_gf2p8affine_epi64_epi8): New intrinsics.
981 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineqb_v64qi)
982 (__builtin_ia32_vgf2p8affineqb_v32qi)
983 (__builtin_ia32_vgf2p8affineqb_v16qi): New builtins.
984 * config/i386/sse.md (vgf2p8affineqb_<mode><mask_name>): New pattern.
986 2017-11-10 Tamar Christina <tamar.christina@arm.com>
990 (arm_option_override): Refactor.
991 (arm_option_reconfigure_globals): New.
992 (arm_options_perform_arch_sanity_checks): New.
993 * config/arm/arm-protos.h (arm_option_reconfigure_globals):
995 (arm_options_perform_arch_sanity_checks): Likewise
997 2017-11-10 Pat Haugen <pthaugen@us.ibm.com>
999 * rs6000/power9.md (power9-qpdiv): Correct DFU pipe usage.
1000 (power9-qpmul): New.
1001 * rs6000/rs6000.md ("type" attr): Add qmul.
1002 (mul<mode>3, fma<mode>4_hw, *fms<mode>4_hw, *nfma<mode>4_hw,
1003 *nfms<mode>4_hw, mul<mode>3_odd, fma<mode>4_odd, *fms<mode>4_odd,
1004 *nfma<mode>4_odd, *nfms<mode>4_odd): Change type to qmul.
1006 2017-11-10 Martin Sebor <msebor@redhat.com>
1009 * builtins.c (compute_objsize): Handle arrays that
1010 compute_builtin_object_size likes to fail for. Make extern.
1011 * builtins.h (compute_objsize): Declare.
1012 (check_strncpy_sizes): New function.
1013 (expand_builtin_strncpy): Call check_strncpy_sizes.
1014 * gimple-fold.c (gimple_fold_builtin_strncpy): Implement
1015 -Wstringop-truncation.
1016 (gimple_fold_builtin_strncat): Same.
1017 * gimple.c (gimple_build_call_from_tree): Set call location.
1018 * tree-ssa-strlen.c (strlen_to_stridx): New global variable.
1019 (maybe_diag_bound_equal_length, is_strlen_related_p): New functions.
1020 (handle_builtin_stxncpy, handle_builtin_strncat): Same.
1021 (handle_builtin_strlen): Use strlen_to_stridx.
1022 (strlen_optimize_stmt): Handle flavors of strncat, strncpy, and
1024 Use strlen_to_stridx.
1025 (pass_strlen::execute): Release strlen_to_stridx.
1026 * doc/invoke.texi (-Wsizeof-pointer-memaccess): Document enhancement.
1027 (-Wstringop-truncation): Document new option.
1029 2017-11-10 Martin Liska <mliska@suse.cz>
1031 PR gcov-profile/82702
1032 * gcov.c (main): Handle intermediate files in a different
1034 (get_gcov_intermediate_filename): New function.
1035 (output_gcov_file): Remove support of intermediate files.
1036 (generate_results): Allocate intermediate file.
1037 (release_structures): Clean-up properly fn_end.
1038 (output_intermediate_file): Start iterating with line 1.
1040 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1042 PR tree-optimization/82929
1043 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1044 ops_swapped_p non-static data member.
1045 (store_immediate_info::store_immediate_info): Clear it.
1046 (imm_store_chain_info::coalesce_immediate_stores): If swapping
1047 ops set ops_swapped_p.
1048 (count_multiple_uses): Handle ops_swapped_p.
1050 2017-11-10 Martin Liska <mliska@suse.cz>
1052 * coverage.c (coverage_init): Stream information about
1053 support of has_unexecuted_blocks.
1054 * doc/gcov.texi: Document that.
1055 * gcov-dump.c (dump_gcov_file): Support it in gcov_dump tool.
1056 * gcov.c (read_graph_file): Likewise.
1057 (output_line_beginning): Fix a small issue with
1060 2017-11-10 Bin Cheng <bin.cheng@arm.com>
1062 * tree-predcom.c (determine_roots_comp): Avoid memory leak by freeing
1063 reference of trivial component.
1065 2017-11-10 Jakub Jelinek <jakub@redhat.com>
1068 * gimple-ssa-store-merging.c
1069 (pass_store_merging::terminate_all_aliasing_chains): For
1070 gimple_store_p stmts also call refs_output_dependent_p.
1072 PR rtl-optimization/82913
1073 * compare-elim.c (try_merge_compare): Punt if def_insn is not
1076 2017-11-09 Jeff Law <law@redhat.com>
1078 * vr-values.h: New file with vr_values class.
1079 * tree-vrp.c: Include vr-values.h
1080 (vrp_value_range_pool, vrp_equiv_obstack, num_vr_values): Move static
1081 data objects into the vr_values class.
1082 (vr_value, values_propagated, vr_phi_edge_counts): Likewise.
1083 (get_value_range): Make it a member function within vr_values class.
1084 (set_defs_to_varying, update_value_range, add_equivalence): Likewise.
1085 (vrp_stmt_computes_nonzero_p, op_with_boolean_value_range_p): Likewise.
1086 (op_with_constant_singleton_value_range): Likewise.
1087 (extract_range_for_var_from_comparison_expr): Likewise.
1088 (extract_range_from_assert, extract_range_from_ssa_name): Likewise.
1089 (extract_range_from_binary_expr): Likewise.
1090 (extract_range_from_unary_expr): Likewise.
1091 (extract_range_from_cond_expr, extrat_range_from_comparison): Likewise.
1092 (check_for_binary_op_overflow, extract_range_basic): Likewise.
1093 (extract_range_from_assignment, adjust_range_with_scev): Likewise.
1094 (dump_all_value_ranges, get_vr_for_comparison): Likewise.
1095 (compare_name_with_value, compare_names): Likewise.
1096 (vrp_evaluate_conditional_warnv_with_ops_using_ranges): Likewise.
1097 (vrp_evaluate_conditional_warnv_with_ops): Likewise. Remove prototype.
1098 (vrp_evaluate_conditional, vrp_visit_cond_stmt): Likewise.
1099 (vrp_visit_switch_stmt, extract_range_from_stmt): Likewise.
1100 (extract_range_from_phi_node): Likewise.
1101 (simplify_truth_ops_using_ranges): Likewise.
1102 (simplify_div_or_mod_using_ranges): Likewise.
1103 (simplify_min_or_max_using_ranges, simplify_abs_using_ranges): Likewise.
1104 (simplify_bit_ops_using_ranges, simplify_cond_using_ranges_1): Likewise.
1105 (simplify_cond_using_ranges_2, simplify_switch_using_ranges): Likewise.
1106 (simplify_float_conversion_using_ranges): Likewise.
1107 (simplify_internal_call_using_ranges): Likewise.
1108 (two_valued_val_range_p, simplify_stmt_using_ranges): Likewise.
1109 (vrp_visit_assignment_or_call): Likewise. Smuggle class instance
1110 poitner via x_vr_values for calls into gimple folder.
1111 (vrp_initialize_lattice): Make this the vr_values ctor.
1112 (vrp_free_lattice): Make this the vr_values dtor.
1113 (set_vr_value): New function.
1114 (class vrp_prop): Add vr_values data member. Add various member
1115 functions as well as member functions that delegate to vr_values.
1116 (check_array_ref): Make a member function within vrp_prop class.
1117 (search_for_addr_array, vrp_initialize): Likewise.
1118 (vrp_finalize): Likewise. Revamp to avoid direct access to
1119 vr_value, values_propagated, etc.
1120 (check_array_bounds): Extract vrp_prop class instance pointer from
1121 walk info structure. Use it to call member functions.
1122 (check_all_array_refs): Make a member function within vrp_prop class.
1123 Smuggle class instance pointer via walk info structure.
1124 (x_vr_values): New local static.
1125 (vrp_valueize): Use x_vr_values to get class instance.
1126 (vr_valueize_1): Likewise.
1127 (class vrp_folder): Add vr_values data member. Add various member
1128 functions as well as member functions that delegate to vr_values.
1129 (fold_predicate_in): Make a mber fucntion within vrp_folder class.
1130 (simplify_stmt_for_jump_threading): Extract smuggled vr_values
1131 class instance from vr_values. Use it to call member functions.
1132 (vrp_dom_walker): Add vr_values data member.
1133 (vrp_dom_walker::after_dom_children): Smuggle vr_values class
1134 instance via x_vr_values.
1135 (identify_jump_threads): Accept vr_values as argument. Store
1136 it into the walker structure.
1137 (evrp_dom_walker): Add vr_values class data member. Add various
1139 (evrp_dom_walker::try_find_new_range): Use vr_values data
1140 member to access the memory allocator.
1141 (evrp_dom_walker::before_dom_children): Store vr_values class
1142 instance into the vrp_folder class.
1143 (evrp_dom_walker::push_value_range): Rework to avoid direct
1144 access to num_vr_values and vr_value.
1145 (evrp_dom_walker::pop_value_range): Likewise.
1146 (execute_early_vrp): Remove call to vrp_initialize_lattice.
1147 Use vr_values to get to dump_all_value_ranges member function.
1148 Remove call to vrp_free_lattice. Call vrp_initialize, vrp_finalize,
1149 and simplify_cond_using_ranges_2 via vrp_prop class instance.
1150 Pass vr_values class instance down to identify_jump_threads.
1151 Remove call to vrp_free_lattice.
1152 (debug_all_value_ranges): Remove.
1154 * tree-vrp.c (vrp_prop): Move class to earlier point in the file.
1155 (vrp_folder): Likewise.
1157 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
1158 Get it from the existing bitmap instead.
1159 (vrp_intersect_ranges_1): Likewise.
1161 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1163 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
1165 (store_immediate_info::store_immediate_info): Add bitnotp argument,
1166 set bit_not_p to it.
1167 (imm_store_chain_info::coalesce_immediate_stores): Break group
1168 if bit_not_p is different.
1169 (count_multiple_uses, split_group,
1170 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
1171 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
1172 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
1173 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
1174 to store_immediate_info ctor.
1176 2017-11-09 Jim Wilson <jimw@sifive.com>
1178 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
1179 (scan_prog_file): Likewise.
1181 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1183 * bb-reorder.c (max_entry_frequency): Remove.
1184 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
1185 connect_traces, push_to_next_round_p): Remove prototypes.
1186 (find_traces_1_round): Use counts only.
1187 (push_to_next_round_p): Likewise.
1188 (find_traces): Likewise.
1189 (rotate_loop): Likewise.
1190 (find_traces_1_round): Likewise.
1191 (connect_traces): Likewise.
1192 (edge_order): Likewise.
1194 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
1196 * config/arm/arm.c (output_return_instruction): Add comments to
1197 indicate requirement for cmse_nonsecure_entry return to account
1198 for the size of clearing instruction output here.
1199 (thumb_exit): Likewise.
1200 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
1201 return in hardfloat mode.
1203 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
1205 * config/rs6000/rs6000.c (machine_function): Add a bool,
1206 "toc_is_wrapped_separately".
1207 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
1208 if it wasn't explicitly set or unset, we are optimizing for speed, and
1209 doing separate shrink-wrapping.
1210 (rs6000_get_separate_components): Enable the TOC component if
1211 saving the TOC register in the prologue.
1212 (rs6000_components_for_bb): Handle the TOC component.
1213 (rs6000_emit_prologue_components): Store the TOC register where needed.
1214 (rs6000_set_handled_components): Mark TOC as handled, if handled.
1215 (rs6000_emit_prologue): Don't save the TOC if that is already done.
1217 2017-11-09 Martin Jambor <mjambor@suse.cz>
1219 * ipa-param-manipulation.c: New file.
1220 * ipa-param-manipulation.h: Likewise.
1221 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
1222 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
1223 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
1224 (ipa_parm_adjustment): Likewise.
1225 (ipa_parm_adjustment_vec): Likewise.
1226 (ipa_get_vector_of_formal_parms): Moved declaration to
1227 ipa-param-manipulation.h.
1228 (ipa_get_vector_of_formal_parm_types): Likewise.
1229 (ipa_modify_formal_parameters): Likewise.
1230 (ipa_modify_call_arguments): Likewise.
1231 (ipa_combine_adjustments): Likewise.
1232 (ipa_dump_param_adjustments): Likewise.
1233 (ipa_modify_expr): Likewise.
1234 (ipa_get_adjustment_candidate): Likewise.
1235 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
1236 ipa-param-manipulation.c.
1237 (ipa_get_vector_of_formal_parm_types): Likewise.
1238 (ipa_modify_formal_parameters): Likewise.
1239 (ipa_modify_call_arguments): Likewise.
1240 (ipa_modify_expr): Likewise.
1241 (get_ssa_base_param): Likewise.
1242 (ipa_get_adjustment_candidate): Likewise.
1243 (index_in_adjustments_multiple_times_p): Likewise.
1244 (ipa_combine_adjustments): Likewise.
1245 (ipa_dump_param_adjustments): Likewise.
1246 * tree-sra.c: Also include ipa-param-manipulation.h
1247 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
1250 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1251 Alan Hayward <alan.hayward@arm.com>
1252 David Sherwood <david.sherwood@arm.com>
1254 * doc/sourcebuild.texi (vect_masked_store): Document.
1256 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1257 Alan Hayward <alan.hayward@arm.com>
1258 David Sherwood <david.sherwood@arm.com>
1260 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
1262 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1263 Alan Hayward <alan.hayward@arm.com>
1264 David Sherwood <david.sherwood@arm.com>
1266 * doc/sourcebuild.texi (vect_variable_length): Document.
1268 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1269 Alan Hayward <alan.hayward@arm.com>
1270 David Sherwood <david.sherwood@arm.com>
1272 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
1274 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1275 Alan Hayward <alan.hayward@arm.com>
1276 David Sherwood <david.sherwood@arm.com>
1278 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
1280 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1281 Alan Hayward <alan.hayward@arm.com>
1282 David Sherwood <david.sherwood@arm.com>
1284 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
1285 previously undocumented selectors.
1286 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
1288 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1290 * doc/rtl.texi (const_vector): Say that elements can be
1291 const_wide_ints too.
1292 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
1293 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
1294 (gen_vec_duplicate): Use it instead of CONSTANT_P.
1295 * optabs.c (expand_vector_broadcast): Likewise.
1297 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1298 Alan Hayward <alan.hayward@arm.com>
1299 David Sherwood <david.sherwood@arm.com>
1301 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
1302 scaled index even if the unscaled address was invalid.
1303 Don't increase the complexity of using a scale in that case.
1305 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
1306 Alan Hayward <alan.hayward@arm.com>
1307 David Sherwood <david.sherwood@arm.com>
1309 * doc/rtl.texi: Rewrite the subreg rules so that they partition
1310 the inner register into REGMODE_NATURAL_SIZE bytes rather than
1311 UNITS_PER_WORD bytes.
1312 * emit-rtl.c (validate_subreg): Divide subregs into blocks
1313 based on REGMODE_NATURAL_SIZE of the inner mode.
1314 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
1315 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
1316 * expmed.c (lowpart_bit_field_p): Divide the value up into
1317 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
1318 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
1319 whether something is likely to occupy more than one register.
1321 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
1324 * ipa-inline-transform.c (update_noncloned_frequencies): Use
1325 profile_count::adjust_for_ipa_scaling.
1326 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
1327 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
1329 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
1331 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1333 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
1334 (split_group): Add total_orig and total_new arguments, estimate the
1335 number of statements related to the store group without store merging
1336 and with store merging.
1337 (imm_store_chain_info::output_merged_store): Adjust split_group
1338 callers, punt if estimated number of statements with store merging
1339 is not smaller than estimated number of statements without it.
1341 (handled_load): Remove has_single_use checks.
1342 (pass_store_merging::process_store): Likewise.
1344 2017-11-09 Richard Biener <rguenther@suse.de>
1346 PR tree-optimization/82902
1347 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
1349 2017-11-09 Martin Liska <mliska@suse.cz>
1352 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
1355 2017-11-09 Martin Liska <mliska@suse.cz>
1357 PR tree-optimization/82669
1358 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
1360 2017-11-09 Martin Liska <mliska@suse.cz>
1362 PR gcov-profile/48463
1363 * coverage.c (coverage_begin_function): Output also end locus
1364 of a function and information whether the function is
1366 * gcov-dump.c (tag_function): Parse and print the information.
1367 * gcov.c (INCLUDE_MAP): Add include.
1368 (INCLUDE_SET): Likewise.
1369 (struct line_info): Move earlier in the source file because
1370 of vector<line_info> in function_info structure.
1371 (line_info::line_info): Likewise.
1372 (line_info::has_block): Likewise.
1373 (struct source_info): Add new member index.
1374 (source_info::get_functions_at_location): New function.
1375 (function_info::group_line_p): New function.
1376 (output_intermediate_line): New function.
1377 (output_intermediate_file): Use the mentioned function.
1378 (struct function_start): New.
1379 (struct function_start_pair_hash): Likewise.
1380 (process_file): Add code that identifies group functions.
1381 Assign lines either to global or function scope.
1382 (generate_results): Skip artificial functions.
1383 (find_source): Assign index for each source file.
1384 (read_graph_file): Read new flag artificial and end_line.
1385 (add_line_counts): Assign it either to global of function scope.
1386 (accumulate_line_counts): Isolate core of the function to
1387 accumulate_line_info and call it for both function and global
1389 (accumulate_line_info): New function.
1390 (output_line_beginning): Fix GNU coding style.
1391 (print_source_line): New function.
1392 (output_line_details): Likewise.
1393 (output_function_details): Likewise.
1394 (output_lines): Iterate both source (global) scope and function
1396 (struct function_line_start_cmp): New class.
1397 * doc/gcov.texi: Reflect changes in documentation.
1399 2017-11-09 Jakub Jelinek <jakub@redhat.com>
1402 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
1403 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
1404 and similarly for not instead of neg.
1406 2017-11-08 Andi Kleen <ak@linux.intel.com>
1408 * config/i386/i386.opt: Add -mforce-indirect-call.
1409 * config/i386/predicates.md: Check for flag_force_indirect_call.
1410 * doc/invoke.texi: Document -mforce-indirect-call
1412 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
1414 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
1416 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
1417 config/riscv/riscv.c (predict.h): New include.
1418 (riscv_slow_unaligned_access_p): No longer static.
1419 (riscv_block_move_straight): Add require.
1420 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
1422 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1425 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
1426 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
1427 nonimmediate_operand predicate for operand 1 instead of
1430 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1432 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
1434 * config/aarch64/constraints.md (Uml): New constraint.
1435 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
1438 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1440 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
1441 of two vec_duplicates into a vec_concat.
1443 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1445 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1446 Simplify vec_merge of vec_duplicate and vec_concat.
1447 * config/aarch64/constraints.md (Utq): New constraint.
1448 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
1451 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1453 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
1454 Simplify vec_merge of vec_duplicate and const_vector.
1455 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
1457 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
1458 mode iterator. Update predicate on operand 1 to
1459 handle non-const_vec constants. Delete constraints.
1460 (*aarch64_combinez_be<mode>): Likewise for operand 2.
1462 2017-11-08 Jakub Jelinek <jakub@redhat.com>
1464 PR tree-optimization/78821
1465 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
1467 (store_operand_info::store_operand_info): Initialize it to false.
1468 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
1469 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
1470 store in the group, and if chain_info is non-NULL, to ignore altogether
1472 (compatible_load_p): Fail if bit_not_p does not match.
1473 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
1474 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
1475 (pass_store_merging::process_store): Adjust
1476 terminate_all_aliasing_chains calls to pass NULL in all current spots,
1477 call terminate_all_aliasing_chains newly when adding a store into
1478 a chain with non-NULL chain_info.
1480 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1482 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
1484 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
1486 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1488 (aarch64_layout_frame): Initialise emit_frame_chain.
1489 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
1490 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
1492 2017-11-08 Martin Liska <mliska@suse.cz>
1494 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
1495 of gimple_call_internal_p.
1497 2017-11-07 Tom de Vries <tom@codesourcery.com>
1499 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
1501 2017-11-07 Tom de Vries <tom@codesourcery.com>
1503 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
1506 2017-11-08 Martin Liska <mliska@suse.cz>
1509 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
1511 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1513 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
1515 2017-11-07 Jakub Jelinek <jakub@redhat.com>
1518 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
1519 (*cmp<mode>_ccz_1): New insn with $k alternative.
1522 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
1523 RTX_COMM_COMPARE as commutative as well.
1524 (ix86_binary_operator_ok): Formatting fix.
1525 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
1526 *<code><mode>3<mask_name><round_saeonly_name>,
1527 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
1528 *<s>mul<mode>3_highpart<mask_name>,
1529 *vec_widen_umult_even_v16si<mask_name>,
1530 *vec_widen_umult_even_v8si<mask_name>,
1531 *vec_widen_umult_even_v4si<mask_name>,
1532 *vec_widen_smult_even_v16si<mask_name>,
1533 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
1534 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
1535 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
1536 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
1537 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
1538 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
1539 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
1540 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
1541 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
1542 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
1543 ix86_binary_operator_ok. Formatting fixes.
1544 (*<plusminus_insn><mode>3<mask_name><round_name>,
1545 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
1548 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
1550 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
1551 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
1552 eq and ne if TARGET_ISEL.
1553 (cmp): New code_iterator.
1554 (UNS, UNSU_, UNSIK): New code_attrs.
1555 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
1556 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
1558 ("eq<mode>3"): ... this.
1559 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
1561 ("ne<mode>3"): ... this.
1563 2017-11-07 Julia Koval <julia.koval@intel.com>
1566 * common/config/i386/i386-common.c
1567 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
1568 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
1569 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
1570 * config/i386/i386.opt: Ditto.
1571 * config/i386/i386.c (ix86_target_string): Ditto.
1572 (ix86_option_override_internal): Ditto.
1573 (ix86_init_mpx_builtins): Move MPX to args2.
1574 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
1575 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1576 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1577 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1578 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1579 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1580 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
1582 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
1585 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
1586 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
1587 (zero-extendsidi peephole2): Remove peephole.
1589 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
1592 * stor-layout.c: Include attribs.h.
1593 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
1594 explicit lookup of "aligned" attribute.
1596 2017-11-07 Andrew Waterman <andrew@sifive.com>
1598 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
1599 (riscv_expand_block_move): Likewise.
1600 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
1602 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
1603 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
1604 * config/riscv/riscv.c (riscv_block_move_straight): New function.
1605 (riscv_adjust_block_mem): Likewise.
1606 (riscv_block_move_loop): Likewise.
1607 (riscv_expand_block_move): Likewise.
1608 gcc/config/riscv/riscv.md (movmemsi): New pattern.
1610 2017-11-07 Michael Clark <michaeljclark@mac.com>
1612 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
1613 (MUSL_DYNAMIC_LINKER): Likewise.
1615 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
1617 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
1618 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
1621 2017-11-07 Richard Biener <rguenther@suse.de>
1623 * match.pd: Fix build.
1625 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
1626 Jackson Woodruff <jackson.woodruff@arm.com>
1628 PR tree-optimization/71026
1629 * match.pd: Canonicalize negate in division.
1631 2017-11-07 Sudakshina Das <sudi.das@arm.com>
1634 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
1636 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1638 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
1639 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
1641 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
1643 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
1644 non-scalar integral types.
1645 * match.pd (negate_expr_p): Handle MINUS_EXPR.
1646 (-(A-B), -(~A)): New transformations.
1648 2017-11-07 Tom de Vries <tom@codesourcery.com>
1650 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
1651 semicolon after "do {} while (0)".
1652 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1653 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1654 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1655 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1656 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1658 2017-11-07 Tom de Vries <tom@codesourcery.com>
1660 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
1661 after "do {} while (0)".
1662 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1663 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1664 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1665 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1666 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
1668 2017-11-07 Tom de Vries <tom@codesourcery.com>
1671 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
1673 (arm_rtx_costs_internal): Add missing semicolon after
1674 HANDLE_NARROW_SHIFT_ARITH call.
1676 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1678 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
1679 disable isel if it was not set explicitly.
1681 2017-11-06 James Bowman <james.bowman@ftdichip.com>
1683 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
1685 (add_type_attribute) likewise.
1687 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
1689 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
1690 of struct ix86_frame.
1691 (ix86_initial_elimination_offset): Likewise.
1692 (ix86_expand_split_stack_prologue): Likewise.
1694 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
1696 * tree-vrp.h (enum value_range_type): Update stale comment.
1698 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1699 Alan Hayward <alan.hayward@arm.com>
1700 David Sherwood <david.sherwood@arm.com>
1702 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
1703 (aarch64_expand_vec_perm_const): Take the number of units too.
1704 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
1705 (aarch64_expand_vec_perm_const): Likewise.
1706 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
1707 (vec_perm<mode>): Update accordingly.
1709 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1710 Alan Hayward <alan.hayward@arm.com>
1711 David Sherwood <david.sherwood@arm.com>
1713 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
1714 Take the number of units too.
1715 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
1716 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
1717 but check for a vector mode before rather than after the call.
1718 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
1719 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
1720 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
1721 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
1722 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
1723 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
1724 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
1725 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
1726 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
1727 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
1728 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
1729 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
1730 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
1731 (aarch64_sqdmull2_n<mode>): Update accordingly.
1733 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1734 Alan Hayward <alan.hayward@arm.com>
1735 David Sherwood <david.sherwood@arm.com>
1737 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
1738 the number of units too.
1739 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
1740 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
1741 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
1742 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
1743 (vec_store_lanesxi<mode>): Update accordingly.
1745 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1746 Alan Hayward <alan.hayward@arm.com>
1747 David Sherwood <david.sherwood@arm.com>
1749 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
1750 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
1751 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
1752 of units rather than the mode.
1753 * config/aarch64/iterators.md (nunits): New mode attribute.
1754 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
1755 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
1756 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
1757 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
1758 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
1759 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
1760 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
1761 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
1762 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
1763 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
1764 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
1765 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
1766 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
1767 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
1768 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
1769 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
1770 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
1771 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
1772 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
1773 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
1774 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
1775 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
1776 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
1777 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
1778 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
1779 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
1780 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
1781 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
1782 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
1783 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
1784 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
1785 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
1786 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
1787 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
1788 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
1789 (aarch64_simd_vec_setv2di): Likewise.
1791 2017-11-06 Carl Love <cel@us.ibm.com>
1793 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
1795 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
1796 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
1798 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
1800 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
1801 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
1802 Add power 8 macro expansions.
1803 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
1804 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
1805 power 8 instructions. (VSX_XXBR): Add iterator.
1807 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
1809 * config/arm/arm.md (predicable_short_it): Change default to "no",
1810 improve documentation, remove uses that are identical to the default.
1811 (enabled_for_depr_it): Rename to enabled_for_short_it.
1812 * gcc/config/arm/arm-fixed.md (predicable_short_it):
1813 Remove default uses.
1814 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
1815 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
1816 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
1817 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
1819 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1822 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
1823 float128 helper macros, which are no longer used after deleting
1824 the old 'q' built-in functions, and moving the round to odd
1825 built-in functions to being special built-in functions.
1826 (BU_FLOAT128_2): Likewise.
1827 (BU_FLOAT128_1_HW): Likewise.
1828 (BU_FLOAT128_2_HW): Likewise.
1829 (BU_FLOAT128_3_HW): Likewise.
1830 (FABSQ): Delete old 'q' built-in functions.
1831 (COPYSIGNQ): Likewise.
1832 (SQRTF128_ODD): Move round to odd built-in functions to be
1833 special built-in functions, so that we can handle
1834 -mabi=ieeelongdouble.
1835 (TRUNCF128_ODD): Likewise.
1836 (ADDF128_ODD): Likewise.
1837 (SUBF128_ODD): Likewise.
1838 (MULF128_ODD): Likewise.
1839 (DIVF128_ODD): Likewise.
1840 (FMAF128_ODD): Likewise.
1841 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
1842 built-in names to 'f128'.
1843 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
1844 old 'q' built-in functions, as the machine independent code for
1845 'f128' built-in functions handles this.
1846 (rs6000_expand_builtin): Add expansion for float128 round to odd
1847 functions, keying off on -mabi=ieeelongdouble of whether to use
1848 the KFmode or TFmode variant.
1849 (rs6000_init_builtins): Initialize the _Float128 round to odd
1851 * doc/extend.texi (PowerPC Built-in Functions): Document the old
1852 _Float128 'q' built-in functions are now mapped into the new
1853 'f128' built-in functions.
1855 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
1857 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
1858 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
1860 2017-11-06 David Malcolm <dmalcolm@redhat.com>
1863 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
1864 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
1865 * toplev.c: Include "ipa-fnsummary.h".
1866 (toplev::finalize): Call ipa_fnsummary_c_finalize.
1868 2017-11-06 Jakub Jelinek <jakub@redhat.com>
1870 PR tree-optimization/82838
1871 * gimple-ssa-store-merging.c
1872 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
1873 on a separate gimple_seq which is then appended to seq.
1875 2017-11-06 Jeff Law <law@redhat.com>
1878 * config/i386/i386.c (PROBE_INTERVAL): Remove.
1879 (get_probe_interval): New functions.
1880 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
1881 (ix86_adjust_stack_and_probe): Likewise.
1882 (output_adjust_stack_and_probe): Likewise.
1883 (ix86_emit_probe_stack_range): Likewise.
1884 (ix86_expand_prologue): Likewise.
1886 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1888 PR tree-optimization/82816
1889 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
1890 if the modes of the two types are the same.
1891 (convert_plusminus_to_widen): Likewise.
1893 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1895 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
1897 (usadv16qi): New define_expand.
1898 (usadv8hi): New define_expand.
1900 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
1903 * ipa-inline-transform.c (update_noncloned_frequencies): Always
1905 (inline_transform): Likewise.
1906 * predict.c (counts_to_freqs): Remove useless conditional.
1907 * profile-count.h (profile_count::apply_scale): Move sanity check.
1908 * tree-inline.c (copy_bb): Always scale.
1909 (copy_cfg_body): Likewise.
1911 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
1914 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
1916 (*cmp_ite0): Add enabled_for_depr_it attribute.
1917 (*cmp_ite1): Likewise.
1919 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
1921 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
1924 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
1926 * tree-vrp.c (vrp_int_const_binop): Return true on success and
1927 return the value by pointer.
1928 (extract_range_from_multiplicative_op_1): Update accordingly.
1929 Return as soon as an operation fails.
1931 2017-11-05 Tom de Vries <tom@codesourcery.com>
1934 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
1935 (DEF_SANITIZER_BUILTIN): ... here.
1936 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
1937 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
1939 2017-11-05 Tom de Vries <tom@codesourcery.com>
1942 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
1944 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
1945 ASM_OUTPUT_BEFORE_CASE_LABEL call.
1946 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
1948 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1949 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1950 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
1952 2017-11-05 Tom de Vries <tom@codesourcery.com>
1955 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
1958 2017-11-04 Michael Clark <michaeljclark@mac.com>
1960 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
1961 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
1963 (*addsi3_extended): Likewise.
1964 (*addsi3_extended2): Likewise.
1965 (<optab>si3): Likewise.
1966 (<optab>di3): Likewise.
1967 (<optab><mode>3): Likewise.
1968 (<*optabe>si3_internal): Likewise.
1969 (zero_extendqi<SUPERQI:mode>2): Likewise.
1970 (*add<mode>hi3): Likewise.
1971 (*xor<mode>hi3): Likewise.
1972 (<optab>di3): Likewise.
1973 (*<optab>si3_extend): Likewise.
1974 (*sge<u>_<X:mode><GPR:mode>): Likewise.
1975 (*slt<u>_<X:mode><GPR:mode>): Likewise.
1976 (*sle<u>_<X:mode><GPR:mode>): Likewise.
1978 2017-11-04 Andrew Waterman <andrew@sifive.com>
1980 * config/riscv/riscv.c (riscv_option_override): Conditionally set
1981 TARGET_STRICT_ALIGN based upon -mtune argument.
1983 2017-11-04 Andrew Waterman <andrew@sifive.com>
1985 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
1987 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
1989 config/i386/i386.c (choose_basereg): Use optional scratch
1990 register and add assertion.
1991 (x86_emit_outlined_ms2sysv_save): Use scratch register when
1992 needed, and don't allocate stack.
1993 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
1994 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
1995 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
1997 2017-11-03 Jeff Law <law@redhat.com>
1999 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
2000 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
2001 to probe at the start of a noreturn function.
2003 2017-11-03 Jakub Jelinek <jakub@redhat.com>
2005 PR tree-optimization/78821
2006 * gimple-ssa-store-merging.c: Update the file comment.
2007 (MAX_STORE_ALIAS_CHECKS): Define.
2008 (struct store_operand_info): New type.
2009 (store_operand_info::store_operand_info): New constructor.
2010 (struct store_immediate_info): Add rhs_code and ops data members.
2011 (store_immediate_info::store_immediate_info): Add rhscode, op0r
2012 and op1r arguments to the ctor, initialize corresponding data members.
2013 (struct merged_store_group): Add load_align_base and load_align
2015 (merged_store_group::merged_store_group): Initialize them.
2016 (merged_store_group::do_merge): Update them.
2017 (merged_store_group::apply_stores): Pick the constant for
2018 encode_tree_to_bitpos from one of the two operands, or skip
2019 encode_tree_to_bitpos if neither operand is a constant.
2020 (class pass_store_merging): Add process_store method decl. Remove
2021 bool argument from terminate_all_aliasing_chains method decl.
2022 (pass_store_merging::terminate_all_aliasing_chains): Remove
2023 var_offset_p argument and corresponding handling.
2024 (stmts_may_clobber_ref_p): New function.
2025 (compatible_load_p): New function.
2026 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
2027 if there is overlap and rhs_code is not INTEGER_CST. For
2028 non-overlapping stores terminate group if rhs is not mergeable.
2029 (get_alias_type_for_stmts): Change first argument from
2030 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
2031 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
2032 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
2034 (get_location_for_stmts): Change first argument from
2035 auto_vec<gimple *> & to vec<gimple *> &.
2036 (struct split_store): Remove orig_stmts data member, add orig_stores.
2037 (split_store::split_store): Create orig_stores rather than orig_stmts.
2038 (find_constituent_stmts): Renamed to ...
2039 (find_constituent_stores): ... this. Change second argument from
2040 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
2041 to info structures rather than the statements.
2042 (split_group): Rename ALLOW_UNALIGNED argument to
2043 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
2044 it. Adjust find_constituent_stores caller.
2045 (imm_store_chain_info::output_merged_store): Handle rhs_code other
2046 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
2047 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
2048 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
2049 (mem_valid_for_store_merging): New function.
2050 (handled_load): New function.
2051 (pass_store_merging::process_store): New method.
2052 (pass_store_merging::execute): Use process_store method. Adjust
2053 terminate_all_aliasing_chains caller.
2055 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2057 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2058 Return true for more constants, symbols and label references.
2059 (aarch64_valid_floating_const): Remove unused function.
2061 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2064 * config/aarch64/aarch64.c (aarch64_layout_frame):
2065 Undo forcing of LR at bottom of frame.
2067 2017-11-03 Jeff Law <law@redhat.com>
2070 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
2071 for int_registers_saved.
2073 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
2074 extracted from tree-ssa-dom.c.
2075 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
2076 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
2077 (record_equivalences_from_incoming_edge): Add additional argument
2078 to single_pred_edge_ignoring_loop_edges call.
2079 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
2080 (uncprop_dom_walker::before_dom_children): Add additional argument
2081 to single_pred_edge_ignoring_loop_edges call.
2082 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
2083 single_pred_edge_ignoring_loop_edges rather than open coding.
2084 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
2086 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
2088 * match.pd (-(-A)): Rewrite.
2090 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2092 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
2093 (rs6000_emit_int_cmove): New declaration.
2094 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
2095 (rs6000_emit_sISEL): Delete.
2096 (rs6000_emit_int_cmove): Make non-static.
2097 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
2098 instead of rs6000_emit_sISEL.
2100 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
2102 * asan.c (create_cond_insert_point): Maintain profile.
2103 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
2105 * basic-block.h (struct basic_block_def): Remove frequency.
2106 (EDGE_FREQUENCY): Use to_frequency
2107 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
2109 (find_traces): Update to use to_frequency.
2110 (find_traces_1_round): Likewise; use only IPA counts.
2111 (bb_to_key): Likewise.
2112 (connect_traces): Use IPA counts only.
2113 (copy_bb_p): Update to use to_frequency.
2114 (fix_up_crossing_landing_pad): Likewise.
2115 (sanitize_hot_paths): Likewise.
2116 * bt-load.c (basic_block_freq): Likewise.
2117 * cfg.c (init_flow): Set count_max to uninitialized.
2118 (check_bb_profile): Remove frequencies; check counts.
2119 (dump_bb_info): Do not dump frequencies.
2120 (update_bb_profile_for_threading): Update counts only.
2121 (scale_bbs_frequencies_int): Likewise.
2122 (MAX_SAFE_MULTIPLIER): Remove.
2123 (scale_bbs_frequencies_gcov_type): Update counts only.
2124 (scale_bbs_frequencies_profile_count): Update counts only.
2125 (scale_bbs_frequencies): Update counts only.
2126 * cfg.h (struct control_flow_graph): Add count-max.
2127 (update_bb_profile_for_threading): Update prototype.
2128 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
2129 (find_many_sub_basic_blocks): Likewise.
2130 * cfgcleanup.c (try_forward_edges): Likewise.
2131 (try_crossjump_to_edge): Likewise.
2132 * cfgexpand.c (expand_gimple_cond): Likewise.
2133 (expand_gimple_tailcall): Likewise.
2134 (construct_init_block): Likewise.
2135 (construct_exit_block): Likewise.
2136 * cfghooks.c (verify_flow_info): Check consistency of counts.
2137 (dump_bb_for_graph): Do not dump frequencies.
2138 (split_block_1): Do not update frequencies.
2139 (split_edge): Do not update frequencies.
2140 (make_forwarder_block): Do not update frequencies.
2141 (duplicate_block): Do not update frequencies.
2142 (account_profile_record): Do not update frequencies.
2143 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
2144 for global heuristics.
2145 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
2146 (expected_loop_iterations_unbounded): Use counts only.
2147 * cfgloopmanip.c (scale_loop_profile): Simplify.
2148 (create_empty_loop_on_edge): Simplify
2150 (duplicate_loop_to_header_edge): Simplify
2151 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
2152 (update_br_prob_note): Take care of removing note when profile
2154 (relink_block_chain): Do not dump frequency.
2155 (rtl_account_profile_record): Use to_frequency.
2156 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
2157 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
2158 (cgraph_update_edges_for_call_stmt_node): Likewise.
2159 (cgraph_edge::verify_count_and_frequency): Update.
2160 (cgraph_node::verify_node): Temporarily disable frequency verification.
2161 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
2162 to_cgraph_frequency.
2163 (cgraph_edge::rebuild_edges): Convert to ipa counts.
2164 * cgraphunit.c (init_lowered_empty_function): Do not initialize
2166 (cgraph_node::expand_thunk): Update profile.
2167 * except.c (dw2_build_landing_pads): Do not update frequency.
2168 * final.c (compute_alignments): Use to_frequency.
2169 (dump_basic_block_info): Do not dump frequency.
2170 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
2171 (dump_gimple_bb_header): Do not dump frequency.
2172 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
2174 * gimple-streamer-in.c (input_bb): Do not stream frequency.
2175 * gimple-streamer-out.c (output_bb): Do not stream frequency.
2176 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
2177 (init_before_recovery): Do not update frequency.
2178 (sched_create_recovery_edges): Do not update frequency.
2179 * hsa-gen.c (convert_switch_statements): Do not update frequency.
2180 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
2181 (ipa_cp_c_finalize): Set max_count to uninitialized.
2182 * ipa-fnsummary.c (get_minimal_bb): Use counts.
2183 (param_change_prob): Use counts.
2184 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
2186 * ipa-split.c (consider_split): Use to_frequency.
2187 (split_function): Use to_frequency.
2188 * ira-build.c (loop_compare_func): Likewise.
2189 (mark_loops_for_removal): Likewise.
2190 (mark_all_loops_for_removal): Likewise.
2191 * loop-doloop.c (doloop_modify): Do not update frequency.
2192 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
2194 * lto-streamer-in.c (input_function): Update count_max.
2195 * omp-expand.c (expand_omp_taskreg): Update count_max.
2196 * omp-simd-clone.c (simd_clone_adjust): Update profile.
2197 * predict.c (maybe_hot_frequency_p): Use to_frequency.
2198 (maybe_hot_count_p): Use ipa counts only.
2199 (maybe_hot_bb_p): Simplify.
2200 (maybe_hot_edge_p): Simplify.
2201 (probably_never_executed): Do not take frequency argument.
2202 (probably_never_executed_bb_p): Do not pass frequency.
2203 (probably_never_executed_edge_p): Likewise.
2204 (combine_predictions_for_bb): Check that profile is nonzero.
2205 (propagate_freq): Do not set frequency.
2206 (drop_profile): Simplify.
2207 (counts_to_freqs): Simplify.
2208 (expensive_function_p): Use to_frequency.
2209 (propagate_unlikely_bbs_forward): Simplify.
2210 (determine_unlikely_bbs): Simplify.
2211 (estimate_bb_frequencies): Add hack to silence graphite issues.
2212 (compute_function_frequency): Use ipa counts.
2213 (pass_profile::execute): Update.
2214 (rebuild_frequencies): Use counts only.
2215 (force_edge_cold): Use counts only.
2216 * profile-count.c (profile_count::dump): Dump new count types.
2217 (profile_count::differs_from_p): Check compatiblity.
2218 (profile_count::to_frequency): New function.
2219 (profile_count::to_cgraph_frequency): New function.
2220 * profile-count.h (struct function): Declare.
2221 (enum profile_quality): Add profile_guessed_local and
2222 profile_guessed_global0.
2223 (class profile_proability): Decrease number of bits to 29;
2224 update from_reg_br_prob_note and to_reg_br_prob_note.
2225 (class profile_count: Update comment; decrease number of bits
2226 to 61. Check compatibility.
2227 (profile_count::compatible_p): New private member function.
2228 (profile_count::ipa_p): New member function.
2229 (profile_count::operator<): Handle global zero correctly.
2230 (profile_count::operator>): Handle global zero correctly.
2231 (profile_count::operator<=): Handle global zero correctly.
2232 (profile_count::operator>=): Handle global zero correctly.
2233 (profile_count::nonzero_p): New member function.
2234 (profile_count::force_nonzero): New member function.
2235 (profile_count::max): New member function.
2236 (profile_count::apply_scale): Handle IPA scalling.
2237 (profile_count::guessed_local): New member function.
2238 (profile_count::global0): New member function.
2239 (profile_count::ipa): New member function.
2240 (profile_count::to_frequency): Declare.
2241 (profile_count::to_cgraph_frequency): Declare.
2242 * profile.c (OVERLAP_BASE): Delete.
2243 (compute_frequency_overlap): Delete.
2244 (compute_branch_probabilities): Do not use compute_frequency_overlap.
2245 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
2246 * sched-ebb.c (rank): Use counts only.
2247 * shrink-wrap.c (handle_simple_exit): Use counts only.
2248 (try_shrink_wrapping): Use counts only.
2249 (place_prologue_for_one_component): Use counts only.
2250 * tracer.c (find_best_predecessor): Use to_frequency.
2251 (find_trace): Use to_frequency.
2252 (tail_duplicate): Use to_frequency.
2253 * trans-mem.c (expand_transaction): Do not update frequency.
2254 * tree-call-cdce.c: Do not update frequency.
2255 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
2256 (gimple_merge_blocks): Likewise.
2257 (gimple_split_edge): Likewise.
2258 (gimple_duplicate_sese_region): Likewise.
2259 (gimple_duplicate_sese_tail): Likewise.
2260 (move_sese_region_to_fn): Likewise.
2261 (gimple_account_profile_record): Likewise.
2262 (insert_cond_bb): Likewise.
2263 * tree-complex.c (expand_complex_div_wide): Likewise.
2264 * tree-eh.c (lower_resx): Update profile.
2265 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
2267 (initialize_cfun): Do not initialize frequencies
2268 (freqs_to_counts): Delete.
2269 (copy_cfg_body): Ignore count parameter.
2270 (copy_body): Update.
2271 (expand_call_inline): Update count_max.
2272 (optimize_inline_calls): Update count_max.
2273 (tree_function_versioning): Update count_max.
2274 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
2275 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
2277 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
2278 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
2279 (try_peel_loop): Likewise.
2280 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
2282 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
2283 (tree_transform_and_unroll_loop): Do not use frequencies
2284 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
2285 Use reliable prediction only.
2286 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
2287 * tree-ssa-sink.c (select_best_block): Use to_frequency.
2288 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
2289 probability scaling.
2290 * tree-ssa-threadupdate.c (create_block_for_threading): Do
2291 not update frequency
2292 (any_remaining_duplicated_blocks): Likewise.
2293 (update_profile): Likewise.
2294 (estimated_freqs_path): Delete.
2295 (freqs_to_counts_path): Delete.
2296 (clear_counts_path): Delete.
2297 (ssa_fix_duplicate_block_edges): Likewise.
2298 (duplicate_thread_path): Likewise.
2299 * tree-switch-conversion.c (gen_inbound_check): Use counts.
2300 * tree-tailcall.c (decrease_profile): Do not update frequency.
2301 (eliminate_tail_call): Likewise.
2302 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
2303 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
2304 (optimize_mask_stores): Likewise.
2305 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
2306 * ubsan.c (ubsan_expand_null_ifn): Update profile.
2307 (ubsan_expand_ptr_ifn): Update profile.
2308 * value-prof.c (gimple_ic): Simplify.
2309 * value-prof.h (gimple_ic): Update prototype.
2310 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
2311 * ipa-inline.c (compute_uninlined_call_time): Be sure that
2313 (want_inline_self_recursive_call_p): Likewise.
2314 (resolve_noninline_speculation): Only cummulate defined counts.
2315 (inline_small_functions): Use nonzero_p.
2316 (ipa_inline): Do not access freed node.
2318 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
2320 * config/aarch64/aarch64.c (aarch64_override_options_internal):
2321 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
2323 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
2325 * config/riscv/riscv.c (riscv_legitimize_move): Handle
2326 non-legitimate address.
2328 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2330 * config/rs6000/rs6000.md (*lt0_disi): Delete.
2331 (*lt0_<mode>di, *lt0_<mode>si): New.
2333 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
2335 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
2336 TARGET_PAIRED_FLOAT.
2338 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
2339 Jim Wilson <jim.wilson@linaro.org>
2341 * config/aarch64/aarch64-cores.def (saphira): New CPU.
2342 * config/aarch64/aarch64-tune.md: Regenerated.
2343 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
2344 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
2346 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
2348 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
2349 (arc_expand_prologue): Restore blink for millicode.
2350 * config/arc/linux.h (LINK_EH_SPEC): Defined.
2352 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
2355 * config/i386/i386.c (ix86_vector_duplicate_value): Use
2356 gen_vec_duplicate after forcing the scalar into a register.
2358 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
2360 * combine (try_combine): Print the insns input to try_combine to the
2363 2017-11-02 Steve Ellcey <sellcey@cavium.com>
2366 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
2367 Remove second argument from aarch64_process_target_attr call.
2368 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
2370 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
2372 (aarch64_handle_attr_arch): Remove second argument.
2373 (aarch64_handle_attr_cpu): Ditto.
2374 (aarch64_handle_attr_tune): Ditto.
2375 (aarch64_handle_attr_isa_flags): Ditto.
2376 (aarch64_process_one_target_attr): Ditto.
2377 (aarch64_process_target_attr): Ditto.
2378 (aarch64_option_valid_attribute_p): Remove second argument.
2379 on aarch64_process_target_attr call.
2381 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2383 * diagnostic.c: Include "selftest-diagnostic.h".
2384 (selftest::assert_location_text): New function.
2385 (selftest::test_diagnostic_get_location_text): New function.
2386 (selftest::diagnostic_c_tests): Call it.
2388 2017-11-02 David Malcolm <dmalcolm@redhat.com>
2390 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
2391 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
2392 (class selftest::test_diagnostic_context): Move to...
2393 * selftest-diagnostic.c: New file.
2394 * selftest-diagnostic.h: New file.
2396 2017-11-02 James Bowman <james.bowman@ftdichip.com>
2398 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
2399 offset range for FT32B.
2400 * config/ft32/ft32.h: option "mcompress" enables relaxation.
2401 * config/ft32/ft32.md: Add TARGET_NOPM.
2402 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
2403 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
2405 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
2407 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
2409 2017-11-02 Jeff Law <law@redhat.com>
2411 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
2412 virtual keyword on FINAL OVERRIDE members.
2414 * tree-ssa-propagate.h (ssa_propagation_engine): Group
2415 virtuals together. Add virtual destructor.
2416 (substitute_and_fold_engine): Similarly.
2418 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
2420 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
2422 2017-11-02 Richard Biener <rguenther@suse.de>
2424 PR tree-optimization/82795
2425 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
2427 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2429 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
2431 (gcc_GAS_CHECK_FEATURE): Remove.
2432 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
2433 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
2434 * configure: Regenerate.
2436 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
2438 * config/arc/arc.c (hwloop_optimize): Account for empty
2441 2017-11-02 Richard Biener <rguenther@suse.de>
2444 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
2445 Truncate ARRAY_REF index and element size.
2447 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2449 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
2451 2017-11-01 Jeff Law <law@redhat.com>
2453 * tree-ssa-ccp.c (ccp_folder): New class derived from
2454 substitute_and_fold_engine.
2455 (ccp_folder::get_value): New member function.
2456 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
2457 (ccp_fold_stmt): Remove prototype.
2458 (ccp_finalize): Call substitute_and_fold from the ccp_class.
2459 * tree-ssa-copy.c (copy_folder): New class derived from
2460 substitute_and_fold_engine.
2461 (copy_folder::get_value): Renamed from get_value.
2462 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
2463 * tree-vrp.c (vrp_folder): New class derived from
2464 substitute_and_fold_engine.
2465 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
2466 (vrp_folder::get_value): New member function.
2467 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
2468 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
2469 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
2470 provide a class interface to folder/substitute routines.
2471 (ssa_prop_fold_stmt_fn): Remove typedef.
2472 (ssa_prop_get_value_fn): Likewise.
2473 (subsitute_and_fold): Remove prototype.
2474 (replace_uses_in): Likewise.
2475 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
2476 Renamed from replace_uses_in. Call the virtual member function
2477 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
2478 (substitute_and_fold_dom_walker): Remove initialization of
2479 data member entries for calbacks. Add substitute_and_fold_engine
2480 member and initialize it.
2481 (substitute_and_fold_dom_walker::before_dom_children0: Use the
2482 member functions for get_value, replace_phi_args_in c
2483 replace_uses_in, and fold_stmt calls.
2484 (substitute_and_fold_engine::substitute_and_fold): Renamed from
2485 substitute_and_fold. Remove assert. Update ctor call.
2487 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
2488 (ssa_prop_visit_phi_fn): Likewise.
2489 (class ssa_propagation_engine): New class to provide an interface
2491 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
2493 (ssa_prop_visit_phi): Likewise.
2494 (ssa_propagation_engine::simulate_stmt): Moved into class.
2495 Call visit_phi/visit_stmt from the class rather than via
2496 file scoped static variables.
2497 (ssa_propagation_engine::simulate_block): Moved into class.
2498 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
2499 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
2500 set file scoped statics for the visit_stmt/visit_phi callbacks.
2501 * tree-complex.c (complex_propagate): New class derived from
2502 ssa_propagation_engine.
2503 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
2504 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
2505 (tree_lower_complex): Call ssa_propagate via the complex_propagate
2507 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
2508 ssa_propagation_engine.
2509 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
2510 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
2511 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
2512 * tree-ssa-copy.c (copy_prop): New class derived from
2513 ssa_propagation_engine.
2514 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
2515 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
2516 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
2517 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
2518 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
2519 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
2520 (execute_vrp): Call ssa_propagate from the vrp_prop class.
2522 2017-11-01 Jakub Jelinek <jakub@redhat.com>
2524 PR rtl-optimization/82778
2525 PR rtl-optimization/82597
2526 * compare-elim.c (struct comparison): Add in_a_setter field.
2527 (find_comparison_dom_walker::before_dom_children): Remove killed
2528 bitmap and df_simulate_find_defs call, instead walk the defs.
2529 Compute last_setter and initialize in_a_setter. Merge definitions
2530 with first initialization for a few variables.
2531 (try_validate_parallel): Use insn_invalid_p instead of
2532 recog_memoized. Return insn rather than just the pattern.
2533 (try_merge_compare): Fix up comment. Don't uselessly test if
2534 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
2536 (execute_compare_elim_after_reload): Remove df_chain_add_problem
2539 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2540 Alan Hayward <alan.hayward@arm.com>
2541 David Sherwood <david.sherwood@arm.com>
2543 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
2544 aarch64_hard_regno_nregs to get the number of registers
2547 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2548 Alan Hayward <alan.hayward@arm.com>
2549 David Sherwood <david.sherwood@arm.com>
2551 * config/aarch64/constraints.md (Upl): Rename to...
2553 * config/aarch64/aarch64.md
2554 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
2557 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2558 Alan Hayward <alan.hayward@arm.com>
2559 David Sherwood <david.sherwood@arm.com>
2561 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
2562 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
2565 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2566 Alan Hayward <alan.hayward@arm.com>
2567 David Sherwood <david.sherwood@arm.com>
2569 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
2570 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
2571 (aarch64_evpc_dup): Generate rtl direcly, rather than using
2573 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
2574 of a single element.
2575 * config/aarch64/iterators.md: Add a comment above the permute
2576 unspecs to say that they are generated directly by
2577 aarch64_expand_vec_perm_const.
2578 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
2580 2017-11-01 Nathan Sidwell <nathan@acm.org>
2582 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
2584 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
2586 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
2587 and medany code models, and describe what they do.
2589 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2591 Revert accidental duplicate:
2593 * combine.c (can_change_dest_mode): Reject changes in
2594 REGMODE_NATURAL_SIZE.
2596 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
2598 PR rtl-optimization/64682
2599 PR rtl-optimization/69567
2600 PR rtl-optimization/69737
2601 PR rtl-optimization/82683
2602 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
2603 register mentioned in the note, drop the note, unless it came from I3,
2604 in which case it should go to I3 again.
2606 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2608 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
2609 and return false if not.
2610 (clear_bytes_written_by, live_bytes_read): Update accordingly.
2612 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2614 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
2615 range is known to be empty.
2617 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2618 Alan Hayward <alan.hayward@arm.com>
2619 David Sherwood <david.sherwood@arm.com>
2621 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
2622 and CONST_VECTOR_NUNITS instead of computing the number of units from
2623 the byte sizes of the vector and element.
2624 (simplify_binary_operation_1): Likewise.
2625 (simplify_const_binary_operation): Likewise.
2626 (simplify_ternary_operation): Likewise.
2628 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2629 Alan Hayward <alan.hayward@arm.com>
2630 David Sherwood <david.sherwood@arm.com>
2632 * var-tracking.c (INT_MEM_OFFSET): Replace with...
2633 (int_mem_offset): ...this new function.
2634 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
2635 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
2636 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
2639 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2640 Alan Hayward <alan.hayward@arm.com>
2641 David Sherwood <david.sherwood@arm.com>
2643 * lower-subreg.c (interesting_mode_p): New function.
2644 (compute_costs, find_decomposable_subregs, decompose_register)
2645 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
2646 (resolve_clobber, dump_choices): Use it.
2648 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2649 Alan Hayward <alan.hayward@arm.com>
2650 David Sherwood <david.sherwood@arm.com>
2652 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
2654 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2655 Alan Hayward <alan.hayward@arm.com>
2656 David Sherwood <david.sherwood@arm.com>
2658 * alias.c (find_base_value, find_base_term): Only process integer
2659 truncations. Check the precision rather than the size.
2661 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2662 Alan Hayward <alan.hayward@arm.com>
2663 David Sherwood <david.sherwood@arm.com>
2665 * machmode.h (is_narrower_int_mode): New function
2666 * optabs.c (expand_float, expand_fix): Use it.
2667 * dwarf2out.c (rotate_loc_descriptor): Likewise.
2669 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2670 Alan Hayward <alan.hayward@arm.com>
2671 David Sherwood <david.sherwood@arm.com>
2673 * rtl.h (narrower_subreg_mode): New function.
2674 * ira-color.c (update_costs_from_allocno): Use it.
2676 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2677 Alan Hayward <alan.hayward@arm.com>
2678 David Sherwood <david.sherwood@arm.com>
2680 * optabs-query.h (convert_optab_p): New function, split out from...
2681 (convert_optab_handler): ...here.
2682 (widening_optab_handler): Delete.
2683 (find_widening_optab_handler): Remove permit_non_widening parameter.
2684 (find_widening_optab_handler_and_mode): Likewise. Provide an
2685 override that operates on mode class wrappers.
2686 * optabs-query.c (widening_optab_handler): Delete.
2687 (find_widening_optab_handler_and_mode): Remove permit_non_widening
2688 parameter. Assert that the two modes are the same class and that
2689 the "from" mode is narrower than the "to" mode. Use
2690 convert_optab_handler instead of widening_optab_handler.
2691 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
2692 instead of widening_optab_handler.
2693 * expr.c (expand_expr_real_2): Update calls to
2694 find_widening_optab_handler.
2695 * optabs.c (expand_widen_pattern_expr): Likewise.
2696 (expand_binop_directly): Take the insn_code as a parameter.
2697 (expand_binop): Only call find_widening_optab_handler for
2698 conversion optabs; use optab_handler otherwise. Update calls
2699 to find_widening_optab_handler and expand_binop_directly.
2700 Use convert_optab_handler instead of widening_optab_handler.
2701 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
2702 find_widening_optab_handler and use scalar_mode rather than
2704 (convert_plusminus_to_widen): Likewise.
2706 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2707 Alan Hayward <alan.hayward@arm.com>
2708 David Sherwood <david.sherwood@arm.com>
2710 * machmode.h (fixed_size_mode): New class.
2711 * rtl.h (get_pool_mode): Return fixed_size_mode.
2712 * gengtype.c (main): Add fixed_size_mode.
2713 * target.def (get_raw_result_mode): Return a fixed_size_mode.
2714 (get_raw_arg_mode): Likewise.
2715 * doc/tm.texi: Regenerate.
2716 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
2717 * targhooks.c (default_get_reg_raw_mode): Likewise.
2718 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
2719 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
2720 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
2721 (msp430_get_raw_result_mode): Likewise.
2722 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
2723 * dbxout.c (dbxout_parms): Require fixed-size modes.
2724 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
2725 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
2726 * omp-low.c (lower_oacc_reductions): Likewise.
2727 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
2728 (simplify_subreg): Update accordingly.
2729 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
2730 (force_const_mem): Update accordingly. Return NULL_RTX for modes
2731 that aren't fixed-size.
2732 (get_pool_mode): Return a fixed_size_mode.
2733 (output_constant_pool_2): Take a fixed_size_mode.
2735 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2736 Alan Hayward <alan.hayward@arm.com>
2737 David Sherwood <david.sherwood@arm.com>
2739 * doc/rtl.texi (vec_series): Document.
2740 (const): Say that the operand can be a vec_series.
2741 * rtl.def (VEC_SERIES): New rtx code.
2742 * rtl.h (const_vec_series_p_1): Declare.
2743 (const_vec_series_p): New function.
2744 * emit-rtl.h (gen_const_vec_series): Declare.
2745 (gen_vec_series): Likewise.
2746 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
2747 (gen_vec_series): Likewise.
2748 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
2749 * simplify-rtx.c (simplify_unary_operation): Handle negations
2751 (simplify_binary_operation_series): New function.
2752 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
2753 (test_vector_ops_series): New function.
2754 (test_vector_ops): Call it.
2755 * config/powerpcspe/altivec.md (altivec_lvsl): Use
2756 gen_const_vec_series.
2757 (altivec_lvsr): Likewise.
2758 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
2760 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2761 Alan Hayward <alan.hayward@arm.com>
2762 David Sherwood <david.sherwood@arm.com>
2764 * doc/rtl.texi (const): Update description of address constants.
2765 Say that vector constants are allowed too.
2766 * common.md (E, F): Use CONSTANT_P instead of checking for
2768 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
2769 checking for CONST_VECTOR.
2770 * expmed.c (make_tree): Use build_vector_from_val for a CONST
2772 * expr.c (expand_expr_real_2): Check for vector modes instead
2773 of checking for CONST_VECTOR.
2774 * rtl.h (const_vec_p): New function.
2775 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
2776 (unwrap_const_vec_duplicate): Handle them here too.
2778 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2779 David Malcolm <dmalcolm@redhat.com>
2780 Alan Hayward <alan.hayward@arm.com>
2781 David Sherwood <david.sherwood@arm.com>
2783 * rtl.h (vec_duplicate_p): New function.
2784 * selftest-rtl.c (assert_rtx_eq_at): New function.
2785 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
2786 (assert_rtx_eq_at): Declare.
2787 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
2788 * selftest-run-tests.c (selftest::run_tests): Call it.
2789 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
2790 (simplify_unary_operation_1): Recursively handle vector duplicates.
2791 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
2793 (simplify_subreg): Handle subregs of vector duplicates.
2794 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
2795 (selftest::simplify_rtx_c_tests): New functions.
2797 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2798 Alan Hayward <alan.hayward@arm.com>
2799 David Sherwood <david.sherwood@arm.com>
2801 * emit-rtl.h (gen_const_vec_duplicate): Declare.
2802 (gen_vec_duplicate): Likewise.
2803 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
2805 (gen_const_vector): ...here.
2806 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
2807 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
2808 whose elements are all equal.
2809 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
2810 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
2811 (simplify_relational_operation): Likewise.
2812 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
2814 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
2815 (aarch64_expand_vector_init): Likewise.
2816 * config/arm/arm.c (neon_vdup_constant): Likewise.
2817 (neon_expand_vector_init): Likewise.
2818 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
2819 (arm_block_set_unaligned_vect): Likewise.
2820 (arm_block_set_aligned_vect): Likewise.
2821 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
2822 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
2823 (expand_vec_perm_even_odd_pack): Likewise.
2824 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
2825 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
2826 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
2827 gen_const_vec_duplicate.
2828 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
2829 * config/mips/mips.c (mips_gen_const_int_vector): Use
2830 gen_const_vec_duplicate.
2831 (mips_expand_vector_init): Use CONST0_RTX.
2832 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
2833 (define_split): Use gen_const_vec_duplicate.
2834 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
2835 (define_split): Use gen_const_vec_duplicate.
2836 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
2837 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
2838 * config/spu/spu.c (spu_const): Likewise.
2840 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
2841 Alan Hayward <alan.hayward@arm.com>
2842 David Sherwood <david.sherwood@arm.com>
2844 * combine.c (can_change_dest_mode): Reject changes in
2845 REGMODE_NATURAL_SIZE.
2847 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
2849 * configure.ac (--enable-libssp): New.
2850 (gcc_cv_libc_provides_ssp): Check for explicit setting before
2851 trying to determine target-specific default. Adjust indentation.
2852 * configure: Regenerated.
2853 * doc/install.texi (Configuration): Expand --disable-libssp
2856 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
2858 config/i386/i386.c (ix86_expand_epilogue): Correct stack
2861 2017-10-31 Martin Jambor <mjambor@suse.cz>
2864 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
2866 2017-10-31 David Malcolm <dmalcolm@redhat.com>
2868 * auto-profile.c (autofdo_source_profile::read): Use
2869 UNKNOWN_LOCATION rather than 0.
2870 * diagnostic-core.h (warning_at_rich_loc): Rename to...
2871 (warning_at): ...this overload.
2872 (warning_at_rich_loc_n): Rename to...
2873 (warning_n): ...this overload.
2874 (error_at_rich_loc): Rename to...
2875 (error_at): ...this overload.
2876 (pedwarn_at_rich_loc): Rename to...
2877 (pedwarn): ...this overload.
2878 (permerror_at_rich_loc): Rename to...
2879 (permerror): ...this overload.
2880 (inform_at_rich_loc): Rename to...
2881 (inform): ...this overload.
2882 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
2883 (diagnostic_n_impl_richloc): Rename to...
2884 (diagnostic_n_impl): ...this rich_location *-based decl.
2885 (inform_at_rich_loc): Rename to...
2886 (inform): ...this, and add an assertion.
2887 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
2888 (warning_at_rich_loc): Rename to...
2889 (warning_at): ...this, and add an assertion.
2890 (warning_at_rich_loc_n): Rename to...
2891 (warning_n): ...this, and add an assertion.
2892 (warning_n): Update location_t-based implementation for removal of
2893 location_t-based diagnostic_n_impl.
2894 (pedwarn_at_rich_loc): Rename to...
2895 (pedwarn): ...this, and add an assertion.
2896 (permerror_at_rich_loc): Rename to...
2897 (permerror): ...this, and add an assertion.
2898 (error_n): Update for removal of location_t-based diagnostic_n_impl.
2899 (error_at_rich_loc): Rename to...
2900 (error_at): ...this, and add an assertion.
2901 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
2902 (driver::do_spec_on_infiles): Likewise.
2903 * substring-locations.c (format_warning_va): Update for renaming
2904 of inform_at_rich_loc.
2906 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
2908 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
2909 _Float<N>X built-in functions so that the variant without the
2910 "__builtin_" prefix is only enabled for the GNU C and Objective C
2911 languages when they are in non-strict ANSI/ISO mode.
2912 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
2913 * target.def (floatn_builtin_p): Add a target hook to control
2914 whether _Float<N> and _Float<N>X built-in functions without the
2915 "__builtin_" prefix are enabled, and return true for C and
2916 Objective C in the default hook. Include langhooks.h in
2918 * targhooks.h (default_floatn_builtin_p): Likewise.
2919 * targhooks.c (default_floatn_builtin_p): Likewise.
2920 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
2921 floatn_builtin_p target hook.
2922 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
2924 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
2925 Eric Botcazou <ebotcazou@adacore.com>
2927 PR rtl-optimization/81803
2928 * lra-constraints.c (curr_insn_transform): Also reload the whole
2929 register for a strict subreg no wider than a word if this is for
2930 a WORD_REGISTER_OPERATIONS target.
2932 2017-10-31 Jason Merrill <jason@redhat.com>
2934 * gdbinit.in: Skip over inlines from timevar.h.
2936 2017-10-31 Martin Liska <mliska@suse.cz>
2938 * doc/gcov.texi: Document new option.
2939 * gcov.c (print_usage): Likewise print it.
2940 (process_args): Support the argument.
2941 (format_count): New function.
2942 (format_gcov): Use the function.
2944 2017-10-31 Martin Liska <mliska@suse.cz>
2946 * gcov.c (struct name_map): do not use typedef.
2947 Define operator== and operator<.
2948 (name_search): Remove.
2949 (name_sort): Remove.
2950 (main): Do not allocate names.
2951 (process_file): Add vertical space.
2952 (generate_results): Use std::find.
2953 (release_structures): Do not release memory.
2954 (find_source): Use std::find.
2956 2017-10-31 Martin Liska <mliska@suse.cz>
2958 * gcov.c (struct line_info): Remove it's typedef.
2959 (line_info::line_info): Add proper ctor.
2960 (line_info::has_block): Do not use a typedef.
2961 (struct source_info): Do not use typedef.
2962 (circuit): Likewise.
2963 (get_cycles_count): Likewise.
2964 (output_intermediate_file): Iterate via vector iterator.
2965 (add_line_counts): Use std::vector methods.
2966 (accumulate_line_counts): Likewise.
2967 (output_lines): Likewise.
2969 2017-10-31 Martin Liska <mliska@suse.cz>
2971 * gcov.c (struct source_info): Remove typedef.
2972 (source_info::source_info): Add proper ctor.
2973 (accumulate_line_counts): Use struct, not it's typedef.
2974 (output_gcov_file): Likewise.
2975 (output_lines): Likewise.
2976 (main): Do not allocate an array.
2977 (output_intermediate_file): Use size of vector container.
2978 (process_file): Resize the vector.
2979 (generate_results): Do not preallocate, use newly added vector
2981 (release_structures): Do not release sources.
2982 (find_source): Use vector methods.
2983 (add_line_counts): Do not use typedef.
2985 2017-10-31 Martin Liska <mliska@suse.cz>
2987 * doc/gcov.texi: Document that.
2988 * gcov.c (add_line_counts): Mark lines with a non-executed
2990 (output_line_beginning): Handle such lines.
2991 (output_lines): Pass new argument.
2992 (output_intermediate_file): Print it in intermediate format.
2994 2017-10-31 Martin Liska <mliska@suse.cz>
2996 * color-macros.h: New file.
2997 * diagnostic-color.c: Factor out color related to macros to
2999 * doc/gcov.texi: Document -k option.
3000 * gcov.c (INCLUDE_STRING): Include string.h.
3001 (print_usage): Add -k option.
3002 (process_args): Parse it.
3003 (pad_count_string): New function.
3004 (output_line_beginning): Likewise.
3005 (DEFAULT_LINE_START): New macro.
3006 (output_lines): Support color output.
3008 2017-10-31 Martin Liska <mliska@suse.cz>
3010 PR gcov-profile/82633
3011 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
3012 their interaction with GCOV infrastructure.
3013 * configure.ac: Add -fkeep-{inline,static}-functions to
3015 * configure: Regenerate.
3017 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
3020 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
3022 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
3025 * config/rs6000/rs6000.md (allocate_stack): Force update interval
3026 into a register if it does not fit into an immediate offset field.
3028 2017-10-31 Olivier Hainque <hainque@adacore.com>
3030 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
3032 2017-10-31 Julia Koval <julia.koval@intel.com>
3034 * config.gcc: Add gfniintrin.h.
3035 * config/i386/gfniintrin.h: New.
3036 * config/i386/i386-builtin-types.def
3037 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
3038 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
3039 __builtin_ia32_vgf2p8affineinvqb_v32qi,
3040 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
3041 __builtin_ia32_vgf2p8affineinvqb_v16qi,
3042 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
3043 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
3044 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
3045 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
3046 V64QI_FTYPE_V64QI_V64QI_INT): New types.
3047 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
3048 * config/i386/immintrin.h: Include gfniintrin.h.
3049 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
3051 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
3053 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
3055 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3057 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
3058 (arm_ashldi3_1bit): Remove pattern.
3059 (ashrdi3): Remove shift by 1 expansion.
3060 (arm_ashrdi3_1bit): Remove pattern.
3061 (lshrdi3): Remove shift by 1 expansion.
3062 (arm_lshrdi3_1bit): Remove pattern.
3063 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
3064 cost of ashldi3 by 1.
3065 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
3066 (<shift>di3_neon): Likewise.
3068 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
3070 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
3071 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
3072 and (*aarch64_simd_mov<VQ:mode>).
3073 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
3074 pattern alternative.
3075 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
3076 attributes to match pattern alternative.
3078 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
3080 * config.gcc (powerpc*-*-*): Add emmintrin.h.
3081 * config/rs6000/emmintrin.h: New file.
3082 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
3084 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
3086 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
3087 * (movdi_vfp_cortexa8): Remove pattern.
3089 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3091 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
3093 (Specific, alpha*-dec-osf5.1): Remove.
3094 (Specific, mips-sgi-irix5): Remove.
3095 (Specific, mips-sgi-irix6): Remove.
3097 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3100 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
3101 arguments to clear_bit_region_be.
3103 2017-10-30 Jim Wilson <wilson@tuliptree.org>
3105 * gimplify.c: Include memmodel.h.
3107 2017-10-30 Martin Jambor <mjambor@suse.cz>
3109 * omp-grid.c (grid_attempt_target_gridification): Also insert a
3110 condition whether loop should be executed at all.
3112 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
3114 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
3115 gimple folding of vec_madd() intrinsics.
3116 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
3117 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
3118 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
3120 2017-10-30 Richard Biener <rguenther@suse.de>
3122 PR tree-optimization/82762
3124 2017-10-23 Richard Biener <rguenther@suse.de>
3126 PR tree-optimization/82129
3128 2017-08-01 Richard Biener <rguenther@suse.de>
3130 PR tree-optimization/81181
3131 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
3132 (compute_antic): ... end of iteration here.
3134 2017-10-30 Joseph Myers <joseph@codesourcery.com>
3136 * doc/invoke.texi (C Dialect Options): Document -std=c17,
3137 -std=iso9899:2017 and -std=gnu17.
3138 * doc/standards.texi (C Language): Document C17 support.
3139 * doc/cpp.texi (Overview): Mention -std=c17.
3140 (Standard Predefined Macros): Document C11 and C17 values of
3141 __STDC_VERSION__. Do not refer to C99 support as incomplete.
3142 * doc/extend.texi (Inline): Do not list individual options for
3143 standards newer than C99.
3144 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
3146 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
3149 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
3151 * asan.c (asan_finish_file): Align asan globals array by shadow
3154 2017-10-30 Jakub Jelinek <jakub@redhat.com>
3157 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
3158 (struct store_immediate_info): Add bitregion_start and bitregion_end
3160 (store_immediate_info::store_immediate_info): Add brs and bre
3161 arguments and initialize bitregion_{start,end} from those.
3162 (struct merged_store_group): Add bitregion_start, bitregion_end,
3163 align_base and mask fields. Drop unnecessary struct keyword from
3164 struct store_immediate_info. Add do_merge method.
3165 (clear_bit_region_be): Use memset instead of loop storing zeros.
3166 (merged_store_group::do_merge): New method.
3167 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
3168 stores as long as the surrounding bitregions have no gaps.
3169 (merged_store_group::merge_overlapping): Use do_merge.
3170 (merged_store_group::apply_stores): Test that bitregion_{start,end}
3171 is byte aligned, rather than requiring that start and width are
3172 byte aligned. Drop unnecessary struct keyword from
3173 struct store_immediate_info. Allocate and populate also mask array.
3174 Make start of the arrays relative to bitregion_start rather than
3175 start and size them according to bitregion_{end,start} difference.
3176 (struct imm_store_chain_info): Drop unnecessary struct keyword from
3177 struct store_immediate_info.
3178 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
3179 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
3180 struct keyword from struct store_immediate_info.
3181 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
3182 between stores as long as the surrounding bitregions have no gaps.
3184 (struct split_store): Add orig non-static data member.
3185 (split_store::split_store): Initialize orig to false.
3186 (find_constituent_stmts): Return store_immediate_info *, non-NULL
3187 if there is exactly a single original stmt. Change stmts argument
3188 to pointer from reference, if NULL, don't push anything to it. Add
3189 first argument, use it to optimize skipping over orig stmts that
3190 are known to be before bitpos already. Simplify.
3191 (split_group): Return unsigned int count how many stores are or
3192 would be needed rather than a bool. Add allow_unaligned argument.
3193 Change split_stores argument from reference to pointer, if NULL,
3194 only do a dry run computing how many stores would be produced.
3195 Rewritten algorithm to use both alignment and misalign if
3196 !allow_unaligned and handle bitfield stores with gaps.
3197 (imm_store_chain_info::output_merged_store): Set start_byte_pos
3198 from bitregion_start instead of start. Compute allow_unaligned
3199 here, if true, do 2 split_group dry runs to compute which one
3200 produces fewer stores and prefer aligned if equal. Punt if
3201 new count is bigger or equal than original before emitting any
3202 statements, rather than during that. Remove no longer needed
3203 new_ssa_names tracking. Replace num_stmts with
3204 split_stores.length (). Use 32-bit stack allocated entries
3205 in split_stores auto_vec. Try to reuse original store lhs/rhs1
3206 if possible. Handle bitfields with gaps.
3207 (pass_store_merging::execute): Ignore bitsize == 0 stores.
3208 Compute bitregion_{start,end} for the stores and construct
3209 store_immediate_info with that. Formatting fixes.
3211 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
3214 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
3215 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
3217 2017-10-29 Jim Wilson <wilson@tuliptree.org>
3219 * gimplify.c: Include tm_p.h.
3221 * common.opt (gcoff): Re-add as ignored option.
3222 (gcoff1, gcoff2, gcoff3): Likewise.
3224 * Makefile.in (OBJS): Delete sdbout.o.
3225 (GTFILES): Delete $(srcdir)/sdbout.c.
3226 * debug.h: Delete sdb_debug_hooks.
3227 * final.c: Delete sdbout.h include.
3228 (final_scan_insn): Delete SDB_DEBUG check.
3229 (rest_of_clean_state): Likewise.
3230 * output.h: Delete sdb_begin_function_line.
3233 * toplev.c: Delete sdbout.h include.
3234 (process_options): Delete SDB_DEBUG check.
3235 * tree-core.h (tree_type_common): Delete pointer field of
3237 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
3238 TYPE_SYMTAB_POINTER.
3239 * tree.h (TYPE_SYMTAB_POINTER): Delete.
3240 (TYPE_SYMTAB_IS_POINTER): Delete.
3241 (TYPE_SYMTAB_IS_DIE): Renumber.
3242 * xcoffout.c: Refer to former sdbout.c file.
3243 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
3245 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
3246 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
3247 (Debugging Options): Delete -gcoff.
3248 (-gstabs): Delete SDB reference.
3250 (-gcoff@var{level}): Delete.
3251 * doc/passes.texi (Debugging information output): Delete SDB and
3252 sdbout.c references.
3253 * doc/tm.texi: Regenerate.
3254 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
3255 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
3257 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
3258 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
3259 (SDB_DEBUGGING_INFO): Delete.
3260 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
3261 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
3262 * target.def (output_source_filename): Delete COFF reference.
3264 * common.opt (gcoff): Delete.
3265 (gxcoff+): Update Negative chain.
3266 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
3268 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
3269 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
3270 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
3272 (expand_function_start): Change sdb reference to past tense.
3273 (expand_function_end): Change sdb reference to past tense.
3274 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
3275 * opts.c (debug_type_names): Delete coff entry.
3276 (common_handle_option): Delete OPT_gcoff case.
3277 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
3279 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
3280 * config/cris/cris.h: Delete SDB reference in comment.
3281 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
3282 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
3283 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
3284 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
3286 (ix86_expand_prologue): Likewise.
3287 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
3288 * config/ia64/ia64.h: Likewise.
3289 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
3290 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
3292 * config/mmix/mmix.h: Likewise.
3293 * config/nds32/nds32.c: Likewise.
3294 * config/stormy/storym16.h: Likewise.
3295 * config/visium/visium.h: Likewise.
3296 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
3298 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
3300 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
3301 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
3302 FRAME_POINTER_REGNUM point at high end of local var area.
3304 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
3306 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
3307 Move comment around. Do not reset best_edge for a copiable
3308 destination if the copy would cause a partition change.
3309 (better_edge_p): Remove redundant check.
3311 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3313 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
3315 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3317 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
3318 for math functions that have _Float<N> and _Float<N>X variants.
3319 (mathfn_built_in_2): Add support for math functions that have
3320 _Float<N> and _Float<N>X variants.
3321 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
3322 (expand_builtin_mathfn_ternary): Add support for fma with
3323 _Float<N> and _Float<N>X variants.
3324 (expand_builtin): Likewise.
3325 (fold_builtin_3): Likewise.
3326 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
3327 create math function _Float<N> and _Float<N>X variants as external
3329 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
3330 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
3331 the __builtin_ prefix and if not strict ansi, without the prefix.
3332 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
3333 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
3334 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
3335 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
3336 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
3337 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
3338 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
3339 function signatures for fma _Float<N> and _Float<N>X variants.
3340 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
3341 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
3342 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
3343 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
3344 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
3345 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
3346 * gencfn-macros.c (print_case_cfn): Add support for math functions
3347 that have _Float<N> and _Float<N>X variants.
3348 (print_define_operator_list): Likewise.
3349 (fltfn_suffixes): Likewise.
3351 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
3352 for math functions that have _Float<N> and _Float<N>X variants.
3353 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
3354 and _Float<N>X variants.
3355 (COPYSIGN): Likewise.
3358 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
3359 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
3361 (integer_valued_read_call_p): Likewise.
3362 * fold-const-call.c (fold_const_call_ss): Likewise.
3363 (fold_const_call_sss): Add support for copysign, fmin, and fmax
3364 _Float<N> and _Float<N>X variants.
3365 (fold_const_call_ssss): Add support for fma _Float<N> and
3366 _Float<N>X variants.
3367 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
3368 support for copysign and fma _Float<N> and _Float<N>X variants.
3369 (backprop::process_builtin_call_use): Likewise.
3370 * tree-call-cdce.c (can_test_argument_range); Add support for
3371 sqrt _Float<N> and _Float<N>X variants.
3372 (edom_only_function): Likewise.
3373 (get_no_error_domain): Likewise.
3374 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
3375 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
3376 copysign _Float<N> and _Float<N>X variants.
3377 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
3378 handled by machine independent code.
3379 (FMAF128): Likewise.
3380 * doc/cpp.texi (Common Predefined Macros): Document defining
3381 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
3382 fma _Float<N> and _Float<N>X variants.
3384 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
3387 * config/i386/i386-modes.def (CCFPU): Remove definition.
3388 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
3389 (ix86_cc_modes_compatible): Ditto.
3390 (ix86_expand_carry_flag_compare): Ditto.
3391 (ix86_expand_int_movcc): Ditto.
3392 (ix86_expand_int_addcc): Ditto.
3393 (ix86_reverse_condition): Ditto.
3394 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
3395 Return true/false for unordered/ordered fp comparisons.
3396 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
3397 (ix86_prepare_fp_compare_args): Update for rename.
3398 (ix86_expand_fp_compare): Update for rename. Generate unordered
3399 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
3400 (ix86_expand_sse_compare_and_jump): Ditto.
3401 * config/i386/predicates.md (fcmov_comparison_operator):
3402 Remove CCFPU mode handling.
3403 (ix86_comparison_operator): Ditto.
3404 (ix86_carry_flag_operator): Ditto.
3405 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
3406 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
3407 (*cmpu<mode>_cc_i387): Ditto.
3408 (FPCMP): Remove mode iterator.
3409 (unord): Remove mode attribute.
3410 (unord_subst): New define_subst transformation
3411 (unord): New define_subst attribute.
3413 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
3414 (*cmpi<unord>xf_i387): Ditto.
3415 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
3416 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
3417 using unord_subst transformation.
3418 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
3419 (round_saeonly): Also handle CCFP mode.
3420 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
3421 Remove UNSPEC_SAHF unspec handling.
3423 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3425 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
3427 2017-10-27 Jeff Law <law@redhat.com>
3429 * gimple-ssa-sprintf.c: Include domwalk.h.
3430 (class sprintf_dom_walker): New class, derived from dom_walker.
3431 (sprintf_dom_walker::before_dom_children): New function.
3432 (struct call_info): Moved into sprintf_dom_walker class
3433 (compute_formath_length, handle_gimple_call): Likewise.
3434 (sprintf_length::execute): Call the dominator walker rather
3435 than walking the statements.
3437 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
3438 gimple statement locations.
3439 (check_array_bounds): Corresponding changes. Get the statement's
3440 location directly from wi->stmt.
3442 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
3445 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
3447 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
3449 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
3450 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
3452 2017-10-27 Jakub Jelinek <jakub@redhat.com>
3455 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
3456 * config/i386/i386.c (maybe_get_pool_constant): Removed.
3457 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
3458 maybe_get_pool_constant.
3459 * config/i386/predicates.md (zero_extended_scalar_load_operand):
3462 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3464 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
3465 2.26 caveat. Update gas and gld versions.
3466 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
3469 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
3471 * cgraph.h (set_malloc_flag): Declare.
3472 * cgraph.c (set_malloc_flag_1): New function.
3473 (set_malloc_flag): Likewise.
3474 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
3475 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
3477 (read_ipa_call_summary): Add support for reading is_return_callee.
3478 (write_ipa_call_summary): Stream is_return_callee.
3479 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
3480 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
3481 ipa-prop.h, ipa-fnsummary.h.
3482 (pure_const_names): Change to static.
3483 (malloc_state_e): Define.
3484 (malloc_state_names): Define.
3485 (funct_state_d): Add field malloc_state.
3486 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
3487 (check_retval_uses): New function.
3488 (malloc_candidate_p): Likewise.
3489 (analyze_function): Add support for malloc attribute.
3490 (pure_const_write_summary): Stream malloc_state.
3491 (pure_const_read_summary): Add support for reading malloc_state.
3492 (dump_malloc_lattice): New function.
3493 (propagate_malloc): New function.
3494 (warn_function_malloc): New function.
3495 (ipa_pure_const::execute): Call propagate_malloc and
3496 ipa_free_fn_summary.
3497 (pass_local_pure_const::execute): Add support for malloc attribute.
3498 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
3499 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
3501 2017-10-27 Martin Liska <mliska@suse.cz>
3503 PR gcov-profile/82457
3504 * doc/invoke.texi: Document that one needs a non-strict ISO mode
3505 for fork-like functions to be properly instrumented.
3507 2017-10-27 Richard Biener <rguenther@suse.de>
3510 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
3511 info when we redirected EH.
3513 2017-10-26 Michael Collison <michael.collison@arm.com>
3515 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
3517 (<optab>_trunchf<GPI:mode>2: New pattern.
3518 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
3519 * config/aarch64/iterators.md (wv): New mode attribute.
3520 (vf, VF): New mode attributes.
3521 (vgp, VGP): New mode attributes.
3522 (s): Update attribute with SImode and DImode prefixes.
3524 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3526 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
3527 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
3528 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
3529 (nios2_option_overide): Initialize it. Don't allow R0-relative
3530 addressing with PIC.
3531 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
3532 (nios2_symbolic_constant_p): Likewise.
3533 (nios2_legitimate_address_p): Likewise.
3534 (nios2_r0rel_section_name_p): New.
3535 (nios2_symbol_ref_in_r0rel_data_p): New.
3536 (nios2_emit_move_sequence): Handle r0rel_constant_p.
3537 (r0rel_constant_p): New.
3538 (nios2_print_operand_address): Handle r0rel_constant_p.
3539 (nios2_cdx_narrow_form_p): Likewise.
3540 * config/nios2/nios2.opt (mr0rel-sec=): New option.
3541 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
3542 (Nios II Options): Document -mr0rel-sec.
3544 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
3546 * config/nios2/nios2.c: Include xregex.h.
3547 (nios2_gprel_sec_regex): New.
3548 (nios2_option_overide): Initialize it. Don't allow GP-relative
3549 addressing with PIC.
3550 (nios2_small_section_name_p): Check for regex match.
3551 * config/nios2/nios2.opt (mgprel-sec=): New option.
3552 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
3553 (Nios II Options): Document -mgprel-sec.
3555 2017-10-26 Jim Wilson <wilson@tuliptree.org>
3557 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
3559 2017-10-26 Tom de Vries <tom@codesourcery.com>
3561 PR tree-optimization/82707
3562 * gimple.c (gimple_copy): Fix unsharing of
3563 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
3565 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
3567 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
3568 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
3569 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
3570 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
3571 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
3572 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
3573 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
3574 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
3575 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
3576 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
3577 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
3578 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
3579 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
3580 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
3581 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
3582 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
3583 _mm512_mask_cmpunord_ps_mask): New intrinsics.
3585 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
3587 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
3589 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
3590 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
3591 warning to rs6000.c. Remove the Undocumented flag, since it has
3593 (-mabi=ibmlongdouble): Likewise.
3594 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
3595 already set, set the default format for long double.
3596 (rs6000_debug_reg_global): Print whether long double is IBM or
3598 (rs6000_option_override_internal): Rework setting long double
3599 format. Only warn if the user is changing the long double default
3600 and they did not use -Wno-psabi.
3601 * doc/invoke.texi (PowerPC options): Update the documentation for
3602 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
3604 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3605 Alan Hayward <alan.hayward@arm.com>
3606 David Sherwood <david.sherwood@arm.com>
3608 * rtl.h (wider_subreg_mode): New function.
3609 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
3610 rather than an unsigned int *.
3611 * ira-color.c (regno_max_ref_width): Replace with...
3612 (regno_max_ref_mode): ...this new variable.
3613 (coalesced_pseudo_reg_slot_compare): Update accordingly.
3614 Use wider_subreg_mode.
3615 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
3616 rather than an unsigned int *.
3617 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
3618 (process_alt_operands): Likewise.
3619 (invariant_p): Likewise.
3620 * lra-spills.c (assign_mem_slot): Likewise.
3621 (add_pseudo_to_slot): Likewise.
3622 * lra.c (collect_non_operand_hard_regs): Likewise.
3623 (add_regs_to_insn_regno_info): Likewise.
3624 * reload1.c (regno_max_ref_width): Replace with...
3625 (regno_max_ref_mode): ...this new variable.
3626 (reload): Update accordingly. Update call to
3627 ira_sort_regnos_for_alter_reg.
3628 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
3629 (init_eliminable_invariants): Update to use regno_max_ref_mode.
3630 (scan_paradoxical_subregs): Likewise.
3632 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3634 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
3635 (aarch64_frame): Add emit_frame_chain boolean.
3636 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3637 Move eh_return case to aarch64_layout_frame.
3638 (aarch64_layout_frame): Initialize emit_frame_chain.
3639 (aarch64_expand_prologue): Use emit_frame_chain.
3641 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3643 * config/aarch64/aarch64.c (aarch64_layout_frame):
3644 Ensure LR is always stored at the bottom of the callee-saves.
3645 Remove rarely used frame layout which saves callee-saves at top of
3646 frame, so the store of LR can be used as a valid probe in all cases.
3648 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
3650 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
3651 Improve unaligned TImode/TFmode base/offset split.
3653 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3654 Alan Hayward <alan.hayward@arm.com>
3655 David Sherwood <david.sherwood@arm.com>
3657 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
3658 * combine.c (find_single_use_1): Likewise.
3659 (expand_field_assignment): Likewise.
3660 (move_deaths): Likewise.
3661 * lra-constraints.c (simplify_operand_subreg): Likewise.
3662 (curr_insn_transform): Likewise.
3663 * lra.c (collect_non_operand_hard_regs): Likewise.
3664 (add_regs_to_insn_regno_info): Likewise.
3665 * rtlanal.c (reg_referenced_p): Likewise.
3666 (covers_regno_no_parallel_p): Likewise.
3668 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3670 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
3671 Don't print any bits outside the precision of the value.
3672 * wide-int.cc (test_printing): Add some new tests.
3674 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3676 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
3677 supports -xbrace_comment option.
3678 * configure: Regenerate.
3679 * config.in: Regenerate.
3680 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
3681 (ASM_CPU_SPEC): Use it.
3683 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
3685 * target.def (static_rtx_alignment): New hook.
3686 * targhooks.h (default_static_rtx_alignment): Declare.
3687 * targhooks.c (default_static_rtx_alignment): New function.
3688 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
3689 * doc/tm.texi: Regenerate.
3690 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
3691 instead of targetm.constant_alignment. Remove call to
3693 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3694 (cris_preferred_mininum_alignment): New function, split out from...
3695 (cris_constant_alignment): ...here.
3696 (cris_static_rtx_alignment): New function.
3697 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
3699 (ix86_constant_alignment): ...here.
3700 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3701 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3702 (mmix_static_rtx_alignment): New function.
3703 * config/spu/spu.c (spu_static_rtx_alignment): New function.
3704 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
3706 2017-10-26 Tamar Christina <tamar.christina@arm.com>
3709 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
3710 Add flag_trapping_math and flag_fp_int_builtin_inexact.
3712 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
3714 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
3715 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
3716 mark as a sign-extending load.
3717 (local_pic_load_u): Define.
3719 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
3722 * fold-const.c (operand_equal_for_comparison_p): Also return true
3723 if ARG0 is a simple variant of ARG1 with narrower precision.
3724 (fold_ternary_loc): Always pass unstripped operands to the predicate.
3726 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
3728 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
3730 * i386.h (processor_costs): Add gather_static, gather_per_elt,
3731 scatter_static, scatter_per_elt.
3732 * x86-tune-costs.h: Add new cost entries.
3734 2017-10-25 Richard Biener <rguenther@suse.de>
3736 * tree-ssa-sccvn.h (vn_eliminate): Declare.
3737 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
3738 class pass_fre): Move to ...
3739 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
3740 class pass_fre): ... here and adjust for statistics.
3742 2017-10-25 Jakub Jelinek <jakub@redhat.com>
3745 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
3746 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
3747 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
3748 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
3751 2017-10-25 Richard Biener <rguenther@suse.de>
3753 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
3754 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
3755 eliminate_push_avail, eliminate_insert): Move inside...
3756 (class eliminate_dom_walker): ... this class in preparation
3758 (fini_eliminate): Remove by merging with ...
3759 (eliminate): ... this function. Adjust for class changes.
3760 (pass_pre::execute): Remove fini_eliminate call.
3761 (pass_fre::execute): Likewise.
3763 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3766 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
3767 (VPERMI2, VPERMI2I): New mode iterators.
3768 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
3769 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
3771 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
3772 mode iterator. Remove 3 old define_insn patterns.
3773 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
3774 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
3775 VPERMI2 mode iterator, remove the other two expanders.
3776 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
3777 to use VPERMI2 mode iterator, add another alternative for vpermi2*
3778 instructions, remove the other two patterns.
3779 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
3780 mode iterator, remove the other two patterns.
3781 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
3782 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
3783 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
3784 and adjust argument order accordingly.
3785 (ix86_expand_vec_perm): Adjust caller.
3786 (expand_vec_perm_1): Likewise.
3787 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
3788 (expand_vec_perm_vpermt2_vpshub2): ... this.
3789 (ix86_expand_vec_perm_const_1): Adjust caller.
3790 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
3793 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
3794 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
3795 (vec_shl_<mode>): Remove unused expander.
3796 (avx512bw_<shift_insn><mode>3): New define_insn.
3797 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
3798 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
3800 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
3803 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
3806 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3808 PR rtl-optimization/82396
3809 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
3810 (autopref_multipass_init): Simplify initialization.
3811 (autopref_rank_data): Simplify sort order.
3812 * gcc/sched-int.h (autopref_multipass_data_): Remove
3813 multi_mem_insn_p, min_offset and max_offset.
3815 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
3818 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
3819 Check special value of flag_omit_frame_pointer.
3820 (aarch64_can_eliminate): Likewise.
3821 (aarch64_override_options_after_change_1): Simplify handling of
3822 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
3824 2017-10-24 Richard Biener <rguenther@suse.de>
3826 PR tree-optimization/82697
3827 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
3828 zero for conditional load and unconditional store.
3830 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3832 * doc/install.texi: Document bootstrap-cet.
3834 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
3837 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
3838 ENDBR instruction at function entrance if function is only
3841 2017-10-24 Jakub Jelinek <jakub@redhat.com>
3844 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
3845 patterns to better describe from which operation the CF is computed.
3846 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
3847 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
3848 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
3849 is 0, use _0 suffixed expanders instead of emitting a comparison
3852 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
3854 * config/i386/i386.md(*movsf_internal, *movdf_internal):
3855 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
3857 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
3860 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
3861 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
3862 * loop-iv.c (iv_get_reaching_def): Likewise.
3863 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
3864 variable is promoted and the partition contains undefined values.
3866 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3868 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
3870 (nios2_address_cost): Define.
3871 (nios2_legitimize_address): Recognize (exp + constant) directly.
3872 (TARGET_ADDRESS_COST): Define.
3874 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3876 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
3877 (nios2_symbolic_memory_operand_p): Declare.
3878 (nios2_split_large_constant): Declare.
3879 (nios2_split_symbolic_memory_operand): Declare.
3880 * config/nios2/nios2.c: Adjust includes.
3881 (nios2_symbolic_constant_allowed): New.
3882 (nios2_symbolic_constant_p): New.
3883 (nios2_plus_symbolic_constant_p): New.
3884 (nios2_valid_addr_expr_p): Recognize addresses involving
3886 (nios2_legitimate_address_p): Likewise, also LO_SUM.
3887 (nios2_symbolic_memory_operand_p): New.
3888 (nios2_large_constant_p): New.
3889 (nios2_split_large_constant): New.
3890 (nios2_split_plus_large_constant): New.
3891 (nios2_split_symbolic_memory_operand): New.
3892 (nios2_legitimize_address): Code refactoring. Handle addresses
3893 involving symbolic constants.
3894 (nios2_emit_move_sequence): Likewise.
3895 (nios2_print_operand): Improve error output.
3896 (nios2_print_operand_address): Handle LO_SUM.
3897 (nios2_cdx_narrow_form_p): Likewise.
3898 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
3899 operands involving symbolic constants.
3900 (movhi_internal, movsi_internal): Likewise.
3901 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
3902 (extendhisi2, extendqi<mode>2): Likewise.
3904 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3906 * tree-pass.h (PROP_rtl_split_insns): Define.
3907 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
3909 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
3911 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
3913 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3916 * target.def (const_not_ok_for_debug_p): Default to
3917 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
3918 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
3919 * targhooks.c (default_const_not_ok_for_debug_p): New function.
3920 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
3921 which targetm.const_not_ok_for_debug_p returned true.
3922 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
3924 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
3926 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
3927 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
3928 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
3930 (ix86_const_not_ok_for_debug_p): New function.
3931 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
3932 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
3934 2017-10-23 David Malcolm <dmalcolm@redhat.com>
3937 * system.h: Conditionally include "unique-ptr.h" if
3938 INCLUDE_UNIQUE_PTR is defined.
3939 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
3940 of defining INCLUDE_UNIQUE_PTR before including "system.h".
3942 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
3944 * config/rl78/rl78.md: New define_expand "subdi3".
3946 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
3949 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
3950 DF_REF_INSN if DF_REF_INSN_INFO is false.
3952 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
3954 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
3955 xmm_move instead of sse_move.
3956 (sse_store_index): New function.
3957 (ix86_register_move_cost): Be more sensible about mismatch stall;
3958 model AVX moves correctly; make difference between sse->integer and
3960 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
3961 moves; make difference between SSE and AVX.
3962 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
3963 and zmm_move. Increase size of sse load and store tables;
3964 add unaligned load and store tables; add ssemmx_to_integer.
3965 * x86-tune-costs.h: Update all entries according to real
3966 move latencies from Agner Fog's manual and chip documentation.
3968 2017-10-23 Jakub Jelinek <jakub@redhat.com>
3971 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
3972 * config/i386/constraints.md (Wf): New constraint.
3973 * config/i386/i386.md (UNSPEC_SBB): New unspec.
3974 (cmp<dwi>_doubleword): Removed.
3975 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
3976 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
3977 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
3978 expand with cmp<dwi>_doubleword. For LTU and GEU use
3979 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
3981 * common.opt (gcolumn-info): Enable by default.
3982 * doc/invoke.texi (gcolumn-info): Document new default.
3984 2017-10-23 Richard Biener <rguenther@suse.de>
3986 PR tree-optimization/82672
3987 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
3988 Fold the stmt if we propagated into it.
3990 2017-10-23 Richard Biener <rguenther@suse.de>
3992 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
3993 (bitmap_remove_expr_from_set): ... this. All callers call this
3994 for non-constant values.
3995 (bitmap_set_subtract): Rename to...
3996 (bitmap_set_subtract_expressions): ... this. Adjust and
3998 (bitmap_set_contains_value): Remove superfluous check.
3999 (bitmap_set_replace_value): Inline into single caller ...
4000 (bitmap_value_replace_in_set): ... here and simplify.
4001 (dependent_clean): Merge into ...
4002 (clean): ... this using an overload. Adjust.
4003 (prune_clobbered_mems): Adjust.
4004 (compute_antic_aux): Likewise.
4005 (compute_partial_antic_aux): Likewise.
4007 2017-10-23 Richard Biener <rguenther@suse.de>
4009 PR tree-optimization/82129
4011 2017-08-01 Richard Biener <rguenther@suse.de>
4013 PR tree-optimization/81181
4014 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
4015 (compute_antic): ... end of iteration here.
4017 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4019 * target.def (starting_frame_offset): New hook.
4020 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
4021 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
4022 * doc/tm.texi.in: Regenerate.
4023 * hooks.h (hook_hwi_void_0): Declare.
4024 * hooks.c (hook_hwi_void_0): New function.
4025 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
4026 STARTING_FRAME_OFFSET.
4027 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
4028 * reload1.c (reload): Likewise.
4029 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
4030 instead of STARTING_FRAME_OFFSET.
4031 * function.c (try_fit_stack_local): Likewise.
4032 (assign_stack_local_1): Likewise
4033 (instantiate_virtual_regs): Likewise.
4034 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
4035 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
4036 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
4037 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
4038 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
4039 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
4040 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
4041 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
4042 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
4043 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
4044 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
4045 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
4046 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
4047 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
4048 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
4049 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
4050 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
4051 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
4052 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
4053 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
4054 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
4055 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
4056 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
4057 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
4058 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
4059 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
4060 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
4061 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
4062 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
4063 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
4064 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
4065 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
4066 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
4067 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
4068 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
4069 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
4070 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
4071 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
4072 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
4073 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
4074 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
4075 * config/avr/avr.c (avr_starting_frame_offset): Make static and
4076 return a HOST_WIDE_INT.
4077 (avr_builtin_setjmp_frame_value): Use it instead of
4078 STARTING_FRAME_OFFSET.
4079 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4080 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
4081 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
4083 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4084 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
4085 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
4086 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4087 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
4088 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
4089 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4090 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
4091 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
4092 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4093 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
4094 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
4096 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4097 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
4098 * config/mips/mips.c (mips_compute_frame_info): Refer to
4099 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4100 (mips_starting_frame_offset): New function.
4101 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4102 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
4103 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
4104 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
4105 and return a HOST_WIDE_INT.
4106 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4107 (mmix_initial_elimination_offset): Refer to
4108 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
4109 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
4110 * config/pa/pa.c (pa_starting_frame_offset): New function.
4111 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
4112 (pa_expand_prologue): Likewise.
4113 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4114 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
4115 !FRAME_GROWS_DOWNWARD handling to...
4116 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4117 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
4118 !FRAME_GROWS_DOWNWARD handling to...
4119 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4120 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
4121 !FRAME_GROWS_DOWNWARD handling to...
4122 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4123 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
4125 (rs6000_starting_frame_offset): New function.
4126 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
4127 !FRAME_GROWS_DOWNWARD handling to...
4128 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4129 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
4130 !FRAME_GROWS_DOWNWARD handling to...
4131 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4132 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
4133 !FRAME_GROWS_DOWNWARD handling to...
4134 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
4135 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
4136 (rs6000_starting_frame_offset): New function.
4137 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
4138 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
4139 * config/vax/vax.c (vax_starting_frame_offset): New function.
4140 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
4141 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4142 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
4143 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
4144 (TARGET_STARTING_FRAME_OFFSET): Redefine.
4145 * system.h (STARTING_FRAME_OFFSET): Poison.
4147 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4149 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
4150 SCALAR_TYPE_MODE instead of TYPE_MODE.
4152 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4153 Alan Hayward <alan.hayward@arm.com>
4154 David Sherwood <david.sherwood@arm.com>
4156 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
4158 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4159 Alan Hayward <alan.hayward@arm.com>
4160 David Sherwood <david.sherwood@arm.com>
4162 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
4164 2017-10-23 Richard Biener <rguenther@suse.de>
4166 PR tree-optimization/82129
4167 * tree-ssa-pre.c (bitmap_set_and): Remove.
4168 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
4169 canonicalizing expressions in the set to those with lowest
4170 ID rather than taking that from the first edge.
4172 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
4174 * combine.c (rtx_equal_for_field_assignment_p): Use
4175 byte_lowpart_offset.
4177 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4178 Alan Hayward <alan.hayward@arm.com>
4179 David Sherwood <david.sherwood@arm.com>
4181 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
4182 to a SUBREG_PROMOTED_VAR.
4184 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4185 Alan Hayward <alan.hayward@arm.com>
4186 David Sherwood <david.sherwood@arm.com>
4188 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
4189 (expand_debug_source_expr): Likewise.
4190 * combine.c (combine_simplify_rtx): Likewise.
4191 * cse.c (fold_rtx): Likewise.
4192 * optabs.c (expand_float): Likewise.
4193 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4194 (simplify_binary_operation_1): Likewise.
4196 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4197 Alan Hayward <alan.hayward@arm.com>
4198 David Sherwood <david.sherwood@arm.com>
4200 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
4201 (record_promoted_value): Likewise.
4202 * expr.c (expand_expr_real_2): Likewise.
4203 * ree.c (update_reg_equal_equiv_notes): Likewise.
4204 (combine_set_extension): Likewise.
4205 * rtlanal.c (low_bitmask_len): Likewise.
4206 * simplify-rtx.c (neg_const_int): Likewise.
4207 (simplify_binary_operation_1): Likewise.
4209 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4210 Alan Hayward <alan.hayward@arm.com>
4211 David Sherwood <david.sherwood@arm.com>
4213 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
4214 * regcprop.c (maybe_mode_change): Likewise.
4215 * reload1.c (alter_reg): Likewise.
4217 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4219 * inchash.h (inchash::hash::add_wide_int): New function.
4220 * lto-streamer-out.c (hash_tree): Use it.
4222 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
4224 * inchash.h (inchash::hash::add_wide_int): Rename to...
4225 (inchash::hash::add_hwi): ...this.
4226 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
4227 (polymorphic_call_target_hasher::hash): Likewise.
4228 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
4229 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
4230 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
4231 * lto-streamer-out.c (hash_tree): Likewise.
4232 * optc-save-gen.awk: Likewise.
4233 * tree.c (add_expr): Likewise.
4235 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4238 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
4239 for ordered inequality comparisons even with TARGET_IEEE_FP.
4241 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
4244 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
4245 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4246 Expand with cmp<dwi>_doubleword.
4248 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4250 * extend.texi: Add x86 specific to 'nocf_check' attribute.
4251 List CET intrinsics.
4252 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
4253 specific to -fcf-protection option.
4255 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4257 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
4258 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
4259 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
4260 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
4261 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
4262 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
4263 (extra_objs): Add cet.o for Linux/x86 targets.
4264 (tmake_file): Add i386/t-cet for Linux/x86 targets.
4265 * config/i386/cet.c: New file.
4266 * config/i386/cetintrin.h: Likewise.
4267 * config/i386/t-cet: Likewise.
4268 * config/i386/cpuid.h (bit_SHSTK): New.
4269 (bit_IBT): Likewise.
4270 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
4271 pass IBT and SHSTK bits.
4272 * config/i386/i386-builtin-types.def
4273 (VOID_FTYPE_UNSIGNED_PVOID): New.
4274 (VOID_FTYPE_UINT64_PVOID): Likewise.
4275 * config/i386/i386-builtin.def: Add CET intrinsics.
4276 * config/i386/i386-c.c (ix86_target_macros_internal): Add
4277 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
4278 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
4279 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
4281 * config/i386/i386.c (rest_of_insert_endbranch): New.
4282 (pass_data_insert_endbranch): Likewise.
4283 (pass_insert_endbranch): Likewise.
4284 (make_pass_insert_endbranch): Likewise.
4285 (ix86_notrack_prefixed_insn_p): Likewise.
4286 (ix86_target_string): Add -mibt, -mshstk flags.
4287 (ix86_option_override_internal): Add flag_cf_protection
4289 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
4290 (ix86_print_operand): Add 'notrack' prefix output.
4291 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
4292 (ix86_expand_builtin): Expand CET intrinsics.
4293 (x86_output_mi_thunk): Add 'endbranch' instruction.
4294 * config/i386/i386.h (TARGET_IBT): New.
4295 (TARGET_IBT_P): Likewise.
4296 (TARGET_SHSTK): Likewise.
4297 (TARGET_SHSTK_P): Likewise.
4298 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
4299 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
4300 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
4301 (builtin_setjmp_setup): New pattern.
4302 (builtin_longjmp): Likewise.
4303 (rdssp<mode>): Likewise.
4304 (incssp<mode>): Likewise.
4305 (saveprevssp): Likewise.
4306 (rstorssp): Likewise.
4307 (wrss<mode>): Likewise.
4308 (wruss<mode>): Likewise.
4309 (setssbsy): Likewise.
4310 (clrssbsy): Likewise.
4311 (nop_endbr): Likewise.
4312 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
4314 * config/i386/immintrin.h: Include <cetintrin.h>.
4315 * config/i386/linux-common.h
4316 (file_end_indicate_exec_stack_and_cet): New prototype.
4317 (TARGET_ASM_FILE_END): New.
4319 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4321 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
4322 latencies instead of having separate table; make difference between
4323 integer and float costs.
4324 * i386.h (processor_costs): Remove scalar_stmt_cost,
4325 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
4326 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
4328 * x86-tune-costs.h: Remove entries which has been removed in
4329 procesor_costs from all tables; make cond_taken_branch_cost
4330 and cond_not_taken_branch_cost COST_N_INSNS based.
4332 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
4334 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
4336 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4338 * config/i386/i386.md (isa): Remove fma_avx512f.
4339 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
4340 <avx512>_fmadd_<mode>_mask3<round_name>,
4341 <avx512>_fmsub_<mode>_mask<round_name>,
4342 <avx512>_fmsub_<mode>_mask3<round_name>,
4343 <avx512>_fnmadd_<mode>_mask<round_name>,
4344 <avx512>_fnmadd_<mode>_mask3<round_name>,
4345 <avx512>_fnmsub_<mode>_mask<round_name>,
4346 <avx512>_fnmsub_<mode>_mask3<round_name>,
4347 <avx512>_fmaddsub_<mode>_mask<round_name>,
4348 <avx512>_fmaddsub_<mode>_mask3<round_name>,
4349 <avx512>_fmsubadd_<mode>_mask<round_name>,
4350 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
4351 (*vec_widen_umult_even_v16si<mask_name>,
4352 *vec_widen_smult_even_v16si<mask_name>): Likewise.
4353 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
4355 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4357 * extend.texi: Add 'nocf_check' documentation.
4358 * gimple.texi: Add second parameter to
4359 gimple_build_call_from_tree.
4360 * invoke.texi: Add -fcf-protection documentation.
4361 * rtl.texi: Add REG_CALL_NOTRACK documenation.
4363 2017-10-20 Richard Biener <rguenther@suse.de>
4365 PR tree-optimization/82473
4366 * tree-vect-loop.c (vectorizable_reduction): Properly get at
4367 the largest input type.
4369 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
4371 * c-attribs.c (handle_nocf_check_attribute): New function.
4372 (c_common_attribute_table): Add 'nocf_check' handling.
4373 * gimple-parser.c: Add second argument NULL to
4374 gimple_build_call_from_tree.
4375 * attrib.c (comp_type_attributes): Check nocf_check attribute.
4376 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
4378 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
4379 * common.opt: Add fcf-protection flag.
4380 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
4381 * flag-types.h: Add enum cf_protection_level.
4382 * gimple.c (gimple_build_call_from_tree): Add second parameter.
4383 Add 'nocf_check' attribute propagation to gimple call.
4384 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
4385 (gimple_build_call_from_tree): Update prototype.
4386 (gimple_call_nocf_check_p): New function.
4387 (gimple_call_set_nocf_check): Likewise.
4388 * gimplify.c: Add second argument to gimple_build_call_from_tree.
4389 * ipa-icf.c: Add nocf_check attribute in statement hash.
4390 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
4391 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
4392 * toplev.c (process_options): Add flag_cf_protection handling.
4394 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4396 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
4398 2017-10-20 Richard Biener <rguenther@suse.de>
4400 PR tree-optimization/82603
4401 * tree-if-conv.c (predicate_mem_writes): Make sure to only
4402 remove false predicated stores.
4404 2017-10-20 Richard Biener <rguenther@suse.de>
4406 * graphite-isl-ast-to-gimple.c
4407 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
4408 Remove return value and simplify, dump copied stmt after lhs
4410 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
4411 Reduce dump verbosity.
4412 (gsi_insert_earliest): Likewise.
4413 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
4414 * graphite.c (print_global_statistics): Adjust dumping.
4415 (print_graphite_scop_statistics): Likewise.
4416 (print_graphite_statistics): Do not dump loops here.
4417 (graphite_transform_loops): But here.
4419 2017-10-20 Nicolas Roche <roche@adacore.com>
4421 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
4422 * configure: Regenerate.
4424 2017-10-20 Jakub Jelinek <jakub@redhat.com>
4427 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
4428 functions when optimizing replace GIMPLE_RETURN stmts with
4429 calls to __builtin_unreachable ().
4432 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
4433 for -fsanitize=thread link of executables.
4434 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
4435 link of executables.
4438 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
4440 (<shift_insn><mode>3<mask_name>): Change the last of the 3
4441 define_insns for logical vector shifts to use VI248_AVX512BW
4442 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
4443 condition, useless isa and prefix attributes. Change the first
4444 2 of these define_insns to ...
4445 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
4446 define_insn for avx512vl.
4447 (<shift_insn><mode>3): ... and this, new define_insn without
4448 masking for non-avx512vl.
4451 * config/i386/sse.md (*andnot<mode>3,
4452 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
4453 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
4454 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
4455 not applied use empty suffix even for TARGET_AVX512VL.
4456 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
4457 is applied, supply evex,evex or evex,evex,evex instead of just
4460 2017-10-20 Julia Koval <julia.koval@intel.com>
4462 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
4463 (OPTION_MASK_ISA_GFNI_UNSET): New.
4464 (ix86_handle_option): Handle OPT_mgfni.
4465 * config/i386/cpuid.h (bit_GFNI): New.
4466 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
4467 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
4468 * config/i386/i386.c (ix86_target_string): Add -mgfni.
4469 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
4470 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
4471 * config/i386/i386.opt: Add mgfni.
4473 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
4475 * config/msp430/msp430.c (msp430_option_override): Disable
4476 -fdelete-null-pointer-checks.
4477 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
4479 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4481 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
4482 of x87 and SSE instructions.
4484 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
4486 * asan.c (create_cond_insert_point): Do not update edge count.
4487 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
4488 (afdo_propagate_circuit): Likewise.
4489 (afdo_calculate_branch_prob): Likewise.
4490 (afdo_annotate_cfg): Likewise.
4491 * basic-block.h (struct edge_def): Remove count.
4492 (edge_def::count): New accessor.
4493 * bb-reorder.c (rotate_loop): Update.
4494 (find_traces_1_round): Update.
4495 (connect_traces): Update.
4496 (sanitize_hot_paths): Update.
4497 * cfg.c (unchecked_make_edge): Update.
4498 (make_single_succ_edge): Update.
4499 (check_bb_profile): Update.
4500 (dump_edge_info): Update.
4501 (update_bb_profile_for_threading): Update.
4502 (scale_bbs_frequencies_int): Update.
4503 (scale_bbs_frequencies_gcov_type): Update.
4504 (scale_bbs_frequencies_profile_count): Update.
4505 (scale_bbs_frequencies): Update.
4506 * cfganal.c (connect_infinite_loops_to_exit): Update.
4507 * cfgbuild.c (compute_outgoing_frequencies): Update.
4508 (find_many_sub_basic_blocks): Update.
4509 * cfgcleanup.c (try_forward_edges): Update.
4510 (try_crossjump_to_edge): Update
4511 * cfgexpand.c (expand_gimple_cond): Update
4512 (expand_gimple_tailcall): Update
4513 (construct_exit_block): Update
4514 * cfghooks.c (verify_flow_info): Update
4515 (redirect_edge_succ_nodup): Update
4516 (split_edge): Update
4517 (make_forwarder_block): Update
4518 (duplicate_block): Update
4519 (account_profile_record): Update
4520 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4521 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4522 * cfgloopmanip.c (scale_loop_profile): Update.
4524 (lv_adjust_loop_entry_edge): Update.
4525 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
4526 (force_nonfallthru_and_redirect): Update.
4527 (purge_dead_edges): Update.
4528 (rtl_flow_call_edges_add): Update.
4529 * cgraphunit.c (init_lowered_empty_function): Update.
4530 (cgraph_node::expand_thunk): Update.
4531 * gimple-pretty-print.c (dump_probability): Update.
4532 (dump_edge_probability): Update.
4533 * gimple-ssa-isolate-paths.c (isolate_path): Update.
4534 * haifa-sched.c (sched_create_recovery_edges): Update.
4535 * hsa-gen.c (convert_switch_statements): Update.
4536 * ifcvt.c (dead_or_predicable): Update.
4537 * ipa-inline-transform.c (inline_transform): Update.
4538 * ipa-split.c (split_function): Update.
4539 * ipa-utils.c (ipa_merge_profiles): Update.
4540 * loop-doloop.c (add_test): Update.
4541 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
4542 * lto-streamer-in.c (input_cfg): Update.
4543 (input_function): Update.
4544 * lto-streamer-out.c (output_cfg): Update.
4545 * modulo-sched.c (sms_schedule): Update.
4546 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4547 * predict.c (maybe_hot_edge_p): Update.
4548 (unlikely_executed_edge_p): Update.
4549 (probably_never_executed_edge_p): Update.
4550 (dump_prediction): Update.
4551 (drop_profile): Update.
4552 (propagate_unlikely_bbs_forward): Update.
4553 (determine_unlikely_bbs): Update.
4554 (force_edge_cold): Update.
4555 * profile.c (compute_branch_probabilities): Update.
4556 * reg-stack.c (better_edge): Update.
4557 * shrink-wrap.c (handle_simple_exit): Update.
4558 * tracer.c (better_p): Update.
4559 * trans-mem.c (expand_transaction): Update.
4560 (split_bb_make_tm_edge): Update.
4561 * tree-call-cdce.c: Update.
4562 * tree-cfg.c (gimple_find_sub_bbs): Update.
4563 (gimple_split_edge): Update.
4564 (gimple_duplicate_sese_region): Update.
4565 (gimple_duplicate_sese_tail): Update.
4566 (gimple_flow_call_edges_add): Update.
4567 (insert_cond_bb): Update.
4568 (execute_fixup_cfg): Update.
4569 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
4570 * tree-complex.c (expand_complex_div_wide): Update.
4571 * tree-eh.c (lower_resx): Update.
4572 (unsplit_eh): Update.
4573 (cleanup_empty_eh_move_lp): Update.
4574 * tree-inline.c (copy_edges_for_bb): Update.
4575 (freqs_to_counts): Update.
4576 (copy_cfg_body): Update.
4577 * tree-ssa-dce.c (remove_dead_stmt): Update.
4578 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4579 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
4580 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
4581 (unloop_loops): Update.
4582 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4583 * tree-ssa-loop-split.c (connect_loops): Update.
4584 (split_loop): Update.
4585 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4586 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
4587 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
4588 * tree-ssa-reassoc.c (branch_fixup): Update.
4589 * tree-ssa-tail-merge.c (replace_block_by): Update.
4590 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
4591 (compute_path_counts): Update.
4592 (update_profile): Update.
4593 (recompute_probabilities): Update.
4594 (update_joiner_offpath_counts): Update.
4595 (estimated_freqs_path): Update.
4596 (freqs_to_counts_path): Update.
4597 (clear_counts_path): Update.
4598 (ssa_fix_duplicate_block_edges): Update.
4599 (duplicate_thread_path): Update.
4600 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
4601 (case_bit_test_cmp): Update.
4602 (collect_switch_conv_info): Update.
4603 (gen_inbound_check): Update.
4604 (do_jump_if_equal): Update.
4605 (emit_cmp_and_jump_insns): Update.
4606 * tree-tailcall.c (decrease_profile): Update.
4607 (eliminate_tail_call): Update.
4608 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4609 (vect_do_peeling): Update.
4610 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4611 * ubsan.c (ubsan_expand_null_ifn): Update.
4612 (ubsan_expand_ptr_ifn): Update.
4613 * value-prof.c (gimple_divmod_fixed_value): Update.
4614 (gimple_mod_pow2): Update.
4615 (gimple_mod_subtract): Update.
4616 (gimple_ic): Update.
4617 (gimple_stringop_fixed_value): Update.
4619 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4622 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
4624 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
4626 PR rtl-optimization/82395
4627 * ira-color.c (allocno_priority_compare_func): Fix comparison step
4628 based on non_spilled_static_chain_regno_p.
4630 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
4632 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
4633 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
4634 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
4636 2017-10-19 Martin Sebor <msebor@redhat.com>
4638 PR tree-optimization/82596
4639 * tree.c (array_at_struct_end_p): Handle STRING_CST.
4641 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4643 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
4644 (get_mem_refs_of_builtin_call): Likewise.
4645 * builtins.c (expand_builtin_apply): Adjust call to
4646 allocate_dynamic_stack_space.
4647 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
4648 the third argument to allocate_dynamic_stack_space, otherwise -1.
4649 (expand_builtin): Deal with all alloca variants.
4650 (is_inexpensive_builtin): Likewise.
4651 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
4652 * calls.c (special_function_p): Deal with all alloca variants.
4653 (initialize_argument_information): Adjust call to
4654 allocate_dynamic_stack_space.
4655 (expand_call): Likewise.
4656 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
4657 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
4658 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
4659 use it for the stack usage computation.
4660 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
4661 * function.c (gimplify_parameters): Call build_alloca_call_expr.
4662 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
4663 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
4664 (in_loop_p): Remove first argument and useless check.
4665 (pass_walloca::execute): Remove useless test and adjust call to above.
4666 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
4667 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
4668 (gimplify_call_expr): Deal with all alloca variants.
4669 * hsa-gen.c (gen_hsa_alloca): Likewise.
4670 (gen_hsa_insns_for_call): Likewise.
4671 * ipa-pure-const.c (special_builtin_state): Likewise.
4672 * tree-chkp.c (chkp_build_returned_bound): Likewise.
4673 * tree-object-size.c (alloc_object_size): Likewise.
4674 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
4675 (call_may_clobber_ref_p_1): Likewise.
4676 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
4677 (ccp_fold_stmt): Likewise.
4678 (optimize_stack_restore): Likewise.
4679 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
4680 (mark_all_reaching_defs_necessary_1): Likewise.
4681 (propagate_necessity): Likewise.
4682 (eliminate_unnecessary_stmts): Likewise.
4683 * tree.c (build_common_builtin_nodes): Build
4684 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
4685 (build_alloca_call_expr): New function.
4686 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
4687 (CASE_BUILT_IN_ALLOCA): Likewise.
4688 (build_alloca_call_expr): Declare.
4689 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
4691 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
4694 * dwarf2out.c (new_die_raw): New static inline function.
4695 (new_die): Use it to create the DIE.
4696 (add_AT_external_die_ref): Likewise.
4697 (clone_die): Likewise.
4698 (clone_as_declaration): Likewise.
4699 (dwarf2out_vms_debug_main_pointer): Likewise.
4700 (base_type_die): Likewise. Remove early return for corner cases.
4701 Do not call add_pubtype on the DIE here.
4702 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
4703 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
4704 typedefs for base types with DW_AT_endianity. Make sure a DIE with
4705 native order exists for base types, attach the DIE manually and call
4706 add_pubtype on it. Do not equate a reverse order DIE to the type.
4708 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
4710 * config/arm/arm.c (align_ok_ldrd_strd): New function.
4711 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
4713 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
4715 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4717 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
4718 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
4719 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
4720 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
4721 * opts.c (sanitizer_opts): Add builtin.
4722 * ubsan.c (instrument_builtin): New function.
4723 (pass_ubsan::execute): Call it.
4724 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
4725 * doc/invoke.texi: Document -fsanitize=builtin.
4727 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
4728 builtins, store max (log2 (align), 0) into uchar field instead of
4729 align into uptr field.
4730 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
4731 store uchar 0 field instead of uptr 0 field.
4732 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
4733 instead of passing one address of struct with 2 locations pass
4734 two addresses of structs with 1 location each.
4735 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
4736 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
4737 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
4738 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
4739 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
4740 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
4741 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
4742 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
4744 2017-10-19 Martin Liska <mliska@suse.cz>
4747 * file-find.c (remove_prefix): Remove.
4748 * file-find.h (remove_prefix): Likewise.
4749 * gcc-ar.c: Remove smartness of lookup.
4751 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
4753 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
4754 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
4755 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
4757 2017-10-19 Jakub Jelinek <jakub@redhat.com>
4760 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
4761 (setcc + and to xor + setcc): New peephole2.
4763 2017-10-19 Tom de Vries <tom@codesourcery.com>
4765 * doc/sourcebuild.texi (Test Directives, Variants of
4766 dg-require-support): Add dg-require-stack-size.
4768 2017-10-19 Martin Liska <mliska@suse.cz>
4771 * gimplify.c (gimplify_decl_expr): Do not instrument variables
4772 that have a large alignment.
4773 (gimplify_target_expr): Likewise.
4775 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
4777 PR rtl-optimization/82602
4778 * ira.c (rtx_moveable_p): Return false for volatile asm.
4780 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
4783 * config/i386/i386-modes.def (CCGZ): New CC mode.
4784 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
4785 * config/i386/predicates.md (ix86_comparison_operator):
4787 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
4788 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
4789 with double-word subtraction.
4790 (put_condition_code): Handle CCGZmode.
4792 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
4794 * wide-int.cc (debug (const wide_int &)): New.
4795 (debug (const wide_int *)): New.
4796 (debug (const widest_int &)): New.
4797 (debug (const widest_int *)): New.
4799 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
4802 * lra-constraints.c (curr_insn_transform): Use non-input operand
4803 instead of output one for matched reload.
4805 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4807 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
4808 (tree-ssa-loop-ivopts.h): New header file.
4809 (struct builtin_info): New fields.
4810 (classify_builtin_1): Compute and record base and offset parts for
4811 memset builtin partition by calling strip_offset.
4812 (offset_cmp, fuse_memset_builtins): New functions.
4813 (finalize_partitions): Fuse adjacent memset partitions by calling
4815 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
4816 Expose the interface.
4817 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
4819 2017-10-18 Bin Cheng <bin.cheng@arm.com>
4821 PR tree-optimization/82574
4822 * tree-loop-distribution.c (find_single_drs): New parameter. Check
4823 that data reference must be executed exactly once per iteration
4824 against the outermost loop in nest.
4825 (classify_partition): Update call to above function.
4827 2017-10-18 Richard Biener <rguenther@suse.de>
4829 PR tree-optimization/82591
4830 * graphite.c (graphite_transform_loops): Move code gen message
4832 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4833 Here. Handle scop_to_isl_ast failing.
4834 (scop_to_isl_ast): Limit the number of ISL operations.
4836 2017-10-18 Richard Biener <rguenther@suse.de>
4838 * graphite-isl-ast-to-gimple.c
4839 (translate_isl_ast_to_gimple::set_rename): Simplify.
4840 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
4841 (graphite_copy_stmts_from_block): ... here.
4842 (copy_bb_and_scalar_dependences): Simplify.
4843 (add_parameters_to_ivs_params): Canonicalize.
4844 (generate_entry_out_of_ssa_copies): Simplify.
4845 * graphite-sese-to-poly.c (extract_affine_name): Simplify
4846 by passing in ISL dimension.
4847 (parameter_index_in_region_1): Rename to ...
4848 (parameter_index_in_region): ... this.
4849 (extract_affine): Adjust assert, pass down parameter index.
4850 (add_param_constraints): Use range-info when available.
4851 (build_scop_context): Adjust.
4852 * sese.c (new_sese_info): Adjust.
4853 (free_sese_info): Likewise.
4854 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
4855 Remove unused typedefs.
4856 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
4858 2017-10-18 Martin Liska <mliska@suse.cz>
4860 * combine.c (simplify_compare_const): Add gcc_fallthrough.
4862 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4864 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
4865 (s390_sched_init): Do not reset s390_sched_state if we entered the
4866 current basic block via a fallthru edge and all others are unlikely.
4868 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
4870 * config/s390/s390.c (NUM_SIDES): New variable.
4871 (LONGRUNNING_THRESHOLD): New variable.
4872 (LATENCY_FACTOR): New variable.
4873 (s390_sched_score): Decrease score for long-running instructions on
4875 (s390_sched_variable_issue): Perform bookkeeping for long-running
4878 2017-10-18 Richard Biener <rguenther@suse.de>
4880 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
4881 Simplify with removal of the parameter rename map.
4882 (set_rename): Likewise.
4883 (should_copy_to_new_region): Likewise.
4884 (graphite_copy_stmts_from_block): Likewise.
4885 (copy_bb_and_scalar_dependences): Remove initialization of
4886 unused copied_bb_map.
4888 (copy_internal_parameters): Likewise.
4889 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
4890 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
4891 Use INTEGRAL_TYPE_P.
4892 (parameter_index_in_region_1): Rename to ...
4893 (assign_parameter_index_in_region): ... this. Assert we have
4894 a parameter we handle.
4895 (scan_tree_for_params): Adjust.
4896 * sese.h (parameter_rename_map_t): Remove.
4897 (struct sese_info_t): Remove unused parameter_rename_map and
4898 copied_bb_map members.
4899 * sese.c (new_sese_info): Adjust.
4900 (free_sese_info): Likewise.
4902 2017-10-18 Martin Liska <mliska@suse.cz>
4905 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
4906 on an abnormal edge.
4908 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
4910 * doc/invoke.texi (ffunction-sections and fdata-sections):
4913 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4915 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
4916 the use statement can throw internally.
4918 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
4920 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
4921 any RTX present on the RHS of a SET.
4922 * compare-elim.c (try_eliminate_compare): Restore comment.
4924 2017-10-17 Jakub Jelinek <jakub@redhat.com>
4926 * langhooks.h (struct lang_hooks): Document that tree_size langhook
4927 may be also called on tcc_type nodes.
4928 * langhooks.c (lhd_tree_size): Likewise.
4930 2017-10-17 David Malcolm <dmalcolm@redhat.com>
4932 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
4933 format_warning_at_substring.
4934 (maybe_warn): Convert source_range * param to a location_t. Pass
4935 UNKNOWN_LOCATION rather than NULL to fmtwarn.
4936 (format_directive): Remove code to extract source_ranges and
4937 source_range * in favor of just a location_t.
4938 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
4940 * substring-locations.c (format_warning_va): Convert
4941 source_range * param to a location_t.
4942 (format_warning_at_substring): Likewise.
4943 * substring-locations.h (format_warning_va): Likewise.
4944 (format_warning_at_substring): Likewise.
4946 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
4948 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
4950 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
4951 and scatter/gather ops.
4953 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
4954 vec_gather_load and vec_scatter_store.
4955 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
4956 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
4957 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
4958 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
4959 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
4960 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
4962 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
4964 * reg-stack.c (compare_for_stack_reg): Add bool argument.
4965 Detect FTST instruction and handle its register pops. Only pop
4966 second operand if can_pop_second_op is true.
4967 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
4968 set can_pop_second_op to false in the compare_for_stack_reg call.
4970 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
4971 output_fp_compare for stack register operands.
4972 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
4973 instructions here. Do not emit stack register pops here. Assert
4974 that FCOMPP pops next to top stack register. Rewrite function.
4976 2017-10-17 Nathan Sidwell <nathan@acm.org>
4979 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
4980 use DECL_ASSEMBLER_NAME_RAW.
4983 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
4986 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
4987 Wilco Dijkstra <wilco.dijkstra@arm.com>
4989 * builtins.c (expand_builtin_update_setjmp_buf): Add a
4990 converstion to Pmode from the buf_addr.
4992 2017-10-17 Richard Biener <rguenther@suse.de>
4994 * graphite-dependences.c (scop_get_reads_and_writes): Change
4995 output parameters to references.
4997 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
4999 PR 71026/tree-optimization
5000 * fold-const.c (distribute_real_division): Removed.
5001 (fold_binary_loc): Remove calls to distribute_real_divison.
5003 2017-10-17 Richard Biener <rguenther@suse.de>
5005 * graphite-scop-detection.c
5006 (scop_detection::stmt_has_simple_data_refs_p): Always use
5007 the full nest as region.
5008 (try_generate_gimple_bb): Likewise.
5009 * sese.c (scalar_evolution_in_region): Simplify now that
5010 SCEV can handle instantiation in regions.
5011 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
5012 in the non-loop part of a function if requested.
5014 2017-10-17 Richard Biener <rguenther@suse.de>
5016 PR tree-optimization/82563
5017 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
5019 (graphite_regenerate_ast_isl): Call it.
5020 * graphite-scop-detection.c (build_scops): Remove entry edge split.
5022 2017-10-17 Jakub Jelinek <jakub@redhat.com>
5024 PR tree-optimization/82549
5025 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
5026 Formatting fixes. Instead of calling make_bit_field_ref with negative
5029 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
5031 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
5032 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
5033 _mm_maskz_reduce_ss): New.
5034 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
5035 __builtin_ia32_reducess_mask): Ditto..
5036 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
5037 * config/i386/sse.md (reduces<mode>): Renamed to ...
5038 (reduces<mode><mask_scalar_name>): ... this.
5039 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
5041 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
5042 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
5044 2017-10-16 David Malcolm <dmalcolm@redhat.com>
5046 * Makefile.in (OBJS): Add unique-ptr-tests.o.
5047 * selftest-run-tests.c (selftest::run_tests): Call
5048 selftest::unique_ptr_tests_cc_tests.
5049 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
5050 * unique-ptr-tests.cc: New file.
5052 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
5055 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5057 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5058 (make_hard_regno_born, make_hard_regno_dead): Update
5059 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
5061 2017-10-16 Jeff Law <law@redhat.com>
5063 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5065 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
5067 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
5069 2017-10-16 Olivier Hainque <hainque@adacore.com>
5071 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
5072 with_cpu if we were configured for an e500v2 target cpu name.
5074 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
5076 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
5077 * doc/invoke.texi: Document +nodsp as a valid extension for
5080 2017-10-16 Martin Liska <mliska@suse.cz>
5082 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
5083 (test_set_range): Likewise.
5084 (test_range_functions): Rename to ...
5085 (test_bit_in_range): ... this.
5086 (sbitmap_c_tests): Add new test.
5088 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5090 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
5092 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
5093 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
5095 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5097 * config/aarch64/aarch64-builtins.c
5098 (aarch64_types_quadopu_lane_qualifiers): New.
5099 (TYPES_QUADOPU_LANE): New.
5100 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
5101 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
5102 (aarch64_<sur>dot_laneq<vsi2qi>): New.
5103 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
5104 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
5105 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
5106 (Vdottype, DOTPROD): New.
5107 (sur): Add SDOT and UDOT.
5109 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5111 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
5112 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5113 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
5115 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
5116 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
5117 Enable TARGET_DOTPROD.
5118 (cortex-a75.cortex-a55): Likewise.
5119 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
5121 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5123 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
5124 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
5126 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
5128 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
5129 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
5130 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
5131 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
5132 * config/arm/types.md (neon_dot, neon_dot_q): New.
5133 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
5135 2017-10-16 Tamar Christina <tamar.christina@arm.com>
5137 * config/arm/arm.h (TARGET_DOTPROD): New.
5138 * config/arm/arm.c (arm_arch_dotprod): New.
5139 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
5140 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
5141 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
5142 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
5143 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
5144 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
5145 * doc/invoke.texi (armv8.2-a): Document dotprod
5147 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5149 * i386.c (ix86_vec_cost): New function.
5150 (ix86_rtx_costs): Handle vector operations better.
5151 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
5152 * x86-tune-costs.h: Add new costs to all tables.
5154 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
5156 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
5158 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
5159 divsd, sqrtss and sqrtsd
5160 * x86-tune-costs.h: Add new entries to all costs.
5161 (znver1_cost): Fix to match real instruction latencies.
5163 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5164 Michael Collison <michael.collison@arm.com>
5166 * compare-elim.c: Include emit-rtl.h.
5167 (can_merge_compare_into_arith): New function.
5168 (try_validate_parallel): Likewise.
5169 (try_merge_compare): Likewise.
5170 (try_eliminate_compare): Call the above when no previous clobber
5172 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
5175 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5179 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
5180 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
5184 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
5185 to be any operand_equal_p operands. For & (B - 1) require
5186 B to be power of 2. Recognize
5187 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
5189 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5192 * optabs.c (expand_memory_blockage): Fix call of
5193 targetm.have_memory_blockage.
5195 2017-10-14 Jakub Jelinek <jakub@redhat.com>
5198 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
5199 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
5200 objects to extra_objs instead of overwriting it.
5202 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5204 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
5205 Use any_fp_register_operand as operand[3] predicate. Simplify
5206 equality test for operands[2] and operands[4] memory location.
5207 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
5208 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
5209 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
5210 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
5211 any_fp_register_operand as operand[1] predicate. Simplify
5212 equality test for operands[0] and operands[3] memory location.
5213 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
5214 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
5215 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
5217 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
5219 * target-insns.def: Add memory_blockage.
5220 * optabs.c (expand_memory_blockage): New function.
5221 (expand_asm_memory_barrier): Rename ...
5222 (expand_asm_memory_blockage): ... to this.
5223 (expand_mem_thread_fence): Call expand_memory_blockage
5224 instead of expand_asm_memory_barrier.
5225 (expand_mem_singnal_fence): Ditto.
5226 (expand_atomic_load): Ditto.
5227 (expand_atomic_store): Ditto.
5228 * doc/md.texi (Standard Pattern Names For Generation):
5229 Document memory_blockage instruction pattern.
5231 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
5233 * config/rl78/rl78.c (rl78_emit_libcall): New function.
5234 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
5235 * config/rl78/rl78.md: New define_expand "adddi3".
5237 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5239 * cfghooks.c (verify_flow_info): Disable check that all probabilities
5242 2017-10-13 Jeff Law <law@redhat.com>
5244 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
5246 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5249 * internal-fn.c (expand_mul_overflow): If both operands have
5250 the same highpart of -1 or 0 and the topmost bit of lowpart
5251 is different, overflow is if res <= 0 rather than res < 0.
5253 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
5255 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
5256 TARGET_P9_VECTOR code for unaligned_load case.
5258 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
5260 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
5262 2017-10-13 Nathan Sidwell <nathan@acm.org>
5264 * tree-core.h (tree_contains_struct): Make bool.
5265 * tree.c (tree_contains_struct): Likewise.
5266 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
5267 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
5268 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
5269 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
5270 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
5272 2017-10-13 Richard Biener <rguenther@suse.de>
5274 * graphite-isl-ast-to-gimple.c
5275 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
5276 parameters and dominance check.
5277 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
5278 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
5279 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
5280 Do not update SSA form here or do intermediate IL verification.
5281 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
5282 (graphite_initialize): Remove check on the number of loops in
5283 the function and inline into graphite_transform_loops.
5284 (graphite_finalize): Inline into graphite_transform_loops.
5285 (graphite_transform_loops): Perform SSA update and IL verification
5287 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
5289 2017-10-13 Richard Biener <rguenther@suse.de>
5291 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
5292 graphite_expression_type_precision): Avoid global constructor
5294 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
5295 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
5296 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
5297 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
5298 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
5299 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
5301 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
5304 * config/i386/i386.h (ix86_red_zone_size): New.
5305 * config/i386/i386.md (push peephole2s): Replace
5306 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
5308 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5309 Alan Hayward <alan.hayward@arm.com>
5310 David Sherwood <david.sherwood@arm.com>
5312 * combine.c (can_change_dest_mode): Reject changes in
5313 REGMODE_NATURAL_SIZE.
5315 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5316 Alan Hayward <alan.hayward@arm.com>
5317 David Sherwood <david.sherwood@arm.com>
5319 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
5320 (expand_debug_source_expr): Likewise.
5321 * combine.c (combine_simplify_rtx): Likewise.
5322 * cse.c (fold_rtx): Likewise.
5323 * fwprop.c (canonicalize_address): Likewise.
5324 * targhooks.c (default_shift_truncation_mask): Likewise.
5326 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5327 Alan Hayward <alan.hayward@arm.com>
5328 David Sherwood <david.sherwood@arm.com>
5330 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
5331 (widened_mode): Likewise.
5332 (expand_unop): Likewise.
5333 * ree.c (transform_ifelse): Likewise.
5334 (merge_def_and_ext): Likewise.
5335 (combine_reaching_defs): Likewise.
5336 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
5338 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5339 Alan Hayward <alan.hayward@arm.com>
5340 David Sherwood <david.sherwood@arm.com>
5342 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
5343 * combine.c (gen_lowpart_for_combine): Likewise.
5344 * dwarf2out.c (rtl_for_decl_location): Likewise.
5345 * final.c (alter_subreg): Likewise.
5346 * rtlhooks.c (gen_lowpart_general): Likewise.
5347 (gen_lowpart_if_possible): Likewise.
5349 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
5350 Alan Hayward <alan.hayward@arm.com>
5351 David Sherwood <david.sherwood@arm.com>
5353 * calls.c (expand_call): Use subreg_lowpart_offset.
5354 * cse.c (cse_insn): Likewise.
5355 * regcprop.c (copy_value): Likewise.
5356 (copyprop_hardreg_forward_1): Likewise.
5358 2017-10-13 Jakub Jelinek <jakub@redhat.com>
5361 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
5362 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
5363 =Q constraints to +Q and into insn condition add check
5364 that operands[0] and operands[1] are equal.
5365 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
5366 =Q constraints to +Q and into insn condition add check
5367 that operands[0] is equal to either operands[1] or operands[2].
5370 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
5371 instead of handling MINUS_EXPR twice (once for each argument),
5372 canonicalize operand order and handle just once, use rtype where
5373 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
5376 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
5377 any values of __C while still being pattern recognizable as a simple
5380 2017-10-13 Richard Biener <rguenther@suse.de>
5382 PR tree-optimization/82451
5384 2017-10-02 Richard Biener <rguenther@suse.de>
5386 PR tree-optimization/82355
5387 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5388 a mapping for the enclosing loop but avoid generating one for
5390 (copy_bb_and_scalar_dependences): Remove premature codegen
5391 error on PHIs in blocks duplicated into multiple places.
5392 * graphite-scop-detection.c
5393 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5394 in the region use it as loop and nest to analyze the DR in.
5395 (try_generate_gimple_bb): Likewise.
5396 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5397 (add_loop_constraints): For blocks in a loop not in the region
5398 create a dimension with a single iteration.
5399 * sese.h (gbb_loop_at_index): Remove assert.
5401 * cfgloop.c (loop_preheader_edge): For the loop tree root
5402 return the single successor of the entry block.
5403 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5404 Reset the SCEV hashtable and niters.
5405 * graphite-scop-detection.c
5406 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
5407 assert that we only have POLYNOMIAL_CHREC that vary in loops
5408 contained in the region.
5409 (scop_detection::graphite_can_represent_expr): Adjust.
5410 (scop_detection::stmt_has_simple_data_refs_p): For loops
5411 not in the region set loop to NULL. The nest is now the
5412 entry edge to the region.
5413 (try_generate_gimple_bb): Likewise.
5414 * sese.c (scalar_evolution_in_region): Adjust for
5415 instantiate_scev change.
5416 * tree-data-ref.h (graphite_find_data_references_in_stmt):
5417 Make nest parameter the edge into the region.
5418 (create_data_ref): Likewise.
5419 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
5420 entry edge into a region and adjust instantiate_scev calls.
5421 (create_data_ref): Likewise.
5422 (graphite_find_data_references_in_stmt): Likewise.
5423 (find_data_references_in_stmt): Pass the loop preheader edge
5424 from the nest argument.
5425 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
5426 parameter the edge into the region.
5427 (instantiate_parameters): Use the loop preheader edge as entry.
5428 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
5430 (get_instantiated_value_entry): Make instantiate_below parameter
5431 the edge into the region.
5432 (instantiate_scev_name): Likewise. Adjust dominance checks,
5433 when we cannot use loop-based instantiation instantiate by
5434 walking use-def chains.
5435 (instantiate_scev_poly): Adjust.
5436 (instantiate_scev_binary): Likewise.
5437 (instantiate_scev_convert): Likewise.
5438 (instantiate_scev_not): Likewise.
5439 (instantiate_array_ref): Remove.
5440 (instantiate_scev_3): Likewise.
5441 (instantiate_scev_2): Likewise.
5442 (instantiate_scev_1): Likewise.
5443 (instantiate_scev_r): Do not blindly handle N-operand trees.
5444 Do not instantiate array-refs. Handle all constants and invariants.
5445 (instantiate_scev): Make instantiate_below parameter
5446 the edge into the region.
5447 (resolve_mixers): Use the loop preheader edge for the region
5448 parameter to instantiate_scev_r.
5449 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
5451 2017-10-13 Richard Biener <rguenther@suse.de>
5453 PR tree-optimization/82525
5454 * graphite-isl-ast-to-gimple.c
5455 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
5457 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
5458 Fail code generation when we cannot represent the isl integer.
5459 (binary_op_to_tree): Elide modulo operations that are no-ops
5460 in the type we code generate. Remove now superfluous code
5463 2017-10-13 Richard Biener <rguenther@suse.de>
5465 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
5466 (scop_detection::harmful_loop_in_region): Remove premature
5467 IV type restriction.
5468 (scop_detection::graphite_can_represent_scev): We can handle
5469 pointer IVs just fine.
5471 2017-10-13 Alan Modra <amodra@gmail.com>
5473 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
5474 "Clobbers and Scratch Registers". Add paragraph on
5475 alternative to clobbers for scratch registers and OpenBLAS
5478 2017-10-13 Alan Modra <amodra@gmail.com>
5480 * doc/extend.texi (Clobbers): Correct vax example. Delete old
5481 example of a memory input for a string of known length. Move
5482 commentary out of table. Add a number of new examples
5483 covering array memory inputs.
5485 2017-10-12 Martin Liska <mliska@suse.cz>
5487 PR tree-optimization/82493
5488 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
5489 (test_range_functions): New function.
5490 (sbitmap_c_tests): Likewise.
5491 * selftest-run-tests.c (selftest::run_tests): Run new tests.
5492 * selftest.h (sbitmap_c_tests): New function.
5494 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
5496 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
5498 * config/rs6000/amo.h: Fix spacing issue.
5500 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5503 * config/i386/i386.md (*ashl<mode>3_mask_1,
5504 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
5505 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
5508 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5510 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
5511 (profile_probability): Set max_probability
5512 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
5514 * profile-count.c (profile_probability::differs_from_p): Do not
5515 rely on max_probaiblity == 10000
5517 2017-10-12 Jeff Law <law@redhat.com>
5519 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
5522 2017-10-12 Martin Sebor <msebor@redhat.com>
5526 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
5527 (handle_alias_pairs): Call it.
5528 * common.opt (-Wattribute-alias): New option.
5529 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
5530 * doc/invoke.texi (-Wattribute-alias): Document.
5532 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
5535 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5537 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5539 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5540 (make_hard_regno_born, make_hard_regno_dead): Update
5541 bb_killed_pseudos and bb_gen_pseudos.
5543 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
5545 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
5547 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
5549 * config/alpha/alpha.c (alpha_split_conditional_move):
5550 Use std::swap instead of manually swapping.
5551 (alpha_stdarg_optimize_hook): Ditto.
5552 (alpha_canonicalize_comparison): Ditto.
5554 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5556 * tree-loop-distribution.c (struct builtin_info): New struct.
5557 (struct partition): Refactor fields into struct builtin_info.
5558 (partition_free): Free struct builtin_info.
5559 (build_size_arg_loc, build_addr_arg_loc): Delete.
5560 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
5561 information from struct builtin_info.
5562 (find_single_drs): New function refactored from classify_partition.
5563 Also moved builtin validity checks to this function.
5564 (compute_access_range, alloc_builtin): New functions.
5565 (classify_builtin_st, classify_builtin_ldst): New functions.
5566 (classify_partition): Refactor code into functions find_single_drs,
5567 classify_builtin_st and classify_builtin_ldst.
5568 (distribute_loop): Don't do runtime alias check when distributing
5570 (find_seed_stmts_for_distribution): New function.
5571 (pass_loop_distribution::execute): Refactor code finding seed
5572 stmts into above function. Support distribution for the innermost
5573 two-level loop nest. Adjust dump information.
5575 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5577 * tree-loop-distribution.c: Adjust the general comment.
5578 (NUM_PARTITION_THRESHOLD): New macro.
5579 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
5580 (classify_partition): Skip builtin pattern of loop nest's inner loop.
5581 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
5582 in call to build_partition_graph.
5583 (finalize_partitions): New parameter. Make loop distribution more
5584 conservative by fusing more partitions.
5585 (distribute_loop): Don't do runtime alias check in case of loop nest
5587 (find_seed_stmts_for_distribution): New function.
5588 (prepare_perfect_loop_nest): New function.
5589 (pass_loop_distribution::execute): Refactor code finding seed stmts
5590 and loop nest into above functions. Support loop nest distribution.
5591 Adjust dump information accordingly.
5593 2017-10-12 Bin Cheng <bin.cheng@arm.com>
5595 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
5596 and set PTYPE_SEQUENTIAL for merged partition.
5598 2017-10-12 Richard Biener <rguenther@suse.de>
5600 PR tree-optimization/69728
5602 2017-09-19 Richard Biener <rguenther@suse.de>
5604 PR tree-optimization/69728
5605 * graphite-sese-to-poly.c (schedule_error): New global.
5606 (add_loop_schedule): Handle empty domain by failing the
5608 (build_original_schedule): Handle schedule_error.
5610 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
5611 domain by returning an unchanged schedule.
5613 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5615 * genrecog.c (validate_pattern): For VEC_SELECT verify that
5616 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
5618 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
5620 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
5623 2017-10-12 Jakub Jelinek <jakub@redhat.com>
5626 * expr.c (store_field): Don't optimize away bitsize == 0 store
5627 from CALL_EXPR with addressable return type.
5629 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5631 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
5632 * config/rs6000/rs6000.md (sel): Delete mode attribute.
5633 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
5634 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
5635 TARGET_ISEL instead of TARGET_ISEL<sel>.
5637 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
5639 * config/rs6000/rs6000.c
5640 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
5642 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
5644 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
5646 (reg_or_cint_operand): Fix comment.
5647 (reg_or_zero_operand): New predicate.
5648 * config/rs6000/rs6000-protos.h (output_isel): Delete.
5649 * config/rs6000/rs6000.c (output_isel): Delete.
5650 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
5651 instead of reg_or_cint_operand. Output instruction directly (not via
5653 (isel_unsigned_<mode>): Ditto.
5654 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
5655 gpc_reg_operand. Add an instruction alternative for this. Output
5656 instruction directly.
5657 (*isel_reversed_unsigned_<mode>): Ditto.
5659 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5661 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
5662 (TARGET_CANONICALIZE_COMPARISON): Define.
5664 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
5667 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
5668 Check whether the dest is REG before adding REG_EQUIV note.
5670 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
5673 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
5675 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
5676 (make_hard_regno_born, make_hard_regno_dead): Update
5677 bb_killed_pseudos and bb_gen_pseudos.
5679 2017-10-11 Nathan Sidwell <nathan@acm.org>
5681 * incpath.h (enum incpath_kind): Name enum, prefix values.
5682 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
5683 * incpath.c (heads, tails): Use INC_MAX.
5684 (add_env_var_paths, add_standard_paths): Use incpath_kind.
5685 (merge_include_chains, split_quote_chain,
5686 register_include_chains): Update incpath_kind names.
5687 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
5688 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
5690 (add_framework_path, darwin_register_objc_includes): Likewise.
5691 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
5693 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5695 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
5696 Do not use float_operator operator predicate.
5697 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
5698 * config/i386/predicates.md (float_operator): Remove predicate.
5700 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
5702 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
5703 (*jccxf_i387): Ditto.
5704 (*jcc<mode>_i387): Ditto.
5705 (*jccu<mode>_i387): Ditto.
5706 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
5707 (*jcc_*_i387 splitters): Remove.
5708 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
5709 * config/i386/i386.c (ix86_split_fp_branch): Remove.
5710 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
5713 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5715 * profile-count.h (slow_safe_scale_64bit): New function.
5716 (safe_scale_64bit): New inline.
5717 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
5718 * profile-count.c: Include wide-int.h
5719 (slow_safe_scale_64bit): New.
5721 2017-10-11 Nathan Sidwell <nathan@acm.org>
5723 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
5724 HAS_DECL_ASSEMBLER_NAME_P.
5725 * gimple-expr.c (gimple_decl_printable_name: Check
5726 HAS_DECL_ASSEMBLER_NAME_P too.
5727 * ipa-utils.h (type_in_anonymous_namespace_p): Check
5728 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
5729 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
5730 * passes.c (rest_of_decl_compilation): Check
5731 HAS_DECL_ASSEMBLER_NAME_P too.
5732 * recog.c (verify_changes): Likewise.
5733 * tree-pretty-print.c (dump_decl_name): Likewise.
5734 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
5736 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
5737 (SET_DECL_ASSEMBLER_NAME): Use it.
5738 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
5739 (COPY_DECL_ASSEMBLER_NAME): Likewise.
5740 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
5742 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
5744 * config.gcc (i386, x86_64): Add extra objects.
5745 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
5746 (ix86_min_insn_size): Declare.
5747 (ix86_issue_rate): Declare.
5748 (ix86_adjust_cost): Declare.
5749 (ia32_multipass_dfa_lookahead): Declare.
5750 (ix86_macro_fusion_p): Declare.
5751 (ix86_macro_fusion_pair_p): Declare.
5752 (ix86_bd_has_dispatch): Declare.
5753 (ix86_bd_do_dispatch): Declare.
5754 (ix86_core2i7_init_hooks): Declare.
5755 (ix86_atom_sched_reorder): Declare.
5756 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
5757 (COSTS_N_BYTES): Move to x86-tune-costs.h.
5758 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
5759 (rip_relative_addr_p): Rename to ...
5760 (ix86_rip_relative_addr_p): ... this one; export.
5761 (memory_address_length): Update.
5762 (ix86_issue_rate): Move to x86-tune-sched.c.
5763 (ix86_flags_dependent): Move to x86-tune-sched.c.
5764 (ix86_agi_dependent): Move to x86-tune-sched.c.
5765 (exact_dependency_1): Move to x86-tune-sched.c.
5766 (exact_store_load_dependency): Move to x86-tune-sched.c.
5767 (ix86_adjust_cost): Move to x86-tune-sched.c.
5768 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
5769 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
5770 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
5771 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
5772 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
5773 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
5774 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
5775 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
5776 (min_insn_size): Rename to ...
5777 (ix86_min_insn_size): ... this one; export.
5778 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
5779 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
5780 (core2i7_first_cycle_multipass_backtrack): Move to
5781 x86-tune-sched-core.c.
5782 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
5783 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
5784 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
5785 (ix86_avoid_jump_mispredicts): Update.
5786 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
5787 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
5788 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
5789 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
5790 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
5791 (MAX_INSN): Move to ix86-tune-sched-bd.c.
5792 (MAX_IMM): Move to ix86-tune-sched-bd.c.
5793 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
5794 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
5795 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
5796 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
5797 (MAX_STORE): Move to ix86-tune-sched-bd.c.
5798 (BIG): Move to ix86-tune-sched-bd.c.
5799 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
5800 (enum insn_path): Move to ix86-tune-sched-bd.c.
5801 (get_mem_group): Move to ix86-tune-sched-bd.c.
5802 (is_cmp): Move to ix86-tune-sched-bd.c.
5803 (dispatch_violation): Move to ix86-tune-sched-bd.c.
5804 (is_branch): Move to ix86-tune-sched-bd.c.
5805 (is_prefetch): Move to ix86-tune-sched-bd.c.
5806 (init_window): Move to ix86-tune-sched-bd.c.
5807 (allocate_window): Move to ix86-tune-sched-bd.c.
5808 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
5809 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
5810 (process_end_window): Move to ix86-tune-sched-bd.c.
5811 (allocate_next_window): Move to ix86-tune-sched-bd.c.
5812 (find_constant): Move to ix86-tune-sched-bd.c.
5813 (get_num_immediates): Move to ix86-tune-sched-bd.c.
5814 (has_immediate): Move to ix86-tune-sched-bd.c.
5815 (get_insn_path): Move to ix86-tune-sched-bd.c.
5816 (get_insn_group): Move to ix86-tune-sched-bd.c.
5817 (count_num_restricted): Move to ix86-tune-sched-bd.c.
5818 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
5819 (add_insn_window): Move to ix86-tune-sched-bd.c.
5820 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
5821 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
5822 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
5823 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
5824 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
5825 (do_dispatch): Move to ix86-tune-sched-bd.c.
5826 (has_dispatch): Move to ix86-tune-sched-bd.c.
5827 * config/i386/t-i386: Add new object files.
5828 * config/i386/x86-tune-costs.h: New file.
5829 * config/i386/x86-tune-sched-atom.c: New file.
5830 * config/i386/x86-tune-sched-bd.c: New file.
5831 * config/i386/x86-tune-sched-core.c: New file.
5832 * config/i386/x86-tune-sched.c: New file.
5834 2017-10-11 Liu Hao <lh_mouse@126.com>
5836 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
5837 the generic version below instead.
5838 (should_colorize): Recognize Windows consoles as terminals
5840 * pretty-print.c [__MINGW32__] (write_all): New function.
5841 [__MINGW32__] (find_esc_head): Likewise.
5842 [__MINGW32__] (find_esc_terminator): Likewise.
5843 [__MINGW32__] (eat_esc_sequence): Likewise.
5844 [__MINGW32__] (mingw_ansi_fputs): New function that handles
5846 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
5849 2017-10-11 Richard Biener <rguenther@suse.de>
5851 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
5852 Properly call analyze_scalar_evolution with the loop of the stmt.
5854 2017-10-11 Richard Biener <rguenther@suse.de>
5856 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
5857 * tree-core.h (tree_base): Add chrec_var union member.
5858 * tree.h (CHREC_VAR): Remove.
5859 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
5860 * tree-chrec.h (build_polynomial_chrec): Adjust.
5861 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
5862 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
5864 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
5866 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
5867 * match.pd: ... here.
5868 ((T) X == (T) Y): Relax condition.
5870 2017-10-11 Bin Cheng <bin.cheng@arm.com>
5872 PR tree-optimization/82472
5873 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
5875 (break_alias_scc_partitions): Update postorder number.
5877 2017-10-11 Martin Liska <mliska@suse.cz>
5880 * opts.c (parse_no_sanitize_attribute): Do not use error_value
5882 * opts.h (parse_no_sanitize_attribute): Remove last argument.
5884 2017-10-11 Martin Liska <mliska@suse.cz>
5886 * print-rtl.c (print_insn): Move declaration of idbuf
5887 to same scope as name.
5889 2017-10-11 Martin Liska <mliska@suse.cz>
5894 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5897 2017-10-11 Martin Liska <mliska@suse.cz>
5900 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
5903 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
5905 * config.gcc (default_gnu_indirect_function): Default to yes for
5906 arm*-*-linux* with glibc.
5908 2017-10-11 Richard Biener <rguenther@suse.de>
5910 * tree-scalar-evolution.c (get_scalar_evolution): Handle
5911 default-defs and types we do not want to analyze.
5912 (interpret_loop_phi): Replace unreachable code with an assert.
5913 (compute_scalar_evolution_in_loop): Remove and inline ...
5914 (analyze_scalar_evolution_1): ... here, replacing condition with
5915 what makes the intent clearer. Remove handling of cases
5916 get_scalar_evolution now handles.
5918 2017-10-10 Jim Wilson <wilson@tuliptree.org>
5920 PR rtl-optimization/81434
5921 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
5922 comment for main loop. In sched_group_found if, also add checks for
5923 pass and min_cost_group.
5925 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
5927 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
5928 (rs6000_insn_cost): New function.
5929 * config/rs6000/rs6000.md (cost): New attribute.
5931 2017-10-10 Jakub Jelinek <jakub@redhat.com>
5932 H.J. Lu <hongjiu.lu@intel.com>
5936 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
5937 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
5938 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
5939 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
5940 to OPTION_MASK_ISA_AVX512VL - builtins that have both
5941 OPTION_MASK_ISA_MMX and some other bit set require both
5942 mmx and the ISAs without the mmx bit.
5943 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
5944 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
5945 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
5946 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
5947 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
5948 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
5949 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
5950 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
5951 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
5952 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
5953 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
5954 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
5955 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
5956 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
5957 Add OPTION_MASK_ISA_MMX.
5959 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
5961 * config.gcc (armv7*-*-freebsd*): New target.
5962 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
5964 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
5966 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
5967 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
5970 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
5972 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
5973 a template parameter.
5974 (WIDE_INT_REF_FOR): Update accordingly.
5975 * tree.h (wi::int_traits <const_tree>): Delete.
5976 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
5977 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
5978 (wi::tree_to_wide_ref): New typedef.
5979 (wi::to_wide): New function.
5980 * calls.c (get_size_range): Use wi::to_wide when operating on
5982 * cgraph.c (cgraph_node::create_thunk): Likewise.
5983 * config/i386/i386.c (ix86_data_alignment): Likewise.
5984 (ix86_local_alignment): Likewise.
5985 * dbxout.c (stabstr_O): Likewise.
5986 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
5987 * expr.c (const_vector_from_tree): Likewise.
5988 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
5989 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
5990 (fold_negate_expr_1, int_const_binop_1, const_binop)
5991 (fold_convert_const_int_from_real, optimize_bit_field_compare)
5992 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
5993 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
5994 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
5995 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
5996 (fold_not_const, round_up_loc): Likewise.
5997 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
5998 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
5999 (alloca_call_type): Likewise.
6000 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
6001 * godump.c (go_output_typedef): Likewise.
6002 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
6003 * internal-fn.c (get_min_precision): Likewise.
6004 * ipa-cp.c (ipcp_store_vr_results): Likewise.
6005 * ipa-polymorphic-call.c
6006 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
6007 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
6008 (ipa_modify_call_arguments): Likewise.
6009 * match.pd: Likewise.
6010 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
6011 * print-tree.c (print_node_brief, print_node): Likewise.
6012 * stmt.c (expand_case): Likewise.
6013 * stor-layout.c (layout_type): Likewise.
6014 * tree-affine.c (tree_to_aff_combination): Likewise.
6015 * tree-cfg.c (group_case_labels_stmt): Likewise.
6016 * tree-data-ref.c (dr_analyze_indices): Likewise.
6017 (prune_runtime_alias_test_list): Likewise.
6018 * tree-dump.c (dequeue_and_dump): Likewise.
6019 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
6020 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
6021 * tree-pretty-print.c (dump_generic_node): Likewise.
6022 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
6023 (simple_iv_with_niters): Likewise.
6024 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
6025 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
6026 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
6027 * tree-ssa-loop-niter.c (split_to_var_and_offset)
6028 (refine_value_range_using_guard, number_of_iterations_ne_max)
6029 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
6030 (get_cst_init_from_scev, record_nonwrapping_iv)
6031 (scev_var_range_cant_overflow): Likewise.
6032 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
6033 * tree-ssa-pre.c (compute_avail): Likewise.
6034 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
6035 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
6036 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
6037 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
6038 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
6039 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
6040 (dump_case_nodes, try_switch_expansion): Likewise.
6041 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
6042 (vect_do_peeling): Likewise.
6043 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
6044 * tree-vect-stmts.c (vectorizable_load): Likewise.
6045 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
6046 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
6047 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
6048 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
6049 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
6050 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
6051 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
6052 (evrp_dom_walker::before_dom_children): Likewise.
6053 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
6054 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
6055 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
6056 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
6057 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
6058 (get_range_pos_neg): Likewise.
6059 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
6060 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
6061 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
6062 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
6063 * config/avr/avr.c (avr_fold_builtin): Likewise.
6064 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
6065 * config/msp430/msp430.c (msp430_attr): Likewise.
6066 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
6067 * config/powerpcspe/powerpcspe-c.c
6068 (altivec_resolve_overloaded_builtin): Likewise.
6069 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
6070 (rs6000_expand_ternop_builtin): Likewise.
6071 * config/rs6000/rs6000-c.c
6072 (altivec_resolve_overloaded_builtin): Likewise.
6073 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
6074 (rs6000_expand_ternop_builtin): Likewise.
6075 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
6077 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6079 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
6080 when copying loop nest with only one inner loop.
6082 2017-10-10 Richard Biener <rguenther@suse.de>
6084 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
6085 blocks if SCEV is active.
6086 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
6088 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
6089 (scev_initialize): Assert we are not yet initialized.
6091 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6093 * tree-loop-distribution.c (generate_loops_for_partition): Remove
6094 inner loop's exit stmt by making it always exit the loop, otherwise
6095 we would generate an infinite empty loop.
6097 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6099 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
6100 renaming variables in new preheader if it's deleted.
6102 2017-10-10 Bin Cheng <bin.cheng@arm.com>
6104 * tree-loop-distribution.c (struct partition): Remove unused field
6105 loops of the structure.
6106 (partition_alloc, partition_free): Ditto.
6107 (build_rdg_partition_for_vertex): Ditto.
6109 2017-10-09 Jeff Law <law@redhat.com>
6111 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
6112 return type to match prototype and documentation.
6114 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6116 * config/rs6000/rs6000.c (processor_costs): Move to ...
6117 * config/rs6000/rs6000.h: ... here.
6118 (rs6000_cost): Declare.
6120 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
6122 * except.c (setjmp_fn): New global variable.
6123 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
6124 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
6125 if DONT_USE_BUILTIN_SETJMP is defined.
6127 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6129 * target.def (insn_cost): New hook.
6130 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
6131 * doc/tm.texi: Regenerate.
6132 * rtlanal.c (insn_cost): Use the new hook.
6134 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6136 * combine.c (combine_validate_cost): Compute the new insn_cost,
6137 not just pattern_cost.
6138 (try_combine): Adjust comment.
6140 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
6142 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
6144 * combine.c (uid_insn_cost): Adjust comment.
6145 (combine_validate_cost): Adjust comment. Use pattern_cost instead
6147 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
6148 * dse.c (find_shift_sequence): Ditto.
6149 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
6150 (bb_valid_for_noce_process_p): Use pattern_cost.
6151 * rtl.h (insn_rtx_cost): Delete.
6152 (pattern_cost): New prototype.
6153 (insn_cost): New prototype.
6154 * rtlanal.c (insn_rtx_cost): Rename to...
6155 (pattern_cost): ... this.
6158 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
6160 * config/i386/i386.md (*jcc_2): Remove insn pattern.
6161 (*jcc<mode>_0_r_i387): Ditto.
6162 (*jccxf_r_i387): Ditto.
6163 (*jcc<mode>_r_i387): Ditto.
6164 (*jccu<mode>_r_i387): Ditto.
6165 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
6166 (*jcc): Rename from *jcc_1.
6168 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6170 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
6171 deferred rescans after the lvx/stvx recombination pre-pass.
6173 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
6175 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
6176 memory operation instruction support.
6177 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
6178 (rs6000-ibm-aix[789]*): Likewise.
6179 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
6180 Document new functions.
6182 2017-10-09 Richard Biener <rguenther@suse.de>
6184 PR tree-optimization/82397
6185 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
6186 equality only for semantically equal trees.
6188 2017-10-09 Richard Biener <rguenther@suse.de>
6190 PR tree-optimization/82449
6191 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
6192 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
6193 allow constant addresses.
6194 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
6197 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6199 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
6202 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6205 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
6208 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6211 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
6213 2017-10-09 Jakub Jelinek <jakub@redhat.com>
6216 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
6217 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
6219 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
6221 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
6222 (WI_BINARY_PREDICATE_RESULT): Likewise.
6223 (wi::binary_traits::operator_result): New type.
6224 (wi::binary_traits::predicate_result): Likewise.
6225 (generic_wide_int::operator~, unary generic_wide_int::operator-)
6226 (generic_wide_int::operator==, generic_wide_int::operator!=)
6227 (generic_wide_int::operator&, generic_wide_int::and_not)
6228 (generic_wide_int::operator|, generic_wide_int::or_not)
6229 (generic_wide_int::operator^, generic_wide_int::operator+
6230 (binary generic_wide_int::operator-, generic_wide_int::operator*):
6232 (operator~, unary operator-, operator==, operator!=, operator&)
6233 (operator|, operator^, operator+, binary operator-, operator*): New
6235 * expr.c (get_inner_reference): Use wi::bit_and_not.
6236 * fold-const.c (fold_binary_loc): Likewise.
6237 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
6238 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
6239 (bit_value_binop): Likewise.
6240 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
6241 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
6242 (extract_range_from_binary_expr_1): Likewise.
6243 (masked_increment): Likewise.
6244 (simplify_bit_ops_using_ranges): Likewise.
6246 2017-10-09 Martin Jambor <mjambor@suse.cz>
6249 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
6250 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
6251 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
6253 (hsa_fixup_mov_insn_type): New function.
6254 (hsa_op_with_type::get_in_type): Use it.
6255 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
6256 immediates in an assert.
6257 (hsa_op_with_type::extend_int_to_32bit): New method.
6258 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
6259 types. Convert to dest type if necessary.
6260 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
6261 (reg_for_gimple_ssa): Pass false as min32int to
6262 hsa_type_for_scalar_tree_type.
6263 (gen_hsa_addr): Fixup type when creating addresable temporary.
6264 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
6265 (gen_hsa_unary_operation): Extend operands and convert to dest type if
6266 necessary. Call hsa_fixup_mov_insn_type.
6267 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
6268 extend operands and convert to dest type if necessary.
6269 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
6270 to dest type if necessary.
6271 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
6272 if conversion nt necessary and size matches.
6273 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
6274 to dest type if necessary.
6275 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
6276 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
6278 (gen_hsa_clrsb): Likewise.
6279 (gen_hsa_ffs): Likewise.
6280 (gen_hsa_divmod): Extend operands and convert to dest type if
6282 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
6284 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6286 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
6287 Remove empty default arguments. Use a brace block as output
6289 (conditional return): Ditto.
6291 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
6292 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
6294 (group_ending_nop): Ditto.
6295 (doloop_end): Ditto.
6296 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
6297 (splitters for those): Ditto.
6299 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
6301 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
6302 a conditional jump (and the compare for it) so that pc_rtx is the
6304 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
6305 for the deleted and renamed ctr<mode>_internal[234] patterns.
6306 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
6307 Delete second conditional return pattern.
6308 (ctr<mode>_internal2): Delete this second bdnz pattern.
6309 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
6310 (ctr<mode>_internal4): Delete this second bdz pattern.
6312 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6314 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
6315 (always_initialized_rtx_for_ssa_name_p): New predicate.
6316 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
6317 (finish_out_of_ssa): Free new field of SA.
6318 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
6319 * tree-ssa-coalesce.c: Include tree-ssa.h.
6320 (get_parm_default_def_partitions): Remove extern keyword.
6321 (get_undefined_value_partitions): New function.
6322 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
6323 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
6325 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
6327 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
6329 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
6331 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6333 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
6334 for targets that preffer 128bit.
6336 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
6338 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
6340 2017-10-08 Olivier Hainque <hainque@adacore.com>
6342 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
6343 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
6345 (thumb_set_return_address): Likewise.
6347 2017-10-08 Olivier Hainque <hainque@adacore.com>
6349 * common/config/arm/arm-common.c (arm_except_unwind_info):
6350 Handle DWARF2_UNWIND_INFO.
6352 2017-10-07 Michael Collison <michael.collison@arm.com>
6354 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
6357 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
6359 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
6360 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
6361 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
6362 defined, force the creation of a new block for a dispatch label.
6364 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
6366 * invoke.texi (Wsuggest-attribute=cold): Document.
6367 * common.opt (Wsuggest-attribute=cold): New
6368 * ipa-pure-const.c (warn_function_cold): New function.
6369 * predict.c (compute_function_frequency): Use it.
6370 * predict.h (warn_function_cold): Declare.
6372 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
6374 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
6377 2017-10-06 Martin Liska <mliska@suse.cz>
6379 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
6380 keyword for member functions.
6381 (struct sanopt_tree_couple): New struct.
6382 (struct sanopt_tree_couple_hash): New function.
6383 (struct sanopt_ctx): Add new hash_map.
6384 (has_dominating_ubsan_ptr_check): New function.
6385 (record_ubsan_ptr_check_stmt): Likewise.
6386 (maybe_optimize_ubsan_ptr_ifn): Likewise.
6387 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
6388 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
6390 2017-10-06 Sudakshina Das <sudi.das@arm.com>
6393 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
6394 aarch64_simd_valid_immediate on CONST_VECTORs.
6395 (aarch64_reg_or_bic_imm): Likewise.
6397 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
6399 PR rtl-optimization/82396
6400 * haifa-sched.c (ready_sort_real): Disable qsort checking.
6402 2017-10-06 Sebastian Pop <sebpop@gmail.com>
6404 * graphite-dependences.c (scop_get_reads): Move code to...
6405 (scop_get_must_writes): Move code to...
6406 (scop_get_may_writes): Move code to...
6407 (scop_get_reads_and_writes): ... here.
6408 (scop_get_dependences): Call scop_get_reads_and_writes.
6410 2017-10-06 Jakub Jelinek <jakub@redhat.com>
6412 PR tree-optimization/82434
6413 * fold-const.h (can_native_encode_type_p,
6414 can_native_encode_string_p): Remove.
6415 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
6416 don't encode anything, just return what would be otherwise returned.
6417 (native_encode_fixed, native_encode_complex, native_encode_vector):
6419 (native_encode_string): Likewise. Inline by hand
6420 can_native_encode_string_p.
6421 (can_native_encode_type_p): Remove.
6422 (can_native_encode_string_p): Remove.
6423 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
6424 STRING_CSTs using can_native_encode_string_p, test all
6425 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
6426 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
6427 argument from native_encode_expr.
6428 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
6429 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
6432 2017-10-06 Richard Biener <rguenther@suse.de>
6434 PR tree-optimization/82397
6435 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
6436 operand_equal_p but rely on data_ref_compare_tree for detecting
6438 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
6439 to match up with dr_group_sort_cmp.
6441 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6444 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
6446 * config/s390/s390-builtin-types.def: Regenerate.
6448 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6451 * config/s390/s390-builtin-types.def: Regenerate.
6452 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
6453 Change flag from B_VXE to B_VX.
6454 (s390_vec_min_dbl): Remove B_VXE flag.
6456 2017-10-06 Richard Biener <rguenther@suse.de>
6458 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
6459 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
6460 translate_isl_ast_to_gimple::is_valid_rename,
6461 translate_isl_ast_to_gimple::get_rename,
6462 translate_isl_ast_to_gimple::get_def_bb_for_const,
6463 translate_isl_ast_to_gimple::get_new_name,
6464 translate_isl_ast_to_gimple::collect_all_ssa_names,
6465 translate_isl_ast_to_gimple::copy_loop_phi_args,
6466 translate_isl_ast_to_gimple::collect_all_ssa_names,
6467 translate_isl_ast_to_gimple::copy_loop_phi_args,
6468 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
6469 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
6470 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
6471 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
6472 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
6473 translate_isl_ast_to_gimple::copy_cond_phi_args,
6474 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
6475 translate_isl_ast_to_gimple::edge_for_new_close_phis,
6476 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
6477 translate_isl_ast_to_gimple::rename_uses,
6478 translate_isl_ast_to_gimple::rename_all_uses): Remove.
6479 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
6480 (set_rename_for_each_def): Likewise.
6481 (graphite_copy_stmts_from_block): Handle debug stmt resetting
6482 here. Handle rewriting SCEV analyzable uses here.
6483 (copy_bb_and_scalar_dependences): Generate code for PHI
6485 (graphite_regenerate_ast_isl): Adjust.
6486 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
6487 (add_write, add_read): New functions.
6488 (build_cross_bb_scalars_def): Use it and simplify.
6489 (build_cross_bb_scalars_use): Likewise.
6490 (graphite_find_cross_bb_scalar_vars): Inline into...
6491 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
6492 simulating out-of-SSA. Compute liveout and add dependencies.
6493 (build_scops): Force an empty entry block.
6494 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
6496 (sese_build_liveouts): Declare.
6497 (sese_trivially_empty_bb_p): Likewise.
6498 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
6499 compute liveout and debug_liveout.
6500 (sese_bad_liveouts_use): Remove.
6501 (sese_reset_debug_liveouts_bb): Likewise.
6502 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
6503 (sese_build_liveouts): Build liveout and debug_liveout and store
6505 (new_sese_info): Adjust.
6506 (free_sese_info): Likewise.
6507 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
6508 do not build liveout here.
6509 (move_sese_in_condition): Adjust region entry.
6510 (scev_analyzable_p): Match up with chrec_apply requirements.
6511 (sese_trivially_empty_bb_p): New.
6512 * tree-into-ssa.c (get_reaching_def): Properly support generating
6513 default-defs for incremental rewrite of anonymous names.
6515 2017-10-06 Richard Biener <rguenther@suse.de>
6517 * graphite-sese-to-poly.c (extract_affine): For casts increasing
6518 precision do not perform modulo reduction.
6520 2017-10-06 Richard Biener <rguenther@suse.de>
6522 PR tree-optimization/82436
6523 * tree-vect-slp.c (vect_supported_load_permutation_p): More
6524 conservatively choose the vectorization factor when checking
6525 whether we can perform the required load permutation.
6526 (vect_transform_slp_perm_load): Assert when we may not fail.
6528 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6530 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
6531 message for incompatible -msdata=* and -mcall-* options.
6533 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6535 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
6536 rate for post-reload scheduling.
6538 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6540 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
6542 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6544 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
6545 to improve monte carlo in scimark.
6547 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
6549 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
6550 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
6551 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
6552 pentium4_cost, nocona_cost): Set reassociation width to 1.
6553 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
6554 width to 2 for fp operations and 1 otherwise.
6555 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
6557 (atom_cost): Set reassociation width to 2.
6558 (slm_cost, generic_cost): Set fp reassociation width
6559 to 2 and 1 otherwise.
6560 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
6561 (core_cost): Set fp reassociation width to 4 and vector to 2.
6562 (ix86_reassociation_width): Rewrite using cost table; special case
6563 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
6564 and TARGET_AVX128_OPTIMAL.
6565 * config/i386/i386.h (processor_costs): Add
6566 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
6567 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
6568 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
6569 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
6570 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
6571 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
6573 2017-10-05 Nathan Sidwell <nathan@acm.org>
6575 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
6577 2017-10-05 Tamar Christina <tamar.christina@arm.com>
6579 * config/arm/arm.c (arm_test_fpu_data): New.
6580 (arm_run_selftests): Call arm_test_fpu_data.
6582 2017-10-04 Nathan Sidwell <nathan@acm.org>
6584 * toplev.c (toplev::main): Remove excess parens on pretty_printer
6586 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
6588 2017-10-04 Sudakshina Das <sudi.das@arm.com>
6590 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
6591 check type for aarch64_simd_valid_immediate.
6592 (aarch64_output_simd_mov_immediate): Update prototype.
6593 (aarch64_simd_valid_immediate): Update prototype.
6594 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
6595 support for ORR-immediate.
6596 (and<mode>3): modified pattern to add support for BIC-immediate.
6597 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
6598 now checks for valid immediate for BIC and ORR based on new enum
6600 (aarch64_output_simd_mov_immediate): Function now used to output
6601 BIC/ORR imm as well based on new enum argument.
6602 * config/aarch64/constraints.md (Do): New vector immediate constraint.
6604 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
6605 (aarch64_reg_or_bic_imm): Likewise.
6607 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6609 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
6610 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
6612 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6616 PR rtl-optimization/82396
6617 * haifa-sched.c (autopref_multipass_init): Simplify
6619 (autopref_rank_data): Simplify sort order.
6620 * sched-int.h (autopref_multipass_data_): Remove
6621 multi_mem_insn_p, min_offset and max_offset.
6623 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6625 * doc/sourcebuild.texi: Document vect_peeling_profitable.
6627 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6629 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
6632 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6634 * doc/sourcebuild.texi: Document vect_long_mult.
6636 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
6638 PR tree-optimization/82413
6639 * fold-const.c (build_range_check): Use widest_int when comparing
6640 the maximum ETYPE value with HIGH.
6642 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
6644 PR rtl-optimization/82396
6645 * haifa-sched.c (autopref_multipass_init): Simplify
6647 (autopref_rank_data): Simplify sort order.
6648 * sched-int.h (autopref_multipass_data_): Remove
6649 multi_mem_insn_p, min_offset and max_offset.
6651 2017-10-04 Jakub Jelinek <jakub@redhat.com>
6653 PR tree-optimization/82381
6654 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
6655 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
6658 PR tree-optimization/82374
6659 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
6660 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
6661 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
6662 current_function_decl to the new decl.
6664 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
6666 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
6667 helper macro for IEEE float128 hardware built-in functions.
6668 (SQRTF128_ODD): Add built-in functions with the round-to-odd
6670 (TRUNCF128_ODD): Likewise.
6671 (ADDF128_ODD): Likewise.
6672 (SUBF128_ODD): Likewise.
6673 (MULF128_ODD): Likewise.
6674 (DIVF128_ODD): Likewise.
6675 (FMAF128_ODD): Likewise.
6676 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
6677 UNSPEC_TRUNC_ROUND_TO_ODD.
6678 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
6679 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
6680 floating point round to odd instructions.
6681 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
6682 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
6683 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
6684 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
6685 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
6686 (trunc<mode>sf2_hw): Change the truncate with round to odd
6687 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
6688 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
6689 to odd hardware instructions.
6690 (sub<mode>3_odd): Likewise.
6691 (mul<mode>3_odd): Likewise.
6692 (div<mode>3_odd): Likewise.
6693 (sqrt<mode>2_odd): Likewise.
6694 (fma<mode>4_odd): Likewise.
6695 (fms<mode>4_odd): Likewise.
6696 (nfma<mode>4_odd): Likewise.
6697 (nfms<mode>4_odd): Likewise.
6698 (trunc<mode>df2_odd): Change the truncate with round to odd
6699 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
6701 * doc/extend.texi (PowerPC built-in functions): Update documentation
6702 for existing IEEE float128-bit built-in functions. Add built-in
6703 functions that generate the IEEE 128-bit floating point round to
6706 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
6708 PR rtl-optimization/77729
6709 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
6710 to (X&(C1&~C2))|C2 transformations.
6712 2017-10-03 Martin Jambor <mjambor@suse.cz>
6714 PR tree-optimization/82363
6715 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
6716 mismatch, mark lacc written regardless of racc.
6718 2017-10-03 Jakub Jelinek <jakub@redhat.com>
6720 PR tree-optimization/82381
6721 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
6722 stmt_to_insert nor wheather SSA_NAMEs are default defs.
6723 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
6724 fallthrough into reassoc_stmt_dominates_stmt_p.
6727 * combine.c (combine_instructions): Don't combine in unreachable
6730 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
6733 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
6734 function to not use the have_cpu variable. Do not set cpu_index,
6735 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
6737 (rs6000_valid_attribute_p): Remove duplicate initializations of
6738 old_optimize and func_optimize.
6739 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
6740 (rs6000_activate_target_options): Make global.
6741 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
6744 2017-10-02 Jakub Jelinek <jakub@redhat.com>
6746 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
6747 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
6748 Set *poffset to 0 and *psize and *pmax_size to -1 if
6749 *poffset + *psize overflows in HOST_WIDE_INT.
6751 PR tree-optimization/82387
6752 PR tree-optimization/82388
6753 PR tree-optimization/82389
6754 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
6755 instead of live_bytes non-NULL.
6757 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
6760 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
6763 2017-10-02 Richard Biener <rguenther@suse.de>
6765 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
6766 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
6767 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
6769 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6771 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
6772 requested precision matches the type's.
6773 * calls.c (alloc_max_size): Calculate the new candidate size as
6774 a widest_int and use wi::to_widest when comparing it with the
6775 current candidate size.
6776 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
6777 zero rather than integer_zero_node.
6778 * match.pd: Check for a no-op conversion before using wi::add
6779 rather than after. Use tree_to_uhwi when summing small shift
6780 counts into an unsigned int.
6782 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
6783 Alan Hayward <alan.hayward@arm.com>
6784 David Sherwood <david.sherwood@arm.com>
6787 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
6788 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
6789 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
6790 POINTER_AND_FP_REGS.
6792 2017-10-02 Richard Biener <rguenther@suse.de>
6794 PR tree-optimization/82355
6795 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
6796 a mapping for the enclosing loop but avoid generating one for
6798 (copy_bb_and_scalar_dependences): Remove premature codegen
6799 error on PHIs in blocks duplicated into multiple places.
6800 * graphite-scop-detection.c
6801 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
6802 in the region use it as loop and nest to analyze the DR in.
6803 (try_generate_gimple_bb): Likewise.
6804 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
6805 (add_loop_constraints): For blocks in a loop not in the region
6806 create a dimension with a single iteration.
6807 * sese.h (gbb_loop_at_index): Remove assert.
6809 2017-10-01 Kevin Buettner <kevinb@redhat.com>
6811 * omp-expand.c (adjust_context_scope): New function.
6812 (expand_parallel_call): Call adjust_context_scope.
6814 2017-10-01 Jeff Law <law@redhat.com>
6816 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
6817 dom_opt_dom_walker class with direct access to private members.
6818 Add comments. Call test_for_singularity.
6819 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
6820 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
6821 m_dummy_cond anymore.
6822 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
6824 (pass_dominator:execute): Build the dummy_cond here and pass it
6825 to the dom_opt_dom_walker ctor.
6826 (test_for_singularity): New function.
6828 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
6829 Maya Rashish <coypu@sdf.org>
6831 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
6832 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
6833 (alpha*-*-netbsd*) Use nbsd_tm_file.
6834 (arm*-*-netbsdelf*) Likewise.
6835 (i[34567]86-*-netbsdelf*) Likewise.
6836 (x86_64-*-netbsd*) Likewise.
6837 (mips*-*-netbsd*) Likewise.
6838 (powerpc-*-netbsd*) Likewise.
6839 (sh*-*-netbsd*) Likewise.
6840 (sparc-*-netbsdelf*) Likewise.
6841 (sparc64-*-netbsd*) Likewise.
6842 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
6844 (vax-*-netbsdelf*) Likewise.
6845 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
6846 (UINT_FAST8_TYPE) Likewise.
6847 (INT_FAST16_TYPE) Check CHAR_FAST16.
6848 (UINT_FAST16_TYPE) Likewise.
6850 2017-09-30 Jakub Jelinek <jakub@redhat.com>
6853 * config/i386/i386.md
6854 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
6855 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
6856 *divmodsi4_zext_2): New define_insn_and_split.
6857 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
6858 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
6859 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
6860 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
6861 New define_insn_and_split.
6862 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
6863 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
6864 operands[1] having DImode when mode is SImode.
6866 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
6867 always SImode for DIV and MOD in REG_EQUAL notes.
6869 2017-09-29 Yury Gribov <tetra2005@gmail.com>
6872 * match.pd: Fix handling of NaNs in pattern.
6874 2017-09-29 Jeff Law <law@redhat.com>
6876 * sbitmap.c (bitmap_bit_in_range_p): New function.
6877 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
6878 * tree-ssa-dse.c (live_bytes_read): New function.
6879 (dse_classify_store): Ignore reads of dead bytes.
6881 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
6882 typos and whitespace errors.
6883 * config/i386/predicates.md (address_no_seg_operand): Likewise.
6884 * config/s390/s390.c (s390_emit_prologue): Likewise.
6886 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6889 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
6890 with a symbol for LRA.
6892 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
6894 PR rtl-optimization/82338
6895 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
6897 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
6899 * genmodes.c (calc_wider_mode): Suppress qsort macro.
6900 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
6901 (qsort_chk): Declare.
6902 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
6903 (qsort_chk): New function.
6905 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6907 PR tree-optimization/82337
6908 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
6909 phi definition if the PHI result appears in an abnormal PHI.
6910 (find_basis_for_base_expr): Don't record a basis if the LHS of the
6911 basis appears in an abnormal PHI.
6913 2017-09-29 Richard Biener <rguenther@suse.de>
6915 * graphite-isl-ast-to-gimple.c
6916 (translate_isl_ast_to_gimple::set_codegen_error): New function.
6917 (binary_op_to_tree): Use it.
6918 (get_rename_from_scev): Likewise.
6919 (copy_loop_phi_nodes): Likewise.
6920 (copy_bb_and_scalar_dependences): Likewise.
6921 (translate_pending_phi_nodes): Likewise.
6923 2017-09-29 Jakub Jelinek <jakub@redhat.com>
6926 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
6927 for movabsq $(i32 << shift), r64.
6929 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
6931 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
6932 index when encoding %esp as %rsp to avoid 0x67 prefix.
6934 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
6936 * config/i386/i386.md (*movsf_internal, *movdf_internal):
6937 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
6939 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
6941 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
6942 Extensions with more than 16 double VFP registers.
6943 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
6944 to_clear_mask and all code related to it. Replace the remaining
6945 entry by a sbitmap and adapt code accordingly.
6947 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
6949 * brig-builtins.def: Change pure attributes to const.
6951 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6953 * config.gcc (default_gnu_indirect_function): Default to yes for
6954 sparc*-*-linux* with glibc.
6956 2017-09-28 Joseph Myers <joseph@codesourcery.com>
6958 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
6959 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
6960 when creating .init_array and .fini_array sections with priority
6963 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
6966 * config/aarch64/aarch64.c
6967 (aarch64_builtin_support_vector_misalignment): Always return false
6968 when misalignment is unknown.
6970 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
6972 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
6973 this function to return false if the definition used by the swap
6974 instruction is artificial, or if the memory address from which the
6975 constant value is loaded is not represented by a base address held
6976 in a register or if the base address register is a frame or stack
6977 pointer. Additionally, return false if the base address of the
6978 loaded constant is a SYMBOL_REF but is not considered to be a
6980 (replace_swapped_load_constant): New function.
6981 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
6982 loaded constant vector with a load of a swapped constant vector.
6984 2017-09-27 Carl Love <cel@us.ibm.com>
6986 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
6987 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
6988 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
6991 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
6993 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
6994 first, always call autopref_rank_data otherwise.
6996 2017-09-27 Richard Biener <rguenther@suse.de>
6998 * graphite-scop-detection.c (find_scop_parameters): Move
6999 loop bound handling ...
7000 (gather_bbs::before_dom_children): ... here, avoiding the need
7001 to build scop_info->loop_nest.
7002 (record_loop_in_sese): Remove.
7003 * sese.h (sese_info_t::loop_nest): Remove.
7004 * sese.c (new_sese_info): Do not allocate loop_nest.
7005 (free_sese_info): Do not free loop_nest.
7007 2017-09-27 Jakub Jelinek <jakub@redhat.com>
7010 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
7011 lhs from calls if the lhs has addressable type.
7013 2017-09-27 Richard Biener <rguenther@suse.de>
7015 * graphite.h (scop::max_alias_set): New member.
7016 * graphite-scop-detection.c: Remove references to non-existing
7017 --param in comments.
7018 (build_alias_sets): Record the maximum alias set used for drs.
7019 (build_scops): Support zero as unlimited for
7020 --param graphite-max-arrays-per-scop.
7021 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
7023 (build_poly_sr_1): ... here. Compute alias set based on the
7024 maximum alias set used for drs rather than
7025 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
7027 2017-09-27 Richard Biener <rguenther@suse.de>
7029 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
7030 --param loop-block-tile-size=0 to disable tiling.
7032 2017-09-27 Richard Biener <rguenther@suse.de>
7034 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
7035 (graphite-max-nb-scop-params): Document special value zero.
7036 * domwalk.h (dom_walker::STOP): New symbolical constant.
7037 (dom_walker::dom_walker): Add optional parameter for bb to
7039 (dom_walker::~dom_walker): Declare.
7040 (dom_walker::before_dom_children): Document STOP return value.
7041 (dom_walker::m_user_bb_to_rpo): New member.
7042 (dom_walker::m_bb_to_rpo): Likewise.
7043 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
7044 mapping here if not provided by the user.
7045 (dom_walker::~dom_walker): Free bb to RPO mapping if not
7046 provided by the user.
7047 (dom_walker::STOP): Define.
7048 (dom_walker::walk): Do not compute bb to RPO mapping here.
7049 Support STOP return value from before_dom_children to stop
7051 * graphite-optimize-isl.c (optimize_isl): If the schedule
7052 is the same still generate code if -fgraphite-identity
7053 or -floop-parallelize-all are given.
7054 * graphite-scop-detection.c: Include cfganal.h.
7055 (gather_bbs::gather_bbs): Get and pass through bb to RPO
7057 (gather_bbs::before_dom_children): Return STOP for BBs
7059 (build_scops): Compute bb to RPO mapping and pass it to
7060 the domwalk. Treat --param graphite-max-nb-scop-params=0
7061 as not limiting the number of params.
7062 * graphite.c (graphite_initialize): Remove limit on the
7063 number of basic-blocks in a function.
7064 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
7065 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
7066 default value of 10.
7068 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7070 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
7071 Adjust code to eliminate needing to do the shift right 32-bits
7072 operation after XSCVDPSPN.
7074 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7076 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
7077 ((X / Y) != 0 -> X >= Y): Likewise.
7079 2017-09-26 Carl Love <cel@us.ibm.com>
7081 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
7082 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
7083 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
7084 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
7085 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
7086 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
7087 definitions and overloading.
7088 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
7089 statement for P9V_BUILTIN_XST_LEN_R.
7090 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
7091 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
7092 define_expand and define_insn for the instructions and builtins.
7093 * doc/extend.texi: Update the built-in documentation file for the new
7095 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
7096 define_insn for the instructions
7098 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
7101 * gcc/config/netbsd-protos.h: New file.
7102 * gcc/config/netbsd.c: New file.
7103 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
7104 * gcc/config/t-netbsd: New file.
7105 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
7106 (tmake_file) Add t-netbsd.
7107 (extra_objs) Add netbsd.o.
7109 2017-09-26 Janus Weil <janus@gcc.gnu.org>
7113 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
7115 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
7117 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
7118 sign extension from a vector register to a GPR by doing a 32-bit
7119 direct move and then an EXTSW.
7120 (extendsi<mode>2 splitter): Likewise.
7121 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
7122 right or vector extract after doing XSCVDPSPN. Use
7123 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
7125 (movdi_from_sf_zero_ext): Likewise.
7126 (reload_gpr_from_vsxsf): Likewise.
7127 (p8_mfvsrd_4_disf): Delete, no longer used.
7128 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
7129 then needing to move the SFmode to a GPR to use the XSCVDPSP
7130 instruction instead of FRSP and XSCVDPSPN.
7131 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
7132 it is adjacent to the other XSCVSPDP insns.
7133 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
7134 SFmode to be in traditional Altivec registers.
7135 (vsx_xscvdpspn): Eliminate useless alternative constraint.
7136 (vsx_xscvspdpn): Likewise.
7137 (vsx_xscvspdpn_scalar): Likewise.
7139 2017-09-26 Martin Jambor <mjambor@suse.cz>
7141 * tree-sra.c (compare_access_positions): Put integral types first,
7142 stabilize sorting of integral types, remove conditions putting
7143 non-full-precision integers last.
7144 (sort_and_splice_var_accesses): Disable scalarization if a
7145 non-integert would be represented by a non-full-precision integer.
7147 2017-09-26 Joseph Myers <joseph@codesourcery.com>
7149 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
7150 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
7151 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
7152 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
7153 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
7154 conditionals inside the function instead of around it. Call
7155 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
7156 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
7158 2017-09-26 Richard Biener <rguenther@suse.de>
7160 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
7162 (scop_detection::build_scop_breadth): ... this. Removed.
7163 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
7164 (scop_detection::harmful_stmt_in_bb): Likewise.
7165 (scop_detection::graphite_can_represent_stmt): Likewise.
7166 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
7167 (scop_detection::can_represent_loop): Remove recursion, fold in ...
7168 (scop_detection::can_represent_loop_1): ... this. Removed.
7169 (scop_detection::harmful_loop_in_region): Simplify after inlining
7170 the above and remove more quadraticness.
7171 (build_scops): Adjust.
7172 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
7175 2017-09-26 Jakub Jelinek <jakub@redhat.com>
7178 * config/i386/i386.c (ix86_print_operand_address_as): Only test
7179 REGNO (base) == SP_REG if base is a REG.
7182 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
7183 if it is different SSA_NAME.
7184 (optimize_range_tests_cmp_bitwise): New function.
7185 (optimize_range_tests): Call it.
7187 2017-09-26 Richard Biener <rguenther@suse.de>
7189 PR tree-optimization/82321
7190 * graphite.c (canonicalize_loop_closed_ssa): Properly check
7191 for the def being inside the loop.
7193 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7195 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
7197 * config/s390/s390-builtins.def: Fix constraint on op4.
7199 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7201 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
7202 independent expanders.
7203 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
7204 ("vec_ordered", "vec_unordered"): New expanders.
7206 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7208 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
7211 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7213 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
7214 vec_unpacks_lo_v16qi.
7215 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
7217 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7219 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
7220 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
7221 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
7223 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7225 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
7227 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
7229 ("*vec_slb<mode>"): New insn pattern.
7230 ("vec_shr_<mode>"): New expander.
7231 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
7232 and force the shift count operand to V16QImode.
7233 ("vec_srb<mode>"): Set shift count mode to V16QI.
7235 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7237 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
7238 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
7239 ("vec_widen_smult_hi_<mode>"): New expander definitions.
7241 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
7244 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
7246 2017-09-26 Richard Biener <rguenther@suse.de>
7248 PR tree-optimization/82320
7249 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
7252 2017-09-25 Jeff Law <law@redhat.com>
7254 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
7255 prototype for new argument.
7256 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
7257 mostly extracted from rs6000_emit_allocate_stack.
7258 (rs6000_emit_probe_stack_range_stack_clash): New function.
7259 (rs6000_emit_allocate_stack): Call
7260 rs6000_emit_probe_stack_range_stack_clash as needed.
7261 (rs6000_emit_probe_stack_range): Add additional argument
7262 to call to gen_probe_stack_range{si,di}.
7263 (output_probe_stack_range): New.
7264 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
7265 (output_probe_stack_range_stack_clash): New.
7266 (rs6000_emit_prologue): Emit notes into dump file as requested.
7267 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
7268 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
7269 Add additional operand and pass it to output_probe_stack_range.
7271 2017-09-25 Bin Cheng <bin.cheng@arm.com>
7273 PR tree-optimization/82163
7274 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
7275 (checking_verify_loop_closed_ssa): New parameter.
7276 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
7277 (check_loop_closed_ssa_stmt): Delete.
7278 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
7279 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
7280 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
7283 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
7285 * brig-builtins.def: Treat HSAIL barrier builtins as
7286 setjmp/longjump style functions.
7288 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7290 * target.def (constant_alignment): New hook.
7291 * defaults.h (CONSTANT_ALIGNMENT): Delete.
7292 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
7293 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
7294 * doc/tm.texi: Regenerate.
7295 * targhooks.h (default_constant_alignment): Declare.
7296 (constant_alignment_word_strings): Likewise.
7297 * targhooks.c (default_constant_alignment): New function.
7298 (constant_alignment_word_strings): Likewise.
7299 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
7300 instead of CONSTANT_ALIGNMENT.
7301 * varasm.c (align_variable, get_variable_align, build_constant_desc)
7302 (force_const_mem): Likewise.
7303 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
7304 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
7305 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
7306 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7307 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
7309 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
7310 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7311 constant_alignment_word_strings.
7312 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
7313 (CONSTANT_ALIGNMENT): Likewise.
7314 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7315 (arm_constant_alignment): New function.
7316 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
7317 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7318 constant_alignment_word_strings.
7319 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
7320 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7321 constant_alignment_word_strings.
7322 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
7323 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7324 (cris_constant_alignment): New function.
7325 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
7326 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7327 (epiphany_constant_alignment): New function.
7328 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
7329 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7330 constant_alignment_word_strings.
7331 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
7332 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7333 constant_alignment_word_strings.
7334 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
7335 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7336 constant_alignment_word_strings.
7337 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
7338 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
7339 * config/i386/i386.c (ix86_constant_alignment): Make static.
7340 Use the same interface as the target hook.
7341 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7342 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
7343 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7344 constant_alignment_word_strings.
7345 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
7346 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
7347 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7348 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
7349 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7350 constant_alignment_word_strings.
7351 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
7352 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7353 constant_alignment_word_strings.
7354 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
7355 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7356 constant_alignment_word_strings.
7357 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
7358 * config/microblaze/microblaze.c (microblaze_constant_alignment):
7360 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7361 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
7362 * config/mips/mips.c (mips_constant_alignment): New function.
7363 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7364 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
7365 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
7366 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7367 (mmix_constant_alignment): Make static. Use the same interface
7369 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
7370 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7371 constant_alignment_word_strings.
7372 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
7373 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7374 constant_alignment_word_strings.
7375 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
7376 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7377 constant_alignment_word_strings.
7378 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
7379 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7380 (rs6000_constant_alignment): New function.
7381 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
7382 * config/riscv/riscv.c (riscv_constant_alignment): New function.
7383 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7384 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
7385 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7386 (rs6000_constant_alignment): New function.
7387 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
7388 * config/s390/s390.c (s390_constant_alignment): New function.
7389 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7390 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
7391 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7392 constant_alignment_word_strings.
7393 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
7394 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7395 (sparc_constant_alignment): New function.
7396 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
7397 * config/spu/spu.c (spu_constant_alignment): New function.
7398 (TARGET_CONSTANT_ALIGNMENT): Redefine.
7399 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
7400 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7401 constant_alignment_word_strings.
7402 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
7403 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7404 constant_alignment_word_strings.
7405 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
7406 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
7407 constant_alignment_word_strings.
7408 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
7409 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7410 (visium_constant_alignment): New function.
7411 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
7412 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
7413 (xtensa_constant_alignment): New function.
7414 * system.h (CONSTANT_ALIGNMENT): Poison.
7416 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
7418 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
7419 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
7420 (rs6000_builtin_valid_without_lhs): New helper function.
7421 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7422 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
7424 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
7426 * target.h (vec_perm_indices): Use unsigned short rather than
7428 (auto_vec_perm_indices): Likewise.
7429 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
7430 Use unsigned int rather than unsigned char.
7431 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
7433 2017-09-25 Richard Biener <rguenther@suse.de>
7435 * cfgloop.h (sort_sibling_loops): Declare.
7436 * cfgloop.c (sort_sibling_loops_cmp): New helper.
7437 (sort_sibling_loops): New function sorting the sibling loop list
7439 * graphite.c (graphite_transform_loops): Sort sibling loops.
7441 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
7443 * target.def (vec_perm_const_ok): Change sel parameter to
7445 * optabs-query.c (can_vec_perm_p): Update accordingly.
7446 * doc/tm.texi: Regenerate.
7447 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
7448 auto_vec_perm_indices and remove separate nelt field.
7449 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
7450 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
7451 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
7452 (aarch64_expand_vec_perm_const): Update accordingly.
7453 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
7454 to vec_perm_indices.
7455 * config/arm/arm.c (expand_vec_perm_d): Change perm to
7456 auto_vec_perm_indices and remove separate nelt field.
7457 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
7458 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
7459 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
7461 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
7462 to vec_perm_indices.
7463 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
7464 sel to vec_perm_indices.
7465 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
7466 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
7467 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
7469 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
7472 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
7475 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
7476 on the FUNCTION_DECL function context if it has a DIE that is a
7479 2017-09-25 Richard Biener <rguenther@suse.de>
7481 PR tree-optimization/82285
7482 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
7485 2017-09-25 Tom de Vries <tom@codesourcery.com>
7489 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
7492 2017-09-25 Richard Biener <rguenther@suse.de>
7494 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
7495 ISL errors other than isl_error_quota happen. Dump if the
7496 schedule is the same.
7497 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
7498 errors instead of aborting inside ISL.
7500 2017-09-25 Iain Sandoe <iain@codesourcery.com>
7503 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
7504 of libgcc_eh for m64.
7505 * config/i386/darwin64.h: Likewise.
7507 2017-09-25 Richard Biener <rguenther@suse.de>
7510 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
7511 attribute for incomplete types nor twice for complete ones.
7513 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
7516 * config/i386/i386.c (ix86_print_operand_address_as): Encode
7517 %esp as %rsp to avoid 0x67 prefix if there is no index or base
7520 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
7523 * config/i386/i386.opt (mprefer-avx256): Use
7524 ix86_target_flags variable.
7525 * config/i386/i386.c (ix86_target_string): Move
7526 -mprefer-avx256 to flag2_opts.
7528 2017-09-22 Jakub Jelinek <jakub@redhat.com>
7531 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
7532 and x != -1 | y != -1 into (x & y) != -1.
7534 2017-09-22 Steve Ellcey <sellcey@cavium.com>
7536 * config.gcc: Add new case statement to set
7537 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
7538 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
7539 s390x-*-linux* case statements. Added aarch64 to the list of
7540 supported architectures.
7542 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7544 PR tree-optimization/82289
7545 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
7546 STMT_VINFO_RELEVANT_P.
7548 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7549 Alan Hayward <alan.hayward@arm.com>
7550 David Sherwood <david.sherwood@arm.com>
7552 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
7553 for VR_RANGE only; don't allow VR_ANTI_RANGE.
7554 (extract_range_from_binary_expr_1): Don't call
7555 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
7557 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7558 Alan Hayward <alan.hayward@arm.com>
7559 David Sherwood <david.sherwood@arm.com>
7561 * target.def (preferred_vector_alignment): New hook.
7562 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
7564 * doc/tm.texi: Regenerate.
7565 * targhooks.h (default_preferred_vector_alignment): Declare.
7566 * targhooks.c (default_preferred_vector_alignment): New function.
7567 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
7569 (DR_TARGET_ALIGNMENT): New macro.
7570 (aligned_access_p): Update commentary.
7571 (vect_known_alignment_in_bytes): New function.
7572 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
7574 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
7575 Calculate the misalignment based on the target alignment rather than
7577 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
7578 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
7579 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
7580 the target alignment, rather than masking the element misalignment
7581 with the number of elements in a vector. Also use the target
7582 alignment when calculating the maximum number of peels.
7583 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
7584 instead of TYPE_ALIGN_UNIT.
7585 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
7586 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
7587 (vect_create_addr_base_for_vector_ref): Update call accordingly.
7588 (vect_create_data_ref_ptr): Likewise.
7589 (vect_setup_realignment): Realign by ANDing with
7590 -DR_TARGET_MISALIGNMENT.
7591 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
7592 the number of peels based on DR_TARGET_ALIGNMENT.
7593 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
7594 with the guaranteed alignment boundary when deciding whether
7596 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
7597 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
7598 (ensure_base_align): Remove stmt_info parameter. Get the
7599 target base alignment from DR_TARGET_ALIGNMENT.
7600 (vectorizable_store): Update call accordingly. Interpret
7601 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
7603 (vectorizable_load): Likewise.
7605 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
7606 Alan Hayward <alan.hayward@arm.com>
7607 David Sherwood <david.sherwood@arm.com>
7609 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
7610 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
7611 (vect_enhance_data_refs_alignment): Likewise.
7613 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7615 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7616 error. Only quit immediately if parsing is complete.
7617 (BEGIN): Initialize fatal_err and parse_done.
7618 (begin fpu, end fpu): Check number of arguments.
7619 (begin arch, end arch): Likewise.
7620 (begin cpu, end cpu): Likewise.
7621 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7622 (optalias): Likewise.
7624 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
7626 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7627 * config/arm/arm-isa.h: Delete. Move definitions to ...
7628 * arm-cpus.in: ... here. Use new feature and fgroup values.
7629 * config/arm/arm.c (arm_option_override): Use lower case for feature
7631 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7632 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7633 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7635 (print_isa_bits_for): New function.
7636 (gen_isa): New function.
7637 (gen_comm_data): Use print_isa_bits_for.
7638 (define feature): New keyword.
7639 (define fgroup): New keyword.
7640 * config/arm/t-arm (TM_H): Remove.
7641 (GTM_H): Add arm-isa.h.
7642 (arm-isa.h): Add rule to generate file.
7643 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7644 case for feature bit names.
7646 2017-09-22 Richard Biener <rguenther@suse.de>
7648 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
7650 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
7651 print of no dependency loops ...
7652 * graphite.c (graphite_transform_loops): ... here.
7653 (canonicalize_loop_closed_ssa_form): Work from inner to outer
7655 (same_close_phi_node, remove_duplicate_close_phi,
7656 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
7657 (canonicalize_loop_closed_ssa): ... here and simplify.
7658 * graphite-optimize-isl.c: Include tree-vectorizer.h.
7659 (optimize_isl): Use dump_printf_loc to tell when we stopped
7660 optimizing because of an ISL timeout.
7662 2017-09-22 Richard Biener <rguenther@suse.de>
7664 PR tree-optimization/82291
7665 * tree-if-conv.c (predicate_mem_writes): Make sure to
7666 remove writes in blocks predicated with false.
7668 2017-09-22 Richard Biener <rguenther@suse.de>
7670 * sese.c: Include cfganal.h.
7671 (if_region_set_false_region): Remove.
7672 (create_if_region_on_edge): Likewise.
7673 (move_sese_in_condition): Re-implement without destroying
7676 2017-09-22 Richard Biener <rguenther@suse.de>
7678 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
7679 Verify both BBs contain loop PHI nodes before dispatching to
7681 (graphite_regenerate_ast_isl): Do not recompute dominators,
7682 do not verify three times. Restructure for clarity.
7683 * graphite-scop-detection.c (same_close_phi_node,
7684 remove_duplicate_close_phi, make_close_phi_nodes_unique,
7685 defined_in_loop_p, canonicalize_loop_closed_ssa,
7686 canonicalize_loop_closed_ssa_form): Simplify, remove excess
7687 checking and SSA rewrite, move to ...
7688 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
7689 (graphite_initialize): Do not pass in ctx, do not reset the
7690 SCEV cache, compute only dominators.
7691 (graphite_transform_loops): Allocate ISL ctx after
7692 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
7693 Maintain post-dominators only around build_scops.
7694 * sese.c (if_region_set_false_region): Make static. Free
7695 and recompute dominators.
7696 (move_sese_in_condition): Assert we don't get called with
7697 post-dominators computed.
7698 * sese.h (if_region_set_false_region): Remove.
7700 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
7702 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
7703 mode attribute for TARGET_AVX512VL.
7705 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
7707 * config/i386/i386.opt (mprefer-avx256): New option.
7708 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
7710 (ix86_preferred_simd_mode): Return 256-bit AVX modes
7711 for TARGET_PREFER_AVX256.
7712 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
7714 2017-09-21 Jeff Law <law@redhat.com>
7716 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
7717 Fix dump output if the only stack space is for pushed registers.
7719 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7721 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
7724 2017-09-21 Martin Sebor <msebor@redhat.com>
7727 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
7728 code (in C++) or code that triggers warnings.
7730 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
7732 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
7734 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
7736 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
7737 * sched-rgn.c: Ditto.
7738 * sel-sched-ir.c: Ditto.
7740 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
7742 * toplev.h (set_random_seed): Adjust return type.
7743 * toplev.c (init_local_tick): Move eager initialization of random_seed
7744 to get_random_seed. Adjust comment.
7745 (init_random_seed): Inline to get_random_seed, delete.
7746 (get_random_seed): Initialize random_seed lazily.
7747 (set_random_seed): Do not return previous value.
7748 (print_switch_value): Do not call get_random_seed.
7750 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
7752 * cgraph.c (delete_function_version): New, broken out from...
7753 (cgraph_node::delete_function_version): ...here. Rename to
7754 cgraph_node::delete_function_version_by_decl. Update all uses.
7755 (cgraph_node::remove): Call delete_function_version.
7757 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7760 * tree-inline.c (expand_call_inline): Emit clobber stmts for
7761 VAR_DECLs to which addressable non-volatile parameters are mapped
7762 and for id->retvar after the return value assignment. Clear
7763 id->retval and id->retbnd after inlining.
7765 2017-09-21 Richard Biener <rguenther@suse.de>
7767 PR tree-optimization/82276
7768 PR tree-optimization/82244
7769 * tree-vrp.c (build_assert_expr_for): Set
7770 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
7772 (remove_range_assertions): Revert earlier change.
7774 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
7777 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
7779 2017-09-21 Richard Biener <rguenther@suse.de>
7781 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
7782 Restore valid IL after code generation errors.
7783 * graphite.c (graphite_transform_loops): Diagnose code
7784 generation issues as MSG_MISSED_OPTIMIZATION and continue
7785 with processing SCOPs.
7787 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7788 Alan Hayward <alan.hayward@arm.com>
7789 David Sherwood <david.sherwood@arm.com>
7791 * calls.c (compute_argument_addresses): Use simplify_gen_binary
7792 rather than choosing between plus_constant and gen_rtx_<CODE>.
7793 * expr.c (emit_push_insn): Likewise.
7794 (expand_expr_real_2): Likewise.
7796 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7797 Alan Hayward <alan.hayward@arm.com>
7798 David Sherwood <david.sherwood@arm.com>
7800 * loop-unroll.c (split_iv): Call copy_rtx on the step.
7802 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7803 Alan Hayward <alan.hayward@arm.com>
7804 David Sherwood <david.sherwood@arm.com>
7806 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
7807 calling tree_to_uhwi.
7809 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7810 Alan Hayward <alan.hayward@arm.com>
7811 David Sherwood <david.sherwood@arm.com>
7813 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
7814 INTEGER_CST rather than a negative test for ADDR_EXPR.
7816 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
7817 Alan Hayward <alan.hayward@arm.com>
7818 David Sherwood <david.sherwood@arm.com>
7820 * tree-vrp.c (extract_range_from_binary_expr_1): Check
7821 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
7823 2017-09-21 Richard Biener <rguenther@suse.de>
7825 PR tree-optimization/71351
7826 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
7827 graphite_create_new_loop_guard): Remove, fold remaining parts
7829 (translate_isl_ast_node_for): ... here and simplify.
7831 2017-09-21 Jakub Jelinek <jakub@redhat.com>
7834 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
7835 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
7836 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
7837 alternative always use QI mode, for -Os imov (=R,R) alternative
7838 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
7841 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7842 Jeff Law <law@redhat.com>
7844 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
7845 (allocate_stack_space): New function, partially extracted from
7847 (s390_emit_prologue): Track offset to most recent stack probe.
7848 Code to allocate space moved into allocate_stack_space.
7849 Dump actions when no stack is allocated.
7850 (s390_prologue_plus_offset): New function.
7851 (s390_emit_stack_probe): Likewise.
7853 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
7855 * common.opt (Wa, Wl, Wp, g, gz=): Add
7857 (gno-column-info): Remove.
7858 (gcolumn-info): Drop RejectNegative.
7860 (gno-record-gcc-switches): Remove.
7861 (grecord-gcc-switches): Drop RejectNegative.
7862 (gno-split-dwarf): Remove.
7863 (gsplit-dwarf): Drop RejectNegative.
7864 (gno-strict-dwarf): Remove.
7865 (gstrict-dwarf): Drop RejectNegative.
7866 * config/darwin.opt (gfull, gused): Add RejectNegative.
7867 * dwarf2out.c (gen_producer_string): Drop
7868 gno-record-gcc-switches handler.
7869 * optc-gen.awk: Add g to prefixes with negative forms.
7870 * opts-common.c (remapping_prefix_p): New.
7871 (find_opt): Check it.
7872 (generate_canonical_option): Test g prefix.
7873 (option_map): Add -gno- mapping.
7874 (add_misspelling_candidates): Check remapping_prefix_p.
7876 2017-09-20 Jeff Law <law@redhat.com>
7878 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
7879 thinko in stack clash protection support.
7881 * explow.c (compute_stack_clash_protection_loop_data): Use
7882 CONST_INT_P instead of explicit test. Verify object is a
7883 CONST_INT_P before looking at INTVAL.
7884 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
7885 instead of explicit test.
7887 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7890 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
7891 address instead of to r1 and r11.
7893 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
7895 * config.gcc: Support "knm".
7896 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
7897 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
7899 * config/i386/i386.c (m_KNM): Define.
7900 (processor_target_table): Add "knm".
7902 (ix86_option_override_internal): Add "knm".
7903 (ix86_issue_rate): Add PROCESSOR_KNM.
7904 (ix86_adjust_cost): Ditto.
7905 (ia32_multipass_dfa_lookahead): Ditto.
7906 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
7907 (fold_builtin_cpu): Add M_INTEL_KNM.
7908 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
7909 (processor_type): Add PROCESSOR_KNM.
7910 * config/i386/x86-tune.def: Add m_KNM.
7911 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
7913 2017-09-20 Richard Biener <rguenther@suse.de>
7915 PR tree-optimization/80213
7916 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
7917 are allowed in empty BBs as well.
7918 (canonicalize_loop_closed_ssa): Also look for other complex
7920 (scop_detection::get_sese): Include the loop-closed PHI block
7922 (scop_detection::merge_sese): Remove code adding extra blocks.
7923 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
7924 (build_scops): Assert the final returned scop is invalid.
7926 2017-09-20 Richard Biener <rguenther@suse.de>
7928 PR tree-optimization/82264
7929 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
7931 (vn_phi_lookup): Likewise.
7932 (vn_phi_insert): Likewise.
7934 2017-09-20 Jakub Jelinek <jakub@redhat.com>
7936 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
7937 that fits into uhwi or shwi, add DW_AT_const_value regardless
7938 of early_dwarf without going through RTL, using add_AT_unsigned
7941 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
7942 (DEBUG_LTO_ABBREV_SECTION): Likewise.
7943 (DEBUG_LTO_MACINFO_SECTION): Likewise.
7944 (DEBUG_MACRO_SECTION): Likewise.
7945 (DEBUG_LTO_MACRO_SECTION): Likewise.
7946 (DEBUG_STR_DWO_SECTION): Likewise.
7947 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
7948 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
7949 (DEBUG_LTO_DWO_LINE_SECTION): Define.
7950 (DEBUG_LTO_LINE_STR_SECTION): Define.
7951 (init_sections_and_labels): Initialize debug_line_str_section
7952 variable. Initialize debug_loc_section for -gdwarf-5 to
7953 DEBUG_LOCLISTS_SECTION. Formatting fixes.
7955 2017-09-20 Richard Biener <rguenther@suse.de>
7957 * graphite-sese-to-poly.c (extract_affine): Properly handle
7958 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
7960 2017-09-20 Richard Biener <rguenther@suse.de>
7962 PR tree-optimization/81373
7963 * graphite-scop-detection.c (build_cross_bb_scalars_def):
7964 Force SESE live-out defs to be handled even if they are
7967 2017-09-19 Jeff Law <law@redhat.com>
7969 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
7970 nothing for stack adjustments with REG_STACK_CHECK.
7971 * sched-deps.c (parse_add_or_inc): Reject insns with
7972 REG_STACK_CHECK from dependency breaking.
7973 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
7974 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
7975 * reg-notes.def (STACK_CHECK): New note.
7977 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
7978 (ix86_expand_prologue): Dump stack clash info as needed.
7979 Call ix86_adjust_stack_and_probe_stack_clash as needed.
7981 * function.c (dump_stack_clash_frame_info): New function.
7982 * function.h (dump_stack_clash_frame_info): Prototype.
7983 (enum stack_clash_probes): New enum.
7985 * config/alpha/alpha.c (alpha_expand_prologue): Also check
7986 flag_stack_clash_protection.
7987 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
7988 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
7989 (arm_frame_pointer_required): Likewise.
7990 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
7991 (ia64_expand_prologue): Likewise.
7992 * config/mips/mips.c (mips_expand_prologue): Likewise.
7993 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
7994 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
7995 (sparc_flat_expand_prologue): Likewise.
7996 * config/spu/spu.c (spu_expand_prologue): Likewise.
7998 * explow.c: Include "params.h".
7999 (anti_adjust_stack_and_probe_stack_clash): New function.
8000 (get_stack_check_protect): Likewise.
8001 (compute_stack_clash_protection_loop_data): Likewise.
8002 (emit_stack_clash_protection_loop_start): Likewise.
8003 (emit_stack_clash_protection_loop_end): Likewise.
8004 (allocate_dynamic_stack_space): Use get_stack_check_protect.
8005 Use anti_adjust_stack_and_probe_stack_clash.
8006 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
8007 (emit_stack_clash_protection_loop_start): Likewise.
8008 (emit_stack_clash_protection_loop_end): Likewise.
8009 * rtl.h (get_stack_check_protect): Prototype.
8010 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
8011 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
8012 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
8014 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
8016 * doc/tm.texi: Rebuilt.
8017 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
8018 get_stack_check_protect.
8019 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
8020 * config/arm/arm.c (arm_expand_prologue): Likewise.
8021 (arm_frame_pointer_required): Likewise.
8022 * config/i386/i386.c (ix86_expand_prologue): Likewise.
8023 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
8024 * config/mips/mips.c (mips_expand_prologue): Likewise.
8025 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
8026 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
8027 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
8028 (sparc_flat_expand_prologue): Likewise.
8030 * common.opt (-fstack-clash-protection): New option.
8031 * flag-types.h (enum stack_check_type): Note difference between
8032 -fstack-check= and -fstack-clash-protection.
8033 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
8034 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
8035 * toplev.c (process_options): Issue warnings/errors for cases
8036 not handled with -fstack-clash-protection.
8037 * doc/invoke.texi (-fstack-clash-protection): Document new option.
8038 (-fstack-check): Note additional problem with -fstack-check=generic.
8039 Note that -fstack-check is primarily for Ada and refer users
8040 to -fstack-clash-protection for stack-clash-protection.
8041 Document new params for stack clash protection.
8043 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
8045 * config/i386/i386.c (ix86_split_long_move): Do not handle
8046 address used for LEA in a special way.
8048 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
8050 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
8052 2017-09-19 Martin Sebor <msebor@redhat.com>
8055 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
8056 of incompatible types.
8058 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
8060 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
8061 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
8062 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8063 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
8065 2017-09-19 Richard Biener <rguenther@suse.de>
8067 PR tree-optimization/82244
8068 * tree-vrp.c (remove_range_assertions): Do not propagate
8069 a constant to abnormals but replace the assert with a copy.
8071 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
8073 PR rtl-optimization/57878
8074 PR rtl-optimization/68988
8075 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
8076 avoidance test involving non_reload_pseudos. Move frequency test
8077 below the general fragmentation avoidance test.
8079 2017-09-19 Richard Biener <rguenther@suse.de>
8081 PR tree-optimization/69728
8082 * graphite-sese-to-poly.c (schedule_error): New global.
8083 (add_loop_schedule): Handle empty domain by failing the
8085 (build_original_schedule): Handle schedule_error.
8087 2017-09-19 Richard Biener <rguenther@suse.de>
8089 * graphite-scop-detection.c (scop_detection::can_represent_loop):
8090 Do not iterate to sibling loops but only to siblings of inner
8093 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
8096 * config/m68k/m68k.md (moveq feeding equality comparison): Check
8097 that the registers are different.
8099 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
8101 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
8102 to processor_model and "amdfam17h" to arch_names_table.
8103 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
8105 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8108 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
8110 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8111 Alan Hayward <alan.hayward@arm.com>
8112 David Sherwood <david.sherwood@arm.com>
8114 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
8116 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
8118 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
8119 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
8120 vect_schedule_slp_instance.
8121 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
8122 Update call to vect_slp_analyze_node_operations. Simplify return
8124 (vect_slp_analyze_bb_1): Update call accordingly.
8125 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
8126 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
8127 (vect_schedule_slp): Update call accordingly.
8129 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8130 Alan Hayward <alan.hayward@arm.com>
8131 David Sherwood <david.sherwood@arm.com>
8133 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
8134 with types that aren't in fact scalar.
8136 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8138 * tree-vect-slp.c (vect_record_max_nunits): New function,
8140 (vect_build_slp_tree_1): ...here.
8141 (vect_build_slp_tree_2): Call it for phis too.
8143 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8145 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
8146 to vect_get_vec_def_for_operand when getting the mask operand.
8148 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8149 Alan Hayward <alan.hayward@arm.com>
8150 David Sherwood <david.sherwood@arm.com>
8152 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
8155 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8156 Alan Hayward <alan.hayward@arm.com>
8157 David Sherwood <david.sherwood@arm.com>
8159 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
8160 calculation for vector booleans.
8162 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
8163 Alan Hayward <alan.hayward@arm.com>
8164 David Sherwood <david.sherwood@arm.com>
8166 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
8168 (vect_transform_stmt): ...here.
8169 (vect_analyze_stmt): Use it instead of calling
8170 vectorizable_live_operation directly.
8172 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
8174 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
8175 non-SIMT targets in acc vector loops.
8177 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
8179 * configure.ac: Add arc and check if assembler supports gdwarf2.
8180 * configure: Regenerate.
8182 2017-09-18 Richard Biener <rguenther@suse.de>
8184 PR tree-optimization/82220
8185 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
8186 epilogue niters from the min_profitable_iters compute.
8188 2017-09-18 Jakub Jelinek <jakub@redhat.com>
8191 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
8192 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
8193 (ix86_init_pic_reg): Revert 2017-09-01 changes.
8195 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
8198 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
8199 switching to a new text section.
8201 2017-09-18 Richard Biener <rguenther@suse.de>
8203 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
8205 (build_alias_set): Reject aliases with no access function.
8207 2017-09-18 Richard Biener <rguenther@suse.de>
8209 PR tree-optimization/79622
8210 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
8212 (build_cross_bb_scalars_use): Likewise.
8214 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
8216 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
8218 2017-09-18 Alan Modra <amodra@gmail.com>
8221 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
8222 stack_pointer_rtx for count 0. Update comments. Break up
8223 large rtl expression.
8225 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
8227 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
8228 Increase to 20 bytes.
8229 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
8230 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
8231 or avx version of the stub.
8233 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
8236 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
8237 compute the minimum stack alignment. Also update preferred stack
8238 boundary for leaf functions.
8240 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
8242 PR tree-optimization/82228
8243 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
8246 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8248 * common/config/nds32/nds32-common.c
8249 (nds32_option_optimization_table): Refine formatting.
8250 (nds32_option_optimization_table): Use -fsched-pressure and
8251 -fomit-frame-pointer for specific optimization level.
8253 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
8255 * config/nds32/nds32.c: Refine formatting and comments.
8256 * config/nds32/nds32.h: Likewise.
8257 * config/nds32/nds32.md: Likewise.
8258 * config/nds32/nds32-cost.c: Likewise.
8259 * config/nds32/nds32-isr.c: Likewise.
8260 * config/nds32/nds32-md-auxiliary.c: Likewise.
8261 * config/nds32/nds32-multiple.md: Likewise.
8262 * config/nds32/nds32-predicates.c: Likewise.
8264 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
8265 Jakub Jelinek <jakub@redhat.com>
8267 Add support for -std=c++2a.
8268 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
8270 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
8272 2017-09-15 Steve Ellcey <sellcey@cavium.com>
8275 * doc/extend.texi (Common Function Attributes): Add
8276 references to ARM, AArch64, and S/390 specific attributes.
8277 (Function Specific Option Pragmas): Add AArch64 and S/390
8278 to list of back ends that support the target pragma.
8280 2017-09-15 Nathan Sidwell <nathan@acm.org>
8282 * doc/standards.texi: Fix C++17 description. Update URLs for
8285 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8287 * common.opt (Wcast-align=strict): New warning option.
8288 * doc/invoke.texi: Document -Wcast-align=strict.
8290 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
8292 * cgraph.h (cgraph_thunk_info): Add comments.
8293 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
8294 assert for VIRTUAL_* arguments stricter.
8296 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
8298 PR tree-optimization/71026
8299 * match.pd: Move RDIV patterns from fold-const.c
8300 * fold-const.c (distribute_real_division): Removed.
8301 (fold_binary_loc): Remove calls to distribute_real_divison.
8303 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8305 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
8306 c++1z and gnu++1z as deprecated. Change other references to
8307 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
8308 Change -Wc++1z-compat to -Wc++17-compat.
8309 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
8310 * dwarf2out.c (highest_c_language): Handle C++17.
8311 (gen_compile_unit_die): Likewise.
8313 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8315 PR rtl-optimization/82192
8316 * combine.c (make_extraction): Don't look through non-paradoxical
8317 SUBREGs or TRUNCATE if pos + len is or might be bigger than
8320 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8321 Alan Hayward <alan.hayward@arm.com>
8322 David Sherwood <david.sherwood@arm.com>
8324 * target.def (function_arg_offset): New hook.
8325 * targhooks.h (default_function_arg_offset): Declare.
8326 * targhooks.c (default_function_arg_offset): New function.
8327 * function.c (locate_and_pad_parm): Use
8328 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
8329 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
8330 (TARGET_FUNCTION_ARG_OFFSET): ...this.
8331 * doc/tm.texi: Regenerate.
8332 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
8333 * config/spu/spu.c (spu_function_arg_offset): New function.
8334 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
8335 * system.h (FUNCTION_ARG_OFFSET): Poison.
8337 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8338 Alan Hayard <alan.hayward@arm.com>
8339 David Sherwood <david.sherwood@arm.com>
8341 * target.def (truly_noop_truncation): New hook.
8342 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
8343 than TRULY_NOOP_TRUNCATION.
8344 * hooks.h (hook_bool_uint_uint_true): Declare.
8345 * hooks.c (hook_bool_uint_uint_true): New function.
8346 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
8347 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
8348 * doc/tm.texi: Regenerate.
8349 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
8350 rather than TRULY_NOOP_TRUNCATION in comments.
8351 (simplify_comparison): Likewise.
8352 (record_truncated_value): Likewise.
8353 * expmed.c (extract_bit_field_1): Likewise.
8354 (extract_split_bit_field): Likewise.
8355 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
8356 instead of TRULY_NOOP_TRUNCATION.
8357 * function.c (assign_parm_setup_block): Likewise.
8358 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
8359 * rtlhooks.c: Include target.h.
8360 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
8361 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
8362 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
8363 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
8364 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
8365 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
8366 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
8367 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
8368 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
8369 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
8370 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
8371 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
8372 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
8373 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
8374 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
8375 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
8376 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
8377 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
8378 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
8379 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
8380 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
8381 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
8382 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
8383 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
8384 * config/mips/mips.c (mips_truly_noop_truncation): New function.
8385 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8386 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8387 rather than TRULY_NOOP_TRUNCATION in comments.
8388 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
8389 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
8390 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
8391 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
8392 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
8393 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
8394 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
8395 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
8396 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
8397 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
8398 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
8399 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8400 rather than TRULY_NOOP_TRUNCATION in comments.
8401 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
8402 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
8403 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
8404 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
8405 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
8406 TRULY_NOOP_TRUNCATION condition.
8407 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
8408 (TRULY_NOOP_TRUNCATION): Delete.
8409 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
8410 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
8411 * config/spu/spu.c (spu_truly_noop_truncation): New function.
8412 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8413 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
8414 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
8415 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
8416 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
8417 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
8418 rather than TRULY_NOOP_TRUNCATION in comments.
8419 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
8420 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
8421 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
8422 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
8423 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
8424 * system.h (TRULY_NOOP_TRUNCATION): Poison.
8426 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
8429 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
8430 (*cmp_ior): Likewise.
8431 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
8432 (*ior_scc_scc_cmp): Likewise.
8433 (*and_scc_scc): Likewise.
8434 (*and_scc_scc_cmp): Likewise.
8436 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
8437 Alan Hayard <alan.hayward@arm.com>
8438 David Sherwood <david.sherwood@arm.com>
8440 * target.def (can_change_mode_class): New hook.
8441 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
8442 (hard_regno_nregs): Likewise.
8443 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
8444 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
8445 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
8446 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
8447 (LOAD_EXTEND_OP): Update accordingly.
8448 * doc/tm.texi: Regenerate.
8449 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
8450 CANNOT_CHANGE_MODE_CLASS.
8451 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
8452 (REG_CAN_CHANGE_MODE_P): ...this new macro.
8453 * combine.c (simplify_set): Update accordingly.
8454 * emit-rtl.c (validate_subreg): Likewise.
8455 * recog.c (general_operand): Likewise.
8456 * regcprop.c (mode_change_ok): Likewise.
8457 * reload1.c (choose_reload_regs): Likewise.
8458 (inherit_piecemeal_p): Likewise.
8459 * rtlanal.c (simplify_subreg_regno): Likewise.
8460 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
8461 instead of CANNOT_CHANGE_MODE_CLASS.
8462 (reload_cse_simplify_operands): Likewise.
8463 * reload.c (push_reload): Use targetm.can_change_mode_class
8464 instead of CANNOT_CHANGE_MODE_CLASS.
8465 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
8466 REG_CANNOT_CHANGE_MODE_P.
8467 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8468 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
8469 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8470 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8471 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8472 (arm_can_change_mode_class): New function.
8473 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8474 than CANNOT_CHANGE_MODE_CLASS in comments.
8475 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8476 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
8477 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
8478 (ix86_can_change_mode_class): ...this new function, inverting the
8479 sense of the return value.
8480 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8481 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8482 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8483 (ia64_can_change_mode_class): New function.
8484 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8485 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
8486 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
8487 (m32c_can_change_mode_class): ...this new function, inverting the
8488 sense of the return value.
8489 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8490 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8491 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
8492 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
8493 (mips_can_change_mode_class): ...this new function, inverting the
8494 sense of the return value.
8495 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8496 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8497 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8498 (msp430_can_change_mode_class): New function.
8499 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8500 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
8501 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8502 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8503 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8504 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
8505 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8506 (pa_cannot_change_mode_class): Replace with...
8507 (pa_can_change_mode_class): ...this new function, inverting the
8508 sense of the return value.
8509 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
8510 than CANNOT_CHANGE_MODE_CLASS in comments.
8511 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8512 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
8513 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8514 (pdp11_cannot_change_mode_class): Replace with...
8515 (pdp11_can_change_mode_class): ...this new function, inverting the
8516 sense of the return value.
8517 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8518 * config/powerpcspe/powerpcspe-protos.h
8519 (rs6000_cannot_change_mode_class_ptr): Delete.
8520 * config/powerpcspe/powerpcspe.c
8521 (rs6000_cannot_change_mode_class_ptr): Delete.
8522 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8523 (rs6000_option_override_internal): Assign to
8524 targetm.can_change_mode_class instead of
8525 rs6000_cannot_change_mode_class_ptr.
8526 (rs6000_cannot_change_mode_class): Replace with...
8527 (rs6000_can_change_mode_class): ...this new function, inverting the
8528 sense of the return value.
8529 (rs6000_debug_cannot_change_mode_class): Replace with...
8530 (rs6000_debug_can_change_mode_class): ...this new function.
8531 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8532 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
8533 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8534 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8535 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
8537 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
8538 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8539 (rs6000_option_override_internal): Assign to
8540 targetm.can_change_mode_class instead of
8541 rs6000_cannot_change_mode_class_ptr.
8542 (rs6000_cannot_change_mode_class): Replace with...
8543 (rs6000_can_change_mode_class): ...this new function, inverting the
8544 sense of the return value.
8545 (rs6000_debug_cannot_change_mode_class): Replace with...
8546 (rs6000_debug_can_change_mode_class): ...this new function.
8547 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8548 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
8549 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
8550 (s390_can_change_mode_class): ...this new function, inverting the
8551 sense of the return value.
8552 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8553 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8554 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
8555 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8556 (sh_cannot_change_mode_class): Replace with...
8557 (sh_can_change_mode_class): ...this new function, inverting the
8558 sense of the return value.
8559 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8560 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8561 (sparc_can_change_mode_class): New function.
8562 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8563 * config/spu/spu.c (spu_can_change_mode_class): New function.
8564 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8565 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
8566 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
8567 (visium_can_change_mode_class): New function.
8568 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
8570 2017-09-15 Richard Biener <rguenther@suse.de>
8572 PR tree-optimization/82217
8573 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
8574 but not undefined case.
8576 2017-09-15 Jakub Jelinek <jakub@redhat.com>
8579 * postreload.c (reload_cse_simplify_operands): Skip
8580 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
8582 2017-09-15 Richard Biener <rguenther@suse.de>
8584 PR tree-optimization/68823
8585 * graphite-scop-detection.c (build_alias_set): If we have a
8586 possible dependence check whether we can handle them by just
8587 looking at the DRs DR_ACCESS_FNs.
8588 (build_scops): If build_alias_set fails, fail the SCOP.
8590 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
8592 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
8593 to support float128 built-in functions that require the ISA 3.0
8595 (BU_FLOAT128_3_HW): Likewise.
8596 (SQRTF128): Add support for the IEEE 128-bit square root and fma
8598 (FMAF128): Likewise.
8600 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
8601 support for built-in functions that need the ISA 3.0 IEEE 128-bit
8602 floating point instructions.
8603 (rs6000_invalid_builtin): Likewise.
8604 (rs6000_builtin_mask_names): Likewise.
8605 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
8606 (RS6000_BTM_FLOAT128_HW): Likewise.
8607 (RS6000_BTM_COMMON): Likewise.
8608 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
8610 * doc/extend.texi (RS/6000 built-in functions): Document the
8611 IEEE 128-bit floating point square root and fused multiply-add
8614 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
8616 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
8617 reg (r2) isn't in the set of registers defined in the prologue.
8619 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8620 Alan Hayward <alan.hayward@arm.com>
8621 David Sherwood <david.sherwood@arm.com>
8623 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
8624 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
8625 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
8626 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
8627 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
8629 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
8630 max_vectorization_factor.
8631 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
8633 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8634 Alan Hayward <alan.hayward@arm.com>
8635 David Sherwood <david.sherwood@arm.com>
8637 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
8638 (vect_worthwhile_without_simd_p): Declare.
8639 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
8640 (vectorizable_reduction): Use it.
8641 * tree-vect-stmts.c (vectorizable_shift): Likewise.
8642 (vectorizable_operation): Likewise.
8644 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8645 Alan Hayward <alan.hayward@arm.com>
8646 David Sherwood <david.sherwood@arm.com>
8648 * tree-vectorizer.h (vect_get_num_copies): New function.
8649 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
8650 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8651 (vectorizable_induction): Likewise.
8652 (vectorizable_live_operation): Likewise.
8653 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8654 (vectorizable_bswap): Likewise.
8655 (vectorizable_call): Likewise.
8656 (vectorizable_conversion): Likewise.
8657 (vectorizable_assignment): Likewise.
8658 (vectorizable_shift): Likewise.
8659 (vectorizable_operation): Likewise.
8660 (vectorizable_store): Likewise.
8661 (vectorizable_load): Likewise.
8662 (vectorizable_condition): Likewise.
8663 (vectorizable_comparison): Likewise.
8664 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
8666 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8667 Alan Hayward <alan.hayward@arm.com>
8668 David Sherwood <david.sherwood@arm.com>
8670 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
8671 of vect_init_vector.
8673 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8674 Alan Hayward <alan.hayward@arm.com>
8675 David Sherwood <david.sherwood@arm.com>
8677 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
8678 an inline wrapper that provides a location.
8679 (gimple_build_vector): Likewise.
8680 * gimple-fold.c (gimple_build_vector_from_val): New function.
8681 (gimple_build_vector): Likewise.
8682 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
8683 functions to build the initial value. Always return a gimple value.
8684 (get_initial_defs_for_reduction): Likewise. Only compute
8686 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
8687 vect_init_vector on the results from get_initial_def(s)_for_reduction.
8688 (vectorizable_induction): Use gimple_build_vector rather than
8691 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8692 Alan Hayward <alan.hayward@arm.com>
8693 David Sherwood <david.sherwood@arm.com>
8695 * target.h (vec_perm_indices): New typedef.
8696 (auto_vec_perm_indices): Likewise.
8697 * optabs-query.h: Include target.h
8698 (can_vec_perm_p): Take a vec_perm_indices *.
8699 * optabs-query.c (can_vec_perm_p): Likewise.
8700 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
8701 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8702 * tree-vect-generic.c (lower_vec_perm): Likewise.
8703 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
8704 (vect_grouped_load_supported): Likewise.
8705 (vect_shift_permute_load_chain): Likewise.
8706 (vect_permute_store_chain): Use auto_vec_perm_indices.
8707 (vect_permute_load_chain): Likewise.
8708 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
8709 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
8710 Update uses of can_vec_perm_p.
8711 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
8712 mode with a number of elements. Take a vec_perm_indices *.
8713 (vect_create_epilog_for_reduction): Update accordingly.
8714 Use auto_vec_perm_indices.
8715 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
8716 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
8717 (vect_transform_slp_perm_load): Likewise.
8718 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
8719 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
8720 (vect_gen_perm_mask_checked): Likewise.
8721 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
8722 (vect_gen_perm_mask_checked): Likewise.
8723 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
8724 (vectorizable_store): Likewise.
8725 (vectorizable_load): Likewise.
8726 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
8727 (vectorizable_bswap): Likewise.
8729 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8730 Alan Hayward <alan.hayward@arm.com>
8731 David Sherwood <david.sherwood@arm.com>
8733 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
8734 * tree.c (build_vector): Likewise.
8735 (build_vector_from_ctor): Update accordingly.
8736 (build_vector_from_val): Likewise.
8737 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
8738 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
8739 * tree-vect-generic.c (add_rshift): Likewise.
8740 (expand_vector_divmod): Likewise.
8741 (optimize_vector_constructor): Likewise.
8742 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
8743 (vect_transform_slp_perm_load): Likewise.
8744 (vect_schedule_slp_instance): Likewise.
8745 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
8746 (vectorizable_call): Likewise.
8747 (vect_gen_perm_mask_any): Likewise. Add elements in order.
8748 * expmed.c (make_tree): Likewise.
8749 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
8750 a vector passed to build_vector.
8751 (fold_convert_const): Likewise.
8752 (exact_inverse): Likewise.
8753 (fold_ternary_loc): Likewise.
8754 (fold_relational_const): Likewise.
8755 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
8756 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
8757 (const_unop): Likewise. Store the reduction accumulator in a
8758 variable rather than an array.
8759 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
8760 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
8761 the new vector, rather than constructing it after the input arrays.
8762 (native_interpret_vector): Use auto_vec<tree> when building
8763 a vector passed to build_vector. Add elements in order.
8764 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
8765 auto_vec<tree> when building a vector passed to build_vector.
8766 (vect_create_epilog_for_reduction): Likewise.
8767 (vectorizable_induction): Likewise.
8768 (get_initial_def_for_reduction): Likewise. Fix indentation of
8770 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
8772 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
8773 passed to build_vector.
8775 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
8776 Alan Hayward <alan.hayward@arm.com>
8777 David Sherwood <david.sherwood@arm.com>
8779 * tree-core.h (tree_base::u): Add an "nelts" field.
8780 (tree_vector): Use VECTOR_CST_NELTS as the length.
8781 * tree.c (tree_size): Likewise.
8782 (make_vector): Initialize VECTOR_CST_NELTS.
8783 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
8784 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
8785 TYPE_VECTOR_SUBPARTS.
8786 * expr.c (const_vector_mask_from_tree): Consistently use "units"
8787 as the number of units, setting it from VECTOR_CST_NELTS.
8788 (const_vector_from_tree): Likewise.
8789 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
8790 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
8791 (fold_negate_expr_1): Likewise.
8792 (fold_convert_const): Likewise.
8793 (const_binop): Likewise. Differentiate the number of output and
8795 (const_unop): Likewise.
8796 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
8797 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
8798 in cases that did the opposite.
8800 2017-09-14 Richard Biener <rguenther@suse.de>
8802 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
8805 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
8807 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
8809 2017-09-14 Jakub Jelinek <jakub@redhat.com>
8812 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
8813 if and where to split a bb, except for splitting before debug insn
8814 sequences followed by non-label real insn. Delete debug insns
8815 in between basic blocks.
8817 * combine.c (make_compound_operation_int): Formatting fixes.
8819 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
8820 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8821 * config/netbsd.h (LINK_EH_SPEC): Likewise.
8822 * config/sol2.h (LINK_EH_SPEC): Likewise.
8823 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8824 * config/s390/linux.h (LINK_SPEC): Likewise.
8825 * config/freebsd.h (LINK_EH_SPEC): Likewise.
8826 * config/openbsd.h (LINK_EH_SPEC): Likewise.
8827 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8828 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
8829 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
8830 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
8831 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
8832 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
8834 2017-09-13 Jakub Jelinek <jakub@redhat.com>
8836 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
8838 (ENDFILE_LINUX_SPEC): Likewise.
8839 (LINK_EH_SPEC): Likewise.
8840 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
8841 (LINK_OS_LINUX_SPEC32): Likewise.
8842 (LINK_OS_LINUX_SPEC64): Likewise.
8843 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
8844 (LINK_OS_LINUX_SPEC): Likewise.
8846 2017-09-13 Martin Liska <mliska@suse.cz>
8849 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
8850 CASE_HIGH is NULL_TREE.
8852 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8853 Alan Hayward <alan.hayward@arm.com>
8854 David Sherwood <david.sherwood@arm.com>
8856 * target.def (secondary_memory_needed): New hook.
8857 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
8858 instead of SECONDARY_MEMORY_NEEDED.
8859 (secondary_memory_needed_mode): Likewise.
8860 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
8861 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
8862 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
8863 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
8864 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
8865 * doc/tm.texi: Regenerate.
8866 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
8867 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
8868 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8869 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
8870 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
8871 * config/i386/i386.c (inline_secondary_memory_needed): Put the
8872 mode argument first and change the reg_class arguments to reg_class_t.
8873 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
8874 Make static. Update the call to inline_secondary_memory_needed.
8875 (ix86_register_move_cost): Update the call to
8876 inline_secondary_memory_needed.
8877 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8878 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
8880 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
8881 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8883 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
8884 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
8885 * config/mips/mips.c (mips_secondary_memory_needed): Make static
8886 and match hook interface. Add comment from mips.h.
8887 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8888 * config/mmix/mmix.md (truncdfsf2): Refer to
8889 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8891 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
8892 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
8893 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8894 (pa_secondary_memory_needed): New function.
8895 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
8896 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
8897 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8898 (pdp11_secondary_memory_needed): Make static and match hook interface.
8899 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
8900 * config/powerpcspe/powerpcspe-protos.h
8901 (rs6000_secondary_memory_needed_ptr): Delete.
8902 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
8904 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8905 (rs6000_option_override_internal): Assign to
8906 targetm.secondary_memory_needed rather than
8907 rs6000_secondary_memory_needed_ptr.
8908 (rs6000_secondary_memory_needed): Match hook interface.
8909 (rs6000_debug_secondary_memory_needed): Likewise.
8910 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
8911 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
8912 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
8913 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8914 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
8915 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
8917 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
8918 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8919 (rs6000_option_override_internal): Assign to
8920 targetm.secondary_memory_needed rather than
8921 rs6000_secondary_memory_needed_ptr.
8922 (rs6000_secondary_memory_needed): Match hook interface.
8923 (rs6000_debug_secondary_memory_needed): Likewise.
8924 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
8925 * config/s390/s390.c (s390_secondary_memory_needed): New function.
8926 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8927 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
8928 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
8929 (sparc_secondary_memory_needed): New function.
8930 * lra-constraints.c (check_and_process_move): Refer to
8931 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
8933 (curr_insn_transform): Likewise.
8934 (process_alt_operands): Use targetm.secondary_memory_needed
8935 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8936 (check_secondary_memory_needed_p): Likewise.
8937 (choose_split_class): Likewise.
8938 * reload.c: Unconditionally include code that was previously
8939 conditional on SECONDARY_MEMORY_NEEDED.
8940 (push_secondary_reload): Use targetm.secondary_memory_needed
8941 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8942 (push_reload): Likewise.
8943 * reload1.c: Unconditionally include code that was previously
8944 conditional on SECONDARY_MEMORY_NEEDED.
8945 (choose_reload_regs): Use targetm.secondary_memory_needed
8946 instead of TARGET_SECONDARY_MEMORY_NEEDED.
8947 (gen_reload): Likewise.
8948 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
8950 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
8951 Alan Hayward <alan.hayward@arm.com>
8952 David Sherwood <david.sherwood@arm.com>
8954 * target.def (secondary_memory_needed_mode): New hook:
8955 * targhooks.c (default_secondary_memory_needed_mode): Declare.
8956 * targhooks.h (default_secondary_memory_needed_mode): New function.
8957 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
8958 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
8959 * doc/tm.texi: Regenerate.
8960 * lra-constraints.c (check_and_process_move): Use
8961 targetm.secondary_memory_needed_mode instead of
8962 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
8963 (curr_insn_transform): Likewise.
8964 * reload.c (get_secondary_mem): Likewise.
8965 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8966 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
8968 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8969 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8970 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
8971 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8972 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
8974 * config/powerpcspe/powerpcspe-protos.h
8975 (rs6000_secondary_memory_needed_mode): Delete.
8976 * config/powerpcspe/powerpcspe.c
8977 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8978 (rs6000_secondary_memory_needed_mode): Make static.
8979 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8980 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
8982 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8984 (rs6000_secondary_memory_needed_mode): Make static.
8985 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8986 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
8987 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
8988 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
8989 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
8991 (sparc_secondary_memory_needed_mode): New function.
8992 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
8994 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
8996 * config/aarch64/constraints.md (Umq): New constraint.
8997 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
8999 (mov<mode>): Update condition.
9001 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9003 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
9004 when bitposition is the same.
9006 2017-09-13 Richard Biener <rguenther@suse.de>
9008 * dwarf2out.c (output_die_symbol): Remove.
9009 (output_die): Do not output a DIEs symbol.
9011 2017-09-13 Richard Biener <rguenther@suse.de>
9014 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
9015 default-def to avoid breaking iterator update with the weird
9016 interaction with cgraph_update_edges_for_call_stmt_node.
9018 2017-09-13 Richard Biener <rguenther@suse.de>
9020 * tree-cfg.c (verify_gimple_assign_binary): Add verification
9021 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
9022 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
9023 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
9025 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
9027 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
9028 Disable pc relative literal load irrespective of
9029 TARGET_FIX_ERR_A53_84341 for default.
9031 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
9033 * config/sparc/sparc.c (output_return): Output the source location of
9034 the insn in the delay slot, if any.
9035 (output_sibcall): Likewise.
9037 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
9040 * common.opt (-static-pie): New alias.
9041 (shared): Negate static-pie.
9042 (-no-pie): Update help text.
9044 (static-pie): New option.
9045 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
9046 -static-pie support.
9047 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
9048 (LINK_EH_SPEC): Likewise.
9049 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
9050 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9051 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
9052 * gcc.c (LINK_COMMAND_SPEC): Likewise.
9053 (init_gcc_specs): Likewise.
9054 (init_spec): Likewise.
9055 (display_help): Update help message for -pie.
9056 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
9059 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
9061 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
9062 (movdi_aarch64): Likewise.
9063 (movti_aarch64): Likewise.
9065 2017-09-12 Simon Wright <simon@pushface.org>
9068 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
9069 calculation of the minor version, always output as 0.
9071 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9074 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
9075 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
9076 on it early, rather than manual conversion late. For
9077 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
9078 instead of performing manual conversion.
9080 2017-09-12 Carl Love <cel@us.ibm.com>
9082 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
9083 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
9085 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
9086 VMULOSW): Add definitions.
9087 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9088 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
9089 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
9090 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
9091 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
9093 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9095 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
9097 (movti_aarch64): Likewise.
9098 (movdf_aarch64): Likewise.
9099 (movtf_aarch64): Likewise.
9100 (load_pairdi): Likewise.
9101 (store_pairdi): Likewise.
9102 (load_pairdf): Likewise.
9103 (store_pairdf): Likewise.
9104 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
9105 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
9106 (ldr_got_small_<mode>): Likewise.
9107 (ldr_got_small_28k_<mode>): Likewise.
9108 (ldr_got_tiny): Likewise.
9109 * config/aarch64/iterators.md (ldst_sz): New.
9110 (ldpstp_sz): Likewise.
9111 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
9113 (thunderx_load): Split load_8 to load_16.
9114 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
9116 (thunderx2t99_storepair_basic): Split store_8 to store_16.
9117 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
9118 (xgene1_store_pair): Split store_8 to store_16.
9119 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
9120 (falkor_st_0_st_sd): Split store_8 to store_16.
9122 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
9124 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
9125 and store1/2/3/4 to store_4/8/12/16.
9126 * config/aarch64/aarch64.md: Update for rename.
9127 * config/arm/arm.md: Likewise.: Likewise.
9128 * config/arm/arm.c: Likewise.
9129 * config/arm/thumb1.md: Likewise.
9130 * config/arm/thumb2.md: Likewise.
9131 * config/arm/vfp.md: Likewise.
9132 * config/arm/arm-generic.md: Likewise.
9133 * config/arm/arm1020e.md: Likewise.
9134 * config/arm/arm1026ejs.md: Likewise.
9135 * config/arm/arm1136jfs.md: Likewise.
9136 * config/arm/arm926ejs.md: Likewise.
9137 * config/arm/cortex-a15.md: Likewise.
9138 * config/arm/cortex-a17.md: Likewise.
9139 * config/arm/cortex-a5.md: Likewise.
9140 * config/arm/cortex-a53.md: Likewise.
9141 * config/arm/cortex-a57.md: Likewise.
9142 * config/arm/cortex-a7.md: Likewise.
9143 * config/arm/cortex-a8.md: Likewise.
9144 * config/arm/cortex-a9.md: Likewise.
9145 * config/arm/cortex-m4.md: Likewise.
9146 * config/arm/cortex-m7.md: Likewise.
9147 * config/arm/cortex-r4.md: Likewise.
9148 * config/arm/exynos-m1.md: Likewise.
9149 * config/arm/fa526.md: Likewise.
9150 * config/arm/fa606te.md: Likewise.
9151 * config/arm/fa626te.md: Likewise.
9152 * config/arm/fa726te.md: Likewise.
9153 * config/arm/fmp626.md: Likewise.
9154 * config/arm/iwmmxt.md: Likewise.
9155 * config/arm/ldmstm.md: Likewise.
9156 * config/arm/marvell-pj4.md: Likewise.
9157 * config/arm/xgene1.md: Likewise.
9158 * config/aarch64/thunderx.md: Likewise.
9159 * config/aarch64/thunderx2t99.md: Likewise.
9160 * config/aarch64/falkor.md: Likewise.
9162 2017-09-12 Martin Liska <mliska@suse.cz>
9164 * attribs.c (private_lookup_attribute): New function.
9165 * attribs.h (private_lookup_attribute): Declared here.
9166 (lookup_attribute): Called from this place.
9168 2017-09-12 Richard Biener <rguenther@suse.de>
9170 PR tree-optimization/82157
9171 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
9172 stmts with side-effects.
9174 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9175 Alan Hayward <alan.hayward@arm.com>
9176 David Sherwood <david.sherwood@arm.com>
9178 * target.def (hard_regno_nregs): New hook.
9179 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
9180 * targhooks.h (default_hard_regno_nregs): Declare.
9181 * targhooks.c (default_hard_regno_nregs): New function.
9182 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
9183 (TARGET_HARD_REGNO_NREGS): ...this hook.
9184 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
9185 (CLASS_MAX_NREGS): Likewise.
9186 * doc/tm.texi: Regenerate.
9187 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
9188 instead of HARD_REGNO_NREGS.
9189 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
9190 HARD_REGNO_NREGS in the comment.
9191 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
9192 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
9193 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
9194 Return an unsigned int.
9195 (TARGET_HARD_REGNO_NREGS): Redefine.
9196 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
9197 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
9198 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9199 (arc_hard_regno_nregs): New function.
9200 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
9201 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
9202 (arm_hard_regno_nregs): New function.
9203 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
9204 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
9205 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
9206 (TARGET_HARD_REGNO_NREGS): Redefine.
9207 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
9208 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
9209 (HARD_REGNO_NREGS): Delete.
9210 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
9211 (cr16_hard_regno_nregs): New function.
9212 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
9213 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
9214 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
9215 (cris_hard_regno_nregs): New function.
9216 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
9217 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
9218 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9219 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
9220 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
9221 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
9222 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
9223 (frv_hard_regno_nregs): Make static. Take and return an
9225 (frv_class_max_nregs): Remove outdated copy of documentation.
9226 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
9227 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
9228 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
9229 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
9230 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
9231 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
9232 (TARGET_HARD_REGNO_NREGS): Redefine.
9233 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
9234 (CLASS_MAX_NREGS): Update comment.
9235 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
9236 (ia64_hard_regno_nregs): New function.
9237 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
9238 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
9239 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
9240 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
9241 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
9243 (m32c_hard_regno_nregs): Likewise. Make static.
9244 (TARGET_HARD_REGNO_NREGS): Redefine.
9245 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
9246 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
9247 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
9248 (m68k_hard_regno_nregs): New function.
9249 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
9250 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
9251 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
9252 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
9253 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
9254 Take and return an unsigned int.
9255 (TARGET_HARD_REGNO_NREGS): Redefine.
9256 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
9257 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
9258 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
9259 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
9260 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
9261 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
9262 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
9263 (msp430_hard_regno_nregs): Make static. Take and return an
9265 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
9266 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
9267 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
9268 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
9269 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
9270 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
9271 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
9272 (TARGET_HARD_REGNO_NREGS): Redefine.
9273 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
9274 (PA_HARD_REGNO_NREGS): ...this.
9275 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
9276 (PA_HARD_REGNO_NREGS): ...this.
9277 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9278 (pa_hard_regno_nregs): New function.
9279 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
9280 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
9281 (pdp11_hard_regno_nregs): New function.
9282 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
9283 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
9284 (rs6000_hard_regno_nregs_hook): New function.
9285 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
9286 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
9287 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
9288 Take and return an unsigned int. Move earlier in file.
9289 (TARGET_HARD_REGNO_NREGS): Redefine.
9290 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
9291 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
9292 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
9293 (rl78_hard_regno_nregs): Make static. Take and return an
9295 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
9296 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
9297 (rs6000_hard_regno_nregs_hook): New function.
9298 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
9299 * config/rx/rx.c (rx_hard_regno_nregs): New function.
9300 (TARGET_HARD_REGNO_NREGS): Redefine.
9301 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
9302 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
9303 instead of HARD_REGNO_NREGS.
9304 (s390_hard_regno_nregs): New function.
9305 (s390_hard_regno_mode_ok): Add comment from s390.h.
9306 (TARGET_HARD_REGNO_NREGS): Redefine.
9307 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
9308 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
9309 (sh_hard_regno_nregs): New function.
9310 (sh_pass_in_reg_p): Use it.
9311 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
9312 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
9313 (sparc_hard_regno_nregs): New function.
9314 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
9315 * config/spu/spu.c (spu_hard_regno_nregs): New function.
9316 (spu_function_arg_advance): Use it, supplying a valid register number.
9317 (TARGET_HARD_REGNO_NREGS): Redefine.
9318 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
9319 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
9320 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
9321 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
9322 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
9323 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
9324 (CLASS_MAX_NREGS): Remove copy of old documentation.
9325 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
9326 (visium_hard_regno_nregs): New function.
9327 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
9328 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
9329 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
9330 xtensa_hard_regno_nregs): New function.
9331 * system.h (HARD_REGNO_NREGS): Poison.
9333 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9335 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
9336 hard_regno_nregs instead of HARD_REGNO_NREGS.
9337 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
9338 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
9339 (c6x_expand_epilogue): Likewise.
9340 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
9341 (frv_read_iacc_argument): Likewise.
9342 * config/sh/sh.c: Include regs.h.
9343 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9344 (regs_used): Likewise.
9345 (output_stack_adjust): Likewise.
9346 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
9347 * expmed.c: Include regs.h.
9348 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
9349 * ree.c: Include regs.h.
9350 (combine_reaching_defs): Use hard_regno_nregs instead of
9352 (add_removable_extension): Likewise.
9354 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9356 * regs.h (hard_regno_nregs): Turn into a function.
9357 (end_hard_regno): Update accordingly.
9358 * caller-save.c (setup_save_areas): Likewise.
9359 (save_call_clobbered_regs): Likewise.
9360 (replace_reg_with_saved_mem): Likewise.
9361 (insert_restore): Likewise.
9362 (insert_save): Likewise.
9363 * combine.c (can_change_dest_mode): Likewise.
9364 (move_deaths): Likewise.
9365 (distribute_notes): Likewise.
9366 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
9367 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
9368 (rs6000_split_multireg_move): Likewise.
9369 (rs6000_register_move_cost): Likewise.
9370 (rs6000_memory_move_cost): Likewise.
9371 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
9372 (rs6000_split_multireg_move): Likewise.
9373 (rs6000_register_move_cost): Likewise.
9374 (rs6000_memory_move_cost): Likewise.
9375 * cselib.c (cselib_reset_table): Likewise.
9376 (cselib_lookup_1): Likewise.
9377 * emit-rtl.c (set_mode_and_regno): Likewise.
9378 * function.c (aggregate_value_p): Likewise.
9379 * ira-color.c (setup_profitable_hard_regs): Likewise.
9380 (check_hard_reg_p): Likewise.
9381 (calculate_saved_nregs): Likewise.
9382 (assign_hard_reg): Likewise.
9383 (improve_allocation): Likewise.
9384 (calculate_spill_cost): Likewise.
9385 * ira-emit.c (modify_move_list): Likewise.
9386 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
9387 (ira_hard_reg_in_set_p): Likewise.
9388 * ira.c (setup_reg_mode_hard_regset): Likewise.
9389 (clarify_prohibited_class_mode_regs): Likewise.
9390 (check_allocation): Likewise.
9391 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9392 (lra_setup_reg_renumber): Likewise.
9393 (setup_try_hard_regno_pseudos): Likewise.
9394 (spill_for): Likewise.
9395 (assign_hard_regno): Likewise.
9396 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
9397 * lra-constraints.c (in_class_p): Likewise.
9398 (lra_constraint_offset): Likewise.
9399 (simplify_operand_subreg): Likewise.
9400 (lra_constraints): Likewise.
9401 (split_reg): Likewise.
9402 (split_if_necessary): Likewise.
9403 (invariant_p): Likewise.
9404 (inherit_in_ebb): Likewise.
9405 * lra-lives.c (process_bb_lives): Likewise.
9406 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
9407 (get_hard_regs): Likewise.
9408 (do_remat): Likewise.
9409 * lra-spills.c (assign_spill_hard_regs): Likewise.
9410 * mode-switching.c (create_pre_exit): Likewise.
9411 * postreload.c (reload_combine_recognize_pattern): Likewise.
9412 * recog.c (peep2_find_free_register): Likewise.
9413 * regcprop.c (kill_value_regno): Likewise.
9414 (set_value_regno): Likewise.
9415 (copy_value): Likewise.
9416 (maybe_mode_change): Likewise.
9417 (find_oldest_value_reg): Likewise.
9418 (copyprop_hardreg_forward_1): Likewise.
9419 * regrename.c (check_new_reg_p): Likewise.
9420 (regrename_do_replace): Likewise.
9421 * reload.c (push_reload): Likewise.
9422 (combine_reloads): Likewise.
9423 (find_dummy_reload): Likewise.
9424 (operands_match_p): Likewise.
9425 (find_reloads): Likewise.
9426 (find_equiv_reg): Likewise.
9427 (reload_adjust_reg_for_mode): Likewise.
9428 * reload1.c (count_pseudo): Likewise.
9429 (count_spilled_pseudo): Likewise.
9430 (find_reg): Likewise.
9431 (clear_reload_reg_in_use): Likewise.
9432 (free_for_value_p): Likewise.
9433 (allocate_reload_reg): Likewise.
9434 (choose_reload_regs): Likewise.
9435 (reload_adjust_reg_for_temp): Likewise.
9436 (emit_reload_insns): Likewise.
9437 (delete_output_reload): Likewise.
9438 * rtlanal.c (subreg_get_info): Likewise.
9439 * sched-deps.c (sched_analyze_reg): Likewise.
9440 * sel-sched.c (init_regs_for_mode): Likewise.
9441 (mark_unavailable_hard_regs): Likewise.
9442 (choose_best_reg_1): Likewise.
9443 (verify_target_availability): Likewise.
9444 * valtrack.c (dead_debug_insert_temp): Likewise.
9445 * var-tracking.c (track_loc_p): Likewise.
9446 (emit_note_insn_var_location): Likewise.
9447 * varasm.c (make_decl_rtl): Likewise.
9448 * reginfo.c (choose_hard_reg_mode): Likewise.
9449 (init_reg_modes_target): Refer directly to
9450 this_target_regs->x_hard_regno_nregs.
9452 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9454 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
9455 instead of hard_regno_nregs.
9457 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9459 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
9460 end_hard_regno instead of hard_regno_nregs.
9461 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
9462 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
9463 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
9464 * ira-color.c (improve_allocation): Likewise.
9465 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9466 * lra-lives.c (mark_regno_live): Likewise.
9467 (mark_regno_dead): Likewise.
9468 * lra-remat.c (operand_to_remat): Likewise.
9469 * lra.c (collect_non_operand_hard_regs): Likewise.
9470 * postreload.c (reload_combine_note_store): Likewise.
9471 (move2add_valid_value_p): Likewise.
9472 * reload.c (regno_clobbered_p): Likewise.
9474 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9476 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
9478 * config/v850/v850.c (v850_reorg): Likewise.
9479 * reload.c (refers_to_regno_for_reload_p): Likewise.
9480 (find_equiv_reg): Likewise.
9481 * reload1.c (reload_reg_reaches_end_p): Likewise.
9483 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9485 * caller-save.c (add_used_regs): Use REG_NREGS instead of
9487 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
9488 * config/arm/arm.c (output_move_neon): Likewise.
9489 (arm_attr_length_move_neon): Likewise.
9490 (neon_split_vcombine): Likewise.
9491 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
9492 (c6x_mark_reg_written): Likewise.
9493 (c6x_dwarf_register_span): Likewise.
9494 * config/i386/i386.c (ix86_save_reg): Likewise.
9495 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
9496 (rws_access_reg): Likewise.
9497 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
9498 * mode-switching.c (create_pre_exit): Likewise.
9499 * ree.c (combine_reaching_defs): Likewise.
9500 (add_removable_extension): Likewise.
9501 * regcprop.c (find_oldest_value_reg): Likewise.
9502 (copyprop_hardreg_forward_1): Likewise.
9503 * reload.c (reload_inner_reg_of_subreg): Likewise.
9504 (push_reload): Likewise.
9505 (combine_reloads): Likewise.
9506 (find_dummy_reload): Likewise.
9507 (reload_adjust_reg_for_mode): Likewise.
9508 * reload1.c (find_reload_regs): Likewise.
9509 (forget_old_reloads_1): Likewise.
9510 (reload_reg_free_for_value_p): Likewise.
9511 (reload_adjust_reg_for_temp): Likewise.
9512 (emit_reload_insns): Likewise.
9513 (delete_output_reload): Likewise.
9514 * sel-sched.c (choose_best_reg_1): Likewise.
9515 (choose_best_pseudo_reg): Likewise.
9517 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9518 Alan Hayward <alan.hayward@arm.com>
9519 David Sherwood <david.sherwood@arm.com>
9521 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
9522 * target.def (slow_unaligned_access): New hook.
9523 * targhooks.h (default_slow_unaligned_access): Declare.
9524 * targhooks.c (default_slow_unaligned_access): New function.
9525 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
9526 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
9527 * doc/tm.texi: Regenerate.
9528 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
9529 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
9530 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
9532 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
9533 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
9535 (rs6000_slow_unaligned_access): New function.
9536 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9537 (expand_block_compare): Likewise.
9538 (expand_strn_compare): Likewise.
9539 (rs6000_rtx_costs): Likewise.
9540 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
9541 (riscv_slow_unaligned_access): Likewise.
9542 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
9543 (riscv_slow_unaligned_access_p): ...this and make static.
9544 (riscv_option_override): Update accordingly.
9545 (riscv_slow_unaligned_access): New function.
9546 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9547 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
9548 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
9549 (rs6000_slow_unaligned_access): New function.
9550 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
9551 (rs6000_rtx_costs): Likewise.
9552 * config/rs6000/rs6000-string.c (expand_block_compare)
9553 (expand_strn_compare): Use targetm.slow_unaligned_access instead
9554 of SLOW_UNALIGNED_ACCESS.
9555 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
9556 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
9557 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
9558 of SLOW_UNALIGNED_ACCESS.
9559 * expmed.c (simple_mem_bitfield_p): Likewise.
9560 * expr.c (alignment_for_piecewise_move): Likewise.
9561 (emit_group_load_1): Likewise.
9562 (emit_group_store): Likewise.
9563 (copy_blkmode_from_reg): Likewise.
9564 (emit_push_insn): Likewise.
9565 (expand_assignment): Likewise.
9566 (store_field): Likewise.
9567 (expand_expr_real_1): Likewise.
9568 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
9569 * lra-constraints.c (simplify_operand_subreg): Likewise.
9570 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
9571 * gimple-ssa-store-merging.c: Likewise in block comment at start
9573 * tree-ssa-strlen.c: Include target.h.
9574 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
9575 of SLOW_UNALIGNED_ACCESS.
9576 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
9578 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
9580 PR rtl-optimization/82185
9581 * expmed.c (emit_store_flag_int): Only test tem if it has been
9584 2017-09-12 Richard Biener <rguenther@suse.de>
9587 * match.pd ((FTYPE) N CMP CST): Fix typo.
9589 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
9591 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
9593 (mips_near_type_p): Add 'short_call' attribute as a synonym
9595 * doc/extend.texi (short_call): Document new function attribute.
9597 2017-09-12 Jakub Jelinek <jakub@redhat.com>
9600 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
9601 assertion check that in the condition.
9602 (get_atomic_generic_size): Likewise. Before testing if parameter
9603 has pointer type, if it has array type, call for C++
9604 default_conversion to perform array-to-pointer conversion.
9606 2017-09-12 Richard Biener <rguenther@suse.de>
9608 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
9609 for operations we cannot scalarize.
9611 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
9613 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
9614 vectors heap vectors. Clean up comments.
9615 Make visited_bbs a reference.
9616 (profitable_jump_thread_path): Make GC
9617 vectors heap vectors. Clean up comments.
9619 (convert_and_register_jump_thread_path): Make GC vectors heap
9621 (check_subpath_and_update_thread_path): Same. Clean up comments.
9622 Make visited_bbs a reference.
9623 (handle_phi): Abstract common code to to
9624 register_jump_thread_path_if_profitable.
9625 Rename VAR_BB to DEF_BB.
9627 Make GC vectors heap vectors.
9628 Make visited_bbs a reference.
9629 (handle_assignment): Same.
9630 (register_jump_thread_path_if_profitable): New.
9631 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
9633 Make GC vectors heap vectors. Clean up comments.
9634 Make visited_bbs a reference.
9635 (find_jump_threads_backwards): Make visited_bbs live in the stack.
9636 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
9639 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
9642 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
9643 words of E_DImode object are reachable by xtensa_uimm8x4 access.
9645 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
9647 Revert r251800 and r251799.
9649 2017-09-11 Martin Jambor <mjambor@suse.cz>
9652 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
9653 arguments in advance.
9654 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
9655 use it to find predecessor edges.
9656 (naive_outof_ssa): Collect vector of predecessors.
9658 2017-09-08 Jason Merrill <jason@redhat.com>
9660 PR c++/70029 - ICE with ref-qualifier and -flto
9661 * langhooks.h (struct lang_hooks_for_types): Add
9662 copy_lang_qualifiers.
9663 * attribs.c (build_type_attribute_qual_variant): Use it.
9664 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
9666 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
9667 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
9669 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
9672 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
9673 (*mulsi3_sp64): New instruction.
9674 (mulsi3): New expander.
9676 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
9678 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
9680 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
9682 * sancov.c: Include memmodel.h.
9684 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
9687 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
9690 2017-09-07 Carl Love <cel@us.ibm.com>
9692 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
9693 the sldi instruction.
9695 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
9697 * sancov.c: Include tm_p.h.
9699 2017-09-07 Jakub Jelinek <jakub@redhat.com>
9702 * output.h (switch_to_other_text_partition): New declaration.
9703 * varasm.c (switch_to_other_text_partition): New function.
9704 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
9705 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
9706 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
9707 to the other text partition before emitting LCL label and switch back
9708 after emitting the word after it.
9710 2017-09-07 Richard Biener <rguenther@suse.de>
9712 * passes.def (pass_split_crit_edges): Remove instance before PRE.
9713 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
9714 critical edges here, after loop init.
9715 (pass_data_pre): Remove PROP_no_crit_edges flags.
9716 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
9717 for valueization of call args to avoid leaking VN_TOP.
9718 (visit_use): Assert we do not visit default defs.
9719 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
9720 Use error_mark_node to more easily detect leaking VN_TOP.
9721 All default-defs are varying, not VN_TOP. Mark them visited.
9722 (run_scc_vn): Make code match comment.
9724 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9726 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
9727 OPTION_MASK_FLOAT128_KEYWORD.
9728 (POWERPC_MASKS): Likewise.
9729 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
9730 support for the -mfloat128-type option, and make -mfloat128
9731 default on PowerPC Linux systems. Define or undefine
9732 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
9733 Define __float128 to be __ieee128 if IEEE 128-bit support is
9734 enabled, or undefine it.
9735 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
9736 Delete defining __FLOAT128_TYPE__.
9737 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
9738 -mfloat128-type option and make -mfloat128 default on PowerPC
9740 (TARGET_FLOAT128_TYPE): Likewise.
9741 (-mfloat128-type): Likewise.
9742 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9743 Delete the -mfloat128-type option and make -mfloat128 default on
9744 PowerPC Linux systems. Always use __ieee128 to be the keyword for
9745 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
9746 128-bit floating point is enabled. Change tests from using
9747 -mfloat128-type to -mfloat128.
9748 (rs6000_mangle_type): Use the correct mangling for the __float128
9749 type even if normal long double is restricted to 64-bits.
9750 (floatn_mode): Enable the _Float128 type by default on VSX Linux
9752 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
9753 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
9754 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
9756 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
9757 documentation for -mfloat128.
9759 2017-09-06 Olivier Hainque <hainque@adacore.com>
9761 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
9763 2017-09-06 Wish Wu <wishwu007@gmail.com>
9764 Jakub Jelinek <jakub@redhat.com>
9766 * asan.c (initialize_sanitizer_builtins): Add
9767 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
9768 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
9769 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
9770 BT_FN_VOID_UINT64_PTR variables.
9771 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
9772 (BT_FN_VOID_UINT16_UINT16): Likewise.
9773 (BT_FN_VOID_UINT32_UINT32): Likewise.
9774 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
9775 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
9776 (BT_FN_VOID_UINT64_PTR): Likewise.
9777 * common.opt (flag_sanitize_coverage): New variable.
9778 (fsanitize-coverage=trace-pc): Remove.
9779 (fsanitize-coverage=): Add.
9780 * flag-types.h (enum sanitize_coverage_code): New enum.
9781 * fold-const.c (fold_range_test): Disable non-short-circuit
9782 optimization if flag_sanitize_coverage.
9783 (fold_truth_andor): Likewise.
9784 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
9785 * opts.c (COVERAGE_SANITIZER_OPT): Define.
9786 (coverage_sanitizer_opts): New array.
9787 (get_closest_sanitizer_option): Add OPTS argument, handle also
9788 OPT_fsanitize_coverage_.
9789 (parse_sanitizer_options): Adjusted to also handle
9790 OPT_fsanitize_coverage_.
9791 (common_handle_option): Add OPT_fsanitize_coverage_.
9792 * sancov.c (instrument_comparison, instrument_switch): New function.
9793 (sancov_pass): Add trace-cmp support.
9794 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
9795 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
9796 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
9797 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
9798 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
9799 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
9800 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
9801 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
9802 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
9803 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
9805 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9807 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
9808 error. Only quit immediately if parsing is complete.
9809 (BEGIN): Initialize fatal_err and parse_done.
9810 (begin fpu, end fpu): Check number of arguments.
9811 (begin arch, end arch): Likewise.
9812 (begin cpu, end cpu): Likewise.
9813 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
9814 (optalias): Likewise.
9816 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
9818 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
9819 * config/arm/arm-isa.h: Delete. Move definitions to ...
9820 * arm-cpus.in: ... here. Use new feature and fgroup values.
9821 * config/arm/arm.c (arm_option_override): Use lower case for feature
9823 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
9824 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
9825 * config/arm/parsecpu.awk (END): Add new command 'isa'.
9827 (print_isa_bits_for): New function.
9828 (gen_isa): New function.
9829 (gen_comm_data): Use print_isa_bits_for.
9830 (define feature): New keyword.
9831 (define fgroup): New keyword.
9832 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
9833 (arm-isa.h): Add rule to generate file.
9834 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
9835 case for feature bit names.
9837 2017-09-06 Richard Biener <rguenther@suse.de>
9839 * tree-ssa-pre.c (NECESSARY): Remove.
9840 (create_expression_by_pieces): Do not touch pass-local flags.
9841 (insert_into_preds_of_block): Likewise.
9842 (do_pre_regular_insertion): Likewise.
9843 (eliminate_insert): Likewise.
9844 (eliminate_dom_walker::before_dom_children): Likewise.
9845 (fini_eliminate): Do not look at inserted_exprs.
9846 (mark_operand_necessary): Remove.
9847 (remove_dead_inserted_code): Replace with simple work-list
9848 algorithm based on inserted_exprs and SSA uses.
9849 (pass_pre::execute): Re-order fini_eliminate and
9850 remove_dead_inserted_code.
9852 2017-09-06 Olivier Hainque <hainque@adacore.com>
9854 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
9855 for VxWorks 7. Adjust surrounding comments.
9857 2017-09-06 Richard Biener <rguenther@suse.de>
9859 * gimple-ssa-strength-reduction.c
9860 (find_candidates_dom_walker::before_dom_children): Also allow
9863 2017-09-06 Richard Biener <rguenther@suse.de>
9865 PR tree-optimization/82108
9866 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
9867 for gap in the non-permutation SLP case.
9869 2017-09-06 Martin Jambor <mjambor@suse.cz>
9871 PR tree-optimization/82078
9872 * tree-sra.c (sort_and_splice_var_accesses): Move call to
9873 add_access_to_work_queue...
9874 (build_accesses_from_assign): ...here.
9875 (propagate_all_subaccesses): Make sure racc is the group
9876 representative, if there is one.
9878 2017-09-06 Jakub Jelinek <jakub@redhat.com>
9881 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
9884 2017-09-06 Richard Biener <rguenther@suse.de>
9886 * gimple-ssa-strength-reduction.c
9887 (find_candidates_dom_walker::before_doom_children): Use a
9888 type and not a mode check.
9890 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
9893 * config/arm/predicates.md (arm_general_adddi_operand): Create new
9895 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
9897 2017-09-05 Jeff Law <law@redhat.com>
9899 PR tree-optimization/64910
9900 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
9901 cases where we have 3 or more operands.
9903 2017-09-05 Jakub Jelinek <jakub@redhat.com>
9906 * omp-low.c (lower_omp_for): Recompute tree invariant if
9907 gimple_omp_for_initial/final is ADDR_EXPR.
9910 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
9911 into gimple val before gimplification fo the COND_EXPR.
9913 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
9915 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
9916 REGION_COPY argument.
9917 (thread_through_all_blocks): Remove unused argument to
9918 duplicate_thread_path.
9920 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9921 Alan Hayward <alan.hayward@arm.com>
9922 David Sherwood <david.sherwood@arm.com>
9924 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
9925 Take a scalar_mode rather than a machine_mode.
9926 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9927 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
9928 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
9929 (aarch64_gen_adjusted_ldpstp): Likewise.
9930 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
9932 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9933 Alan Hayward <alan.hayward@arm.com>
9934 David Sherwood <david.sherwood@arm.com>
9936 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
9937 Take a scalar_int_mode instead of a machine_mode.
9938 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9939 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9940 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9941 (aarch64_simd_attr_length_rglist): Delete.
9942 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
9943 a scalar_int_mode instead of a machine_mode.
9944 (aarch64_add_offset): Likewise.
9945 (aarch64_internal_mov_immediate): Likewise
9946 (aarch64_add_constant_internal): Likewise.
9947 (aarch64_add_constant): Likewise.
9948 (aarch64_movw_imm): Likewise.
9949 (aarch64_rtx_arith_op_extract_p): Likewise.
9950 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
9951 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
9952 Remove assert that the mode isn't a vector.
9953 (aarch64_output_scalar_simd_mov_immediate): Likewise.
9954 (aarch64_expand_mov_immediate): Update calls after above changes.
9955 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
9956 (aarch64_and_bitmask_imm): Check for scalar integer modes.
9957 (aarch64_move_imm): Likewise.
9958 (aarch64_can_const_movi_rtx_p): Likewise.
9959 (aarch64_strip_extend): Likewise.
9960 (aarch64_extr_rtx_p): Likewise.
9961 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
9962 a CONST_INT when the mode parameter is VOIDmode.
9963 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
9965 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9967 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
9968 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9969 (bitwise_type_for_mode): Update accordingly.
9971 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9973 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
9974 * stor-layout.c (mode_for_size_tree): Likewise.
9975 (mode_for_array): Update accordingly.
9976 (layout_decl): Likewise.
9977 (compute_record_mode): Likewise. Only set the mode once.
9979 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9981 * target.def (get_mask_mode): Change return type to opt_mode.
9983 * doc/tm.texi: Regenerate.
9984 * targhooks.h (default_get_mask_mode): Return an opt_mode.
9985 * targhooks.c (default_get_mask_mode): Likewise.
9986 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
9987 * optabs-query.c (can_vec_mask_load_store_p): Update use of
9988 targetm.get_mask_mode.
9989 * tree.c (build_truth_vector_type): Likewise.
9991 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
9993 * machmode.h (mode_for_vector): Return an opt_mode.
9994 * stor-layout.c (mode_for_vector): Likewise.
9995 (mode_for_int_vector): Update accordingly.
9996 (layout_type): Likewise.
9997 * config/i386/i386.c (emit_memmov): Likewise.
9998 (ix86_expand_set_or_movmem): Likewise.
9999 (ix86_expand_vector_init): Likewise.
10000 (ix86_get_mask_mode): Likewise.
10001 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
10003 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
10004 * expmed.c (extract_bit_field_1): Likewise.
10005 * expr.c (expand_expr_real_2): Likewise.
10006 * optabs-query.c (can_vec_perm_p): Likewise.
10007 (can_vec_mask_load_store_p): Likewise.
10008 * optabs.c (expand_vec_perm): Likewise.
10009 * targhooks.c (default_get_mask_mode): Likewise.
10010 * tree-vect-stmts.c (vectorizable_store): Likewise.
10011 (vectorizable_load): Likewise.
10012 (get_vectype_for_scalar_type_and_size): Likewise.
10014 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10016 * machmode.h (mode_for_int_vector): New function.
10017 * stor-layout.c (mode_for_int_vector): Likewise.
10018 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
10019 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
10020 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10021 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10022 (s390_expand_vcond): Likewise.
10024 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10026 * machmode.h (opt_machine_mode): New type.
10027 (opt_mode<T>): Allow construction from anything that can be
10029 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
10030 (mode_for_size): Return an opt_machine_mode.
10031 * stor-layout.c (mode_for_size): Likewise.
10032 (mode_for_size_tree): Update call accordingly.
10033 (bitwise_mode_for_mode): Likewise.
10034 (make_fract_type): Likewise.
10035 (make_accum_type): Likewise.
10036 * caller-save.c (replace_reg_with_saved_mem): Update call
10038 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10039 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10040 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10041 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
10042 * expmed.c (extract_bit_field_1): Likewise.
10043 * reload.c (get_secondary_mem): Likewise.
10044 * varasm.c (assemble_integer): Likewise.
10045 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
10048 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10050 * machmode.h (decimal_float_mode_for_size): New function.
10051 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
10052 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
10053 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
10054 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
10055 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
10056 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
10058 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10060 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
10061 (get_builtin_sync_mode): Likewise.
10062 (expand_ifn_atomic_compare_exchange): Likewise.
10063 (expand_builtin_atomic_clear): Likewise.
10064 (expand_builtin_atomic_test_and_set): Likewise.
10065 (fold_builtin_atomic_always_lock_free): Likewise.
10066 * calls.c (compute_argument_addresses): Likewise.
10067 (emit_library_call_value_1): Likewise.
10068 (store_one_arg): Likewise.
10069 * combine.c (combine_instructions): Likewise.
10070 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
10071 * config/arm/arm.c (arm_function_value): Likewise.
10072 (aapcs_allocate_return_reg): Likewise.
10073 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
10074 * config/i386/i386.c (construct_container): Likewise.
10075 (ix86_gimplify_va_arg): Likewise.
10076 (ix86_expand_sse_cmp): Likewise.
10077 (emit_memmov): Likewise.
10078 (emit_memset): Likewise.
10079 (expand_small_movmem_or_setmem): Likewise.
10080 (ix86_expand_pextr): Likewise.
10081 (ix86_expand_pinsr): Likewise.
10082 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
10083 * config/microblaze/microblaze.c (microblaze_block_move_straight):
10085 * config/mips/mips.c (mips_function_value_1) Likewise.
10086 (mips_block_move_straight): Likewise.
10087 (mips_expand_ins_as_unaligned_store): Likewise.
10088 * config/powerpcspe/powerpcspe.c
10089 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10090 (rs6000_darwin64_record_arg_flush): Likewise.
10091 * config/rs6000/rs6000.c
10092 (rs6000_darwin64_record_arg_advance_flush): Likewise.
10093 (rs6000_darwin64_record_arg_flush): Likewise.
10094 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
10095 (sparc_function_value_1): Likewise.
10096 * config/spu/spu.c (adjust_operand): Likewise.
10097 (spu_emit_branch_or_set): Likewise.
10098 (arith_immediate_p): Likewise.
10099 * emit-rtl.c (gen_lowpart_common): Likewise.
10100 * expr.c (expand_expr_real_1): Likewise.
10101 * function.c (assign_parm_setup_block): Likewise.
10102 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
10103 * reload1.c (alter_reg): Likewise.
10104 * stor-layout.c (mode_for_vector): Likewise.
10105 (layout_type): Likewise.
10107 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
10109 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
10110 (spu_convert_move): Likewise.
10111 * lower-subreg.c (resolve_simple_move): Likewise.
10113 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10116 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
10117 define_insn to a define_expand.
10118 (altivec_vsum2sws_direct): New define_insn.
10119 (altivec_vsumsws): Convert from a define_insn to a define_expand.
10121 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10123 * config/arm/arm.c (arm_option_params_internal): Improve setting of
10126 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
10131 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
10132 realign stack if stack alignment needed is less than incoming
10135 2017-09-05 Marek Polacek <polacek@redhat.com>
10138 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
10141 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
10143 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
10145 2017-09-05 Richard Biener <rguenther@suse.de>
10147 PR tree-optimization/82084
10148 * fold-const.c (can_native_encode_string_p): Handle wide characters.
10150 2017-09-05 Richard Biener <rguenther@suse.de>
10152 PR tree-optimization/82102
10153 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
10155 2017-09-05 Martin Liska <mliska@suse.cz>
10157 PR tree-optimization/82032
10158 * tree-cfg.c (generate_range_test): New function.
10159 * tree-cfg.h (generate_range_test): Declared here.
10160 * tree-cfgcleanup.c (convert_single_case_switch): New function.
10161 (cleanup_control_expr_graph): Use it.
10162 * tree-switch-conversion.c (try_switch_expansion): Remove
10164 (emit_case_nodes): Use generate_range_test.
10166 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10169 * config/i386/i386.md (*<btsc><mode>_mask): Add
10170 TARGET_USE_BT to insn constraint.
10171 (*btr<mode>_mask): Ditto.
10173 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
10175 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
10176 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
10178 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
10181 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
10182 TARGET_NEON and TARGET_IWMMXT.
10183 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
10184 TARGET_NEON and TARGET_IWMMXT.
10185 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
10187 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
10189 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
10190 (ix86_rewrite_tls_address): Ditto.
10191 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
10192 (ix86_rewrite_tls_address_1): Ditto.
10193 (ix86_rewrite_tls_address): Ditto.
10194 * config/i386/predicates.md (tls_address_pattern): New predicate.
10195 * config/i386/i386.md (TLS address splitter): New splitter.
10197 2017-09-04 Richard Biener <rguenther@suse.de>
10199 PR tree-optimization/82084
10200 * fold-const.h (can_native_encode_string_p): Declare.
10201 * fold-const.c (can_native_encode_string_p): Factor out from ...
10202 (native_encode_string): ... here.
10203 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
10204 vectorizing stores from constants we later cannot handle.
10206 2017-09-04 Marek Polacek <polacek@redhat.com>
10209 * doc/invoke.texi: Update -Wtautological-compare documentation.
10211 2017-09-04 Jeff Law <law@redhat.com>
10213 PR tree-optimization/64910
10214 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
10215 swap the first and last operand if the last is a constant.
10217 2017-09-04 Marek Polacek <polacek@redhat.com>
10220 * convert.c (do_narrow): When sanitizing signed integer overflows,
10221 bail out for signed types.
10222 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
10224 2017-09-04 Richard Biener <rguenther@suse.de>
10226 PR tree-optimization/82060
10227 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
10228 Move devirtualization after stmt folding and before EH/AB/noreturn
10229 cleanup to get the stmt refs canonicalized. Use a bool instead
10230 of gimple_modified_p since that doesn't work for NOPs. Schedule
10231 NOPs generated by folding for removal.
10233 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10234 Alan Hayward <alan.hayward@arm.com>
10235 David Sherwood <david.sherwood@arm.com>
10237 * coretypes.h (pad_direction): New enum.
10238 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
10239 (FUNCTION_ARG_PADDING): Likewise.
10240 * target.def (function_arg_padding): New hook.
10241 * targhooks.h (default_function_arg_padding): Declare.
10242 * targhooks.c (default_function_arg_padding): New function.
10243 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
10244 (TARGET_FUNCTION_ARG_PADDING): ...this.
10245 * doc/tm.texi: Regenerate.
10246 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
10247 instead of direction.
10248 (compute_argument_addresses): Likewise.
10249 (load_register_parameters): Likewise.
10250 (emit_library_call_value_1): Likewise.
10251 (store_one_arg): Use targetm.calls.function_arg_padding instead
10252 of FUNCTION_ARG_PADDING.
10253 (must_pass_in_stack_var_size_or_pad): Likewise.
10254 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
10255 (emit_group_store): Likewise.
10256 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
10257 instead of FUNCTION_ARG_PADDING.
10258 (emit_push_insn): Likewise, and propagate enum change throughout
10260 * function.h (direction): Delete.
10261 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
10263 * function.c (assign_parm_find_stack_rtl): Likewise.
10264 (assign_parm_setup_block_p): Likewise.
10265 (assign_parm_setup_block): Likewise.
10266 (gimplify_parameters): Likewise.
10267 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
10268 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
10270 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
10271 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10272 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
10273 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
10274 (aarch64_function_arg_padding): ...this new function.
10275 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
10276 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10277 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
10278 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10279 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
10280 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10281 (arm_pad_arg_upward): Replace with...
10282 (arm_function_arg_padding): ...this new function.
10283 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
10285 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
10286 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
10287 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10288 (ia64_hpux_function_arg_padding): Replace with...
10289 (ia64_function_arg_padding): ...this new function. Use pad_direction
10290 instead of direction. Check for TARGET_HPUX.
10291 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
10292 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10293 (iq2000_function_arg_padding): New function.
10294 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
10295 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
10296 (mips_function_arg_padding): ...this new function.
10297 (mips_pad_reg_upward): Update accordingly.
10298 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10299 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
10300 targetm.calls.function_arg_padding.
10301 (FUNCTION_ARG_PADDING): Delete.
10302 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
10303 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
10304 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10305 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
10306 (nios2_block_reg_padding): Return pad_direction instead of direction.
10307 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
10308 instead of direction.
10309 (nios2_function_arg_padding): Likewise. Make static.
10310 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10311 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
10312 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
10313 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
10314 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10315 (pa_function_arg_padding): Make static. Return pad_direction instead
10317 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
10318 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10319 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
10320 instead of direction. Use targetm.calls.function_arg_padding.
10321 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
10322 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10323 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
10324 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
10325 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
10327 (function_arg_padding): Rename to...
10328 (rs6000_function_arg_padding): ...this. Make static. Return
10329 pad_direction instead of direction.
10330 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10331 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
10332 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
10333 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
10334 instead of direction. Use targetm.calls.function_arg_padding.
10335 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
10336 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
10337 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
10338 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
10339 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10340 (function_arg_padding): Rename to...
10341 (rs6000_function_arg_padding): ...this. Make static. Return
10342 pad_direction instead of direction.
10343 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
10344 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
10345 * config/s390/s390.c (s390_function_arg_padding): New function.
10346 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10347 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
10348 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
10349 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
10350 (function_arg_padding): Rename to...
10351 (sparc_function_arg_padding): ...this. Make static. Return
10352 pad_direction instead of direction.
10353 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
10354 * config/spu/spu.c (spu_function_arg_padding): New function.
10355 (TARGET_FUNCTION_ARG_PADDING): Redefine.
10356 * system.h (FUNCTION_ARG_PADDING): Poison.
10358 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10359 Alan Hayward <alan.hayward@arm.com>
10360 David Sherwood <david.sherwood@arm.com>
10362 * target.def (modes_tieable_p): New hook.
10363 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
10364 (TARGET_MODES_TIEABLE_P): ...this.
10365 * doc/tm.texi.in: Regenerate.
10366 * hooks.h (hook_bool_mode_mode_true): Declare.
10367 * hooks.c (hook_bool_mode_mode_true): New function.
10368 * combine.c (subst): Use targetm.modes_tieable_p instead of
10370 * dse.c (find_shift_sequence): Likewise.
10371 * expmed.c (extract_low_bits): Likewise.
10372 * lower-subreg.c: Include target.h.
10373 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
10375 * rtlanal.c (rtx_cost): Likewise.
10376 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
10377 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
10378 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
10379 (TARGET_MODES_TIEABLE_P): Redefine.
10380 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
10381 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
10382 (TARGET_MODES_TIEABLE_P): Redefine.
10383 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
10384 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
10385 (arc_modes_tieable_p): New function.
10386 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
10387 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
10388 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
10389 (arm_modes_tieable_p): Make static.
10390 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
10391 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
10392 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
10393 (TARGET_MODES_TIEABLE_P): Redefine.
10394 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
10395 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
10396 (TARGET_MODES_TIEABLE_P): Redefine.
10397 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
10398 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
10399 (cr16_modes_tieable_p): New function.
10400 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
10401 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
10402 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
10403 (TRULY_NOOP_TRUNCATION): Update comment.
10404 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
10405 (TRULY_NOOP_TRUNCATION): Update comment.
10406 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
10407 (frv_modes_tieable_p): New function.
10408 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
10409 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
10410 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
10411 (TARGET_MODES_TIEABLE_P): Redefine.
10412 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
10413 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
10414 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
10415 (TARGET_MODES_TIEABLE_P): Redefine.
10416 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
10417 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
10418 (ia64_modes_tieable_p): New function.
10419 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
10420 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
10421 (iq2000_modes_tieable_p): New function.
10422 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
10423 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
10424 (lm32_modes_tieable_p): New function.
10425 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
10426 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
10427 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
10428 (TARGET_MODES_TIEABLE_P): Redefine.
10429 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
10430 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
10431 (m32r_modes_tieable_p): New function.
10432 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
10433 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
10434 (m68k_modes_tieable_p): New function.
10435 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
10436 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
10437 (mcore_modes_tieable_p): New function.
10438 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
10439 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
10441 (TARGET_MODES_TIEABLE_P): Redefine.
10442 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
10443 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
10444 * config/mips/mips.c (mips_modes_tieable_p): Make static.
10445 (TARGET_MODES_TIEABLE_P): Redefine.
10446 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
10447 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
10448 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
10449 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
10450 (mn10300_modes_tieable_p): ...this and make static.
10451 (TARGET_MODES_TIEABLE_P): Redefine.
10452 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
10453 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
10454 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
10455 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
10456 (msp430_modes_tieable_p): Make static.
10457 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
10458 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
10459 (TARGET_MODES_TIEABLE_P): Redefine.
10460 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
10461 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
10462 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
10463 (TARGET_MODES_TIEABLE_P): Redefine.
10464 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
10465 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
10466 * config/pa/pa.c (pa_modes_tieable_p): Make static.
10467 (TARGET_MODES_TIEABLE_P): Redefine.
10468 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
10469 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
10470 (pdp11_modes_tieable_p): New function.
10471 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
10472 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
10473 (rs6000_modes_tieable_p): New function.
10474 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10475 * config/powerpcspe/powerpcspe.md: Update comment.
10476 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
10477 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
10478 (TARGET_MODES_TIEABLE_P): Redefine.
10479 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
10480 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
10481 (rl78_modes_tieable_p): New function.
10482 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
10483 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
10484 (rs6000_modes_tieable_p): New function.
10485 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
10486 * config/rs6000/rs6000.md: Update comment.
10487 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
10488 * config/rx/rx.c (rx_modes_tieable_p): New function.
10489 (TARGET_MODES_TIEABLE_P): Redefine.
10490 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
10491 * config/s390/s390.c (s390_modes_tieable_p): New function.
10492 (TARGET_MODES_TIEABLE_P): Redefine.
10493 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
10494 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
10495 (sh_modes_tieable_p): New function.
10496 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
10497 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
10498 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
10499 (sparc_modes_tieable_p): Make static.
10500 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
10501 * config/spu/spu.c (spu_modes_tieable_p): New function.
10502 (TARGET_MODES_TIEABLE_P): Redefine.
10503 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
10504 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
10505 (TARGET_MODES_TIEABLE_P): Redefine.
10506 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
10507 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
10508 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
10509 * config/v850/v850.c (v850_modes_tieable_p): New function.
10510 (TARGET_MODES_TIEABLE_P): Redefine.
10511 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
10512 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
10513 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
10514 (visium_modes_tieable_p): New function.
10515 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
10516 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
10517 (xtensa_modes_tieable_p): New function.
10518 * system.h (MODES_TIEABLE_P): Poison.
10520 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10521 Alan Hayward <alan.hayward@arm.com>
10522 David Sherwood <david.sherwood@arm.com>
10524 * target.def (hard_regno_mode_ok): New hook.
10525 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
10526 (TARGET_HARD_REGNO_MODE_OK): ...this.
10527 * doc/tm.texi.in: Regenerate.
10528 * hooks.h (hook_bool_uint_mode_true): Declare.
10529 * hooks.c (hook_bool_uint_mode_true): New function.
10530 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
10531 HARD_REGNO_MODE_OK.
10532 * genpreds.c (write_insn_preds_c): Add an include of target.h.
10533 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
10534 instead of HARD_REGNO_MODE_OK.
10535 * caller-save.c: Include target.h.
10536 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
10537 HARD_REGNO_MODE_OK.
10538 * combine.c (can_combine_p): Likewise.
10539 (combinable_i3pat): Likewise.
10540 (can_change_dest_mode): Likewise.
10541 * expr.c (init_expr_target): Likewise.
10542 (convert_move): Likewise.
10543 (convert_modes): Likewise.
10544 * ira.c (setup_prohibited_class_mode_regs): Likewise.
10545 (setup_prohibited_mode_move_regs): Likewise.
10546 * ira.h (target_ira): Likewise.
10547 * lra-assigns.c (find_hard_regno_for_1): Likewise.
10548 * lra-constraints.c (process_alt_operands): Likewise.
10549 (split_reg): Likewise.
10550 * recog.c (peep2_find_free_register): Likewise.
10551 * ree.c (combine_reaching_defs): Likewise.
10552 * regcprop.c (maybe_mode_change): Likewise.
10553 * reginfo.c (init_reg_sets_1): Likewise.
10554 (choose_hard_reg_mode): Likewise.
10555 (simplifiable_subregs): Likewise.
10556 * regrename.c (check_new_reg_p): Likewise.
10557 * reload.c (find_valid_class): Likewise.
10558 (find_valid_class_1): Likewise.
10559 (reload_inner_reg_of_subreg): Likewise.
10560 (push_reload): Likewise.
10561 (combine_reloads): Likewise.
10562 (find_dummy_reload): Likewise.
10563 (find_reloads): Likewise.
10564 * reload1.c (find_reg): Likewise.
10565 (set_reload_reg): Likewise.
10566 (allocate_reload_reg): Likewise.
10567 (choose_reload_regs): Likewise.
10568 (reload_adjust_reg_for_temp): Likewise.
10569 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
10570 (simplify_subreg_regno): Likewise.
10571 * sel-sched.c (init_regs_for_mode): Likewise.
10572 * varasm.c (make_decl_rtl): Likewise.
10573 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
10574 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
10575 HARD_REGNO_MODE_OK.
10576 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
10577 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
10578 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10579 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
10580 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
10581 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10582 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
10583 (arc_mode_class): Delete.
10584 (HARD_REGNO_MODE_OK): Delete.
10585 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10586 (arc_hard_regno_mode_ok): Rename old array to...
10587 (arc_hard_regno_mode_ok_modes): ...this.
10588 (arc_conditional_register_usage): Update accordingly.
10589 (arc_mode_class): Make static.
10590 (arc_hard_regno_mode_ok): New function.
10591 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
10592 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
10593 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10594 (arm_hard_regno_mode_ok): Make static.
10595 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
10596 HARD_REGNO_MODE_OK.
10597 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
10598 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
10599 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
10601 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10602 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
10603 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
10604 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
10605 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
10606 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10607 * config/bfin/predicates.md (valid_reg_operand): Use
10608 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
10609 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
10610 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
10611 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10612 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
10613 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
10614 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10615 (cr16_hard_regno_mode_ok): Make static and return a bool.
10616 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
10617 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10618 (cris_hard_regno_mode_ok): New function.
10619 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
10620 (epiphany_mode_class): Delete.
10621 (HARD_REGNO_MODE_OK): Delete.
10622 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
10623 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10624 (hard_regno_mode_ok): Rename to...
10625 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
10626 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
10627 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
10628 HARD_REGNO_MODE_OK.
10629 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
10630 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
10631 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10632 (frv_hard_regno_mode_ok): Make static and return a bool.
10633 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
10634 HARD_REGNO_MODE_OK.
10635 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
10636 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
10637 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
10638 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
10640 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10641 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
10642 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
10643 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
10645 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10646 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
10647 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10648 (ia64_hard_regno_mode_ok): New function.
10649 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
10650 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10651 (iq2000_hard_regno_mode_ok): New function.
10652 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
10653 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10654 (lm32_hard_regno_mode_ok): New function.
10655 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
10656 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
10657 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
10658 instead of HARD_REGNO_MODE_OK.
10659 (m32c_hard_regno_ok): Rename to...
10660 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
10661 (m32c_cannot_change_mode_class): Update accordingly.
10662 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10663 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
10664 (m32r_mode_class): Delete.
10665 (HARD_REGNO_MODE_OK): Delete.
10666 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10667 (m32r_hard_regno_mode_ok): Rename to...
10668 (m32r_hard_regno_modes): ...this.
10669 (m32r_mode_class): Make static.
10670 (m32r_hard_regno_mode_ok): New function.
10671 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
10672 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
10673 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10674 (m68k_hard_regno_mode_ok): Make static.
10675 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
10676 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10677 (mcore_hard_regno_mode_ok): New function.
10678 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
10679 (HARD_REGNO_MODE_OK): Delete.
10680 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
10682 (microblaze_hard_regno_mode_ok_p): ...this and make static.
10683 (microblaze_hard_regno_mode_ok): New function.
10684 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10685 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
10686 (mips_hard_regno_mode_ok): Delete.
10687 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
10688 (mips_hard_regno_mode_ok_p): ...this and make static.
10689 (mips_hard_regno_mode_ok_p): Rename to...
10690 (mips_hard_regno_mode_ok_uncached): ...this.
10691 (mips_hard_regno_mode_ok): New function.
10692 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
10693 of HARD_REGNO_MODE_OK.
10694 (mips_option_override): Update after above name changes.
10695 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10696 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
10697 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
10698 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
10699 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
10700 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10701 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
10702 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
10703 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
10704 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10705 (msp430_hard_regno_mode_ok): Make static and return a bool.
10706 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
10707 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
10708 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
10710 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10711 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
10712 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
10713 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
10714 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
10715 (PA_HARD_REGNO_MODE_OK): ...this
10716 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
10717 (PA_HARD_REGNO_MODE_OK): ...this.
10718 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10719 (pa_hard_regno_mode_ok): New function.
10720 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
10721 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10722 (pdp11_hard_regno_mode_ok): New function.
10723 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
10724 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
10726 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
10728 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10729 (rs6000_hard_regno_mode_ok): Rename to...
10730 (rs6000_hard_regno_mode_ok_uncached): ...this.
10731 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10732 (rs6000_hard_regno_mode_ok): New function.
10733 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
10734 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
10735 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
10736 (riscv_hard_regno_mode_ok): ...this and make static.
10737 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10738 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
10739 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
10740 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10741 (rl78_hard_regno_mode_ok): Make static and return bool.
10742 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
10743 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
10745 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
10746 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10747 (rs6000_hard_regno_mode_ok): Rename to...
10748 (rs6000_hard_regno_mode_ok_uncached): ...this.
10749 (rs6000_init_hard_regno_mode_ok): Update accordingly.
10750 (rs6000_hard_regno_mode_ok): New function.
10751 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
10752 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
10753 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10754 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
10755 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
10756 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
10757 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10758 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
10759 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
10760 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10761 (sh_hard_regno_mode_ok): Make static.
10762 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
10763 instead of HARD_REGNO_MODE_OK.
10764 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
10765 (sparc_mode_class): Delete.
10766 (HARD_REGNO_MODE_OK): Delete.
10767 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10768 (hard_regno_mode_classes): Make static.
10769 (sparc_mode_class): Likewise.
10770 (sparc_hard_regno_mode_ok): New function.
10771 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
10772 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
10773 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
10775 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10776 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
10777 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
10778 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
10779 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
10780 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10781 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
10782 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
10783 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
10784 (visium_hard_regno_mode_ok): New function.
10785 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
10786 instead of HARD_REGNO_MODE_OK.
10787 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
10788 (HARD_REGNO_MODE_OK): Delete.
10789 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
10790 (xtensa_hard_regno_mode_ok_p): ...this and make static.
10791 (xtensa_option_override): Update accordingly.
10792 (TARGET_HARD_REGNO_MODE_OK): Redefine.
10793 (xtensa_hard_regno_mode_ok): New function.
10794 * system.h (HARD_REGNO_MODE_OK): Poison.
10796 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10797 Alan Hayward <alan.hayward@arm.com>
10798 David Sherwood <david.sherwood@arm.com>
10800 * target.def (hard_regno_call_part_clobbered): New hook.
10801 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
10802 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
10803 * doc/tm.texi: Regenerate.
10804 * hooks.h (hook_bool_uint_mode_false): Declare.
10805 * hooks.c (hook_bool_uint_mode_false): New function.
10806 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10807 * cselib.c (cselib_process_insn): Use
10808 targetm.hard_regno_call_part_clobbered instead of
10809 HARD_REGNO_CALL_PART_CLOBBERED.
10810 * ira-conflicts.c (ira_build_conflicts): Likewise.
10811 * ira-costs.c (ira_tune_allocno_costs): Likewise.
10812 * lra-constraints.c (need_for_call_save_p): Likewise.
10813 * lra-lives.c: Include target.h.
10814 (check_pseudos_live_through_calls): Use
10815 targetm.hard_regno_call_part_clobbered instead of
10816 HARD_REGNO_CALL_PART_CLOBBERED.
10817 * regcprop.c: Include target.h.
10818 (copyprop_hardreg_forward_1): Use
10819 targetm.hard_regno_call_part_clobbered instead of
10820 HARD_REGNO_CALL_PART_CLOBBERED.
10821 * reginfo.c (choose_hard_reg_mode): Likewise.
10822 * regrename.c (check_new_reg_p): Likewise.
10823 * reload.c (find_equiv_reg): Likewise.
10824 * reload1.c (emit_reload_insns): Likewise.
10825 * sched-deps.c (deps_analyze_insn): Likewise.
10826 * sel-sched.c (init_regs_for_mode): Likewise.
10827 (mark_unavailable_hard_regs): Likewise.
10828 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
10829 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10830 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
10832 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10833 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10834 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
10836 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
10838 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10839 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10840 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
10842 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10843 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10844 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
10846 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10847 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
10849 * config/powerpcspe/powerpcspe.c
10850 (rs6000_hard_regno_call_part_clobbered): New function.
10851 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10852 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10853 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
10855 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10856 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10857 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
10859 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
10860 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
10861 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
10863 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10864 Alan Hayward <alan.hayward@arm.com>
10865 David Sherwood <david.sherwood@arm.com>
10867 * rtl.h (subreg_memory_offset): Declare.
10868 * emit-rtl.c (subreg_memory_offset): New function.
10869 * expmed.c (store_bit_field_1): Use it.
10870 * expr.c (undefined_operand_subword_p): Likewise.
10871 * simplify-rtx.c (simplify_subreg): Likewise.
10873 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
10875 PR rtl-optimization/57448
10878 * optabs.c (expand_atomic_load): Place compiler memory barriers if
10879 using atomic_load pattern.
10880 (expand_atomic_store): Likewise.
10882 2017-09-04 Jakub Jelinek <jakub@redhat.com>
10885 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
10886 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
10887 handling. Use replace_call_with_value with NULL instead of
10888 gsi_replace, unlink_stmt_vdef and release_defs.
10890 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
10893 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
10895 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
10898 * rtl.h (emit_library_call_value_1): Declare.
10899 (emit_library_call): Replace declaration with a series of overloads.
10900 Remove the parameter count argument.
10901 (emit_library_call_value): Likewise.
10902 * calls.c (emit_library_call_value_1): Make global. Replace varargs
10903 with an "rtx_mode_t *".
10904 (emit_library_call_value): Delete.
10905 (emit_library_call): Likewise.
10906 * asan.c (asan_emit_stack_protection): Update calls accordingly.
10907 (asan_emit_allocas_unpoison): Likewise.
10908 * builtins.c (expand_builtin_powi): Likewise.
10909 (expand_asan_emit_allocas_unpoison): Likewise.
10910 * cfgexpand.c (expand_main_function): Likewise.
10911 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
10912 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
10913 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
10914 * config/arm/arm.c (arm_trampoline_init): Likewise.
10915 (arm_call_tls_get_addr): Likewise.
10916 (arm_expand_divmod_libfunc): Likewise.
10917 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
10918 (smulsi3_highpart): Likewise.
10919 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
10920 (c6x_expand_compare): Likewise.
10921 (c6x_expand_movmem): Likewise.
10922 * config/frv/frv.c (frv_trampoline_init): Likewise.
10923 * config/i386/i386.c (ix86_trampoline_init): Likewise.
10924 (ix86_expand_divmod_libfunc): Likewise.
10925 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
10926 (ia64_expand_compare): Likewise.
10927 (ia64_profile_hook): Likewise.
10928 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
10929 (nonlocal_goto): Likewise.
10930 (restore_stack_nonlocal): Likewise.
10931 * config/m32r/m32r.c (block_move_call): Likewise.
10932 (m32r_trampoline_init): Likewise.
10933 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
10934 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
10935 (m68k_call_m68k_read_tp): Likewise.
10936 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
10937 (microblaze_expand_divide): Likewise.
10938 * config/mips/mips.h (mips_args): Likewise.
10939 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
10940 (MIPS_ICACHE_SYNC): Likewise.
10941 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
10942 (nios2_trampoline_init): Likewise.
10943 * config/pa/pa.c (hppa_tls_call): Likewise.
10944 (pa_trampoline_init): Likewise.
10945 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
10946 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
10947 (expand_strn_compare): Likewise.
10948 (rs6000_generate_compare): Likewise.
10949 (rs6000_expand_float128_convert): Likewise.
10950 (output_profile_hook): Likewise.
10951 (rs6000_trampoline_init): Likewise.
10952 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
10953 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
10954 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
10955 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
10956 (rs6000_generate_compare): Likewise.
10957 (rs6000_expand_float128_convert): Likewise.
10958 (output_profile_hook): Likewise.
10959 (rs6000_trampoline_init): Likewise.
10960 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
10961 * config/sh/sh.c (sh_trampoline_init): Likewise.
10962 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
10963 (sparc_emit_float_lib_cmp): Likewise.
10964 (sparc32_initialize_trampoline): Likewise.
10965 (sparc64_initialize_trampoline): Likewise.
10966 (sparc_profile_hook): Likewise.
10967 * config/spu/spu.c (ea_load_store): Likewise.
10968 * config/spu/spu.md (floatunssidf2): Likewise.
10969 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
10970 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
10971 * config/visium/visium.c (expand_block_move_4): Likewise.
10972 (expand_block_move_2): Likewise.
10973 (expand_block_move_1): Likewise.
10974 (expand_block_set_4): Likewise.
10975 (expand_block_set_2): Likewise.
10976 (expand_block_set_1): Likewise.
10977 (visium_trampoline_init): Likewise.
10978 (visium_profile_hook): Likewise.
10979 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
10980 (xtensa_setup_frame_addresses): Likewise.
10981 (xtensa_trampoline_init): Likewise.
10982 * except.c (sjlj_emit_function_enter): Likewise.
10983 (sjlj_emit_function_exit): Likewise.
10984 * explow.c (allocate_dynamic_stack_space): Likewise.
10985 (probe_stack_range): Likewise.
10986 * expr.c (convert_mode_scalar): Likewise.
10987 * optabs.c (expand_binop): Likewise.
10988 (expand_twoval_binop_libfunc): Likewise.
10989 (expand_unop): Likewise.
10990 (prepare_cmp_insn): Likewise.
10991 (prepare_float_lib_cmp): Likewise.
10992 (expand_float): Likewise.
10993 (expand_fix): Likewise.
10994 (expand_fixed_convert): Likewise.
10995 (maybe_emit_sync_lock_test_and_set): Likewise.
10996 (expand_atomic_compare_and_swap): Likewise.
10997 (expand_mem_thread_fence): Likewise.
10998 (expand_atomic_fetch_op): Likewise.
11000 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
11002 * doc/generic.texi (OpenACC): Adjust URL.
11003 * doc/invoke.texi (C Dialect Options): Ditto.
11005 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
11007 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
11008 predicate for operand 1. Add (m,<S>) constraint.
11009 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
11010 Prevent memory operand 1 with register operand 2.
11012 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
11014 PR rtl-optimization/82024
11015 * combine.c (try_combine): If the combination result is a PARALLEL,
11016 and we only need to retain the SET in there that would be placed
11017 at I2, check that we can place that at I3 instead, before doing so.
11019 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11022 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
11024 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
11025 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
11028 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
11029 Jeff Law <law@redhat.com>
11031 * varasm.c (bss_initializer_p): Do not put constants into .bss
11032 (categorize_decl_for_section): Handle bss_initializer_p returning
11033 false when DECL_INITIAL is NULL.
11035 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11038 * config/s390/s390.c (s390_can_inline_p): New function.
11040 2017-09-01 Jeff Law <law@redhat.com>
11042 PR tree-optimization/82052
11043 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
11044 Always initialize the returned slot after a hash table miss
11045 when INSERT is true.
11047 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
11049 * config/s390/s390.md (mem_signal_fence): Remove.
11050 * doc/md.texi (mem_signal_fence): Remove.
11051 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
11053 * target-insns.def (mem_signal_fence): Remove.
11055 2017-09-01 Jakub Jelinek <jakub@redhat.com>
11058 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
11061 * asan.c (create_odr_indicator): Strip name encoding from assembler
11062 name before appending it after __odr_asan_.
11064 2017-09-01 Martin Liska <mliska@suse.cz>
11066 PR tree-optimization/82059
11067 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
11068 frequency only when an edge is redirected.
11070 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11072 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
11073 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
11074 (arc_conditional_register_usage): Remove ARC600 lp_count
11076 (arc_file_start): Emit Tag_ARC_CPU_variation.
11077 (arc_can_use_doloop_p): New conditions to use ZOLs.
11078 (hwloop_fail): New function.
11079 (hwloop_optimize): Likewise.
11080 (hwloop_pattern_reg): Likewise.
11081 (arc_doloop_hooks): New struct, to be used with reorg_loops.
11082 (arc_reorg_loops): New function, calls reorg_loops.
11083 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
11084 (arc600_corereg_hazard): Remove ZOL checking, case handled by
11086 (arc_loop_hazard): Remove function, functionality moved into
11088 (arc_hazard): Remove arc_loop_hazard call.
11089 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
11090 into hwloop_optimize.
11091 (arc_label_align): Remove ZOL handling.
11092 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
11093 * config/arc/arc.md (doloop_begin): Remove pattern.
11094 (doloop_begin_i): Likewise.
11095 (doloop_end_i): Likewise.
11096 (doloop_fallback): Likewise.
11097 (doloop_fallback_m): Likewise.
11098 (doloop_end): Reimplement expand.
11099 (arc_lp): New pattern for LP instruction.
11100 (loop_end): New pattern.
11101 (loop_fail): Likewise.
11102 (decrement_and_branch_until_zero): Likewise.
11103 * config/arc/arc.opt (mlpc-width): New option.
11104 * doc/invoke.texi (mlpc-width): Document option.
11106 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11108 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
11109 (arc_ccfsm_advance): Fix checking for delay slots.
11110 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
11112 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11114 * config/arc/arc.md (movqi_insn): Add stores to save constant long
11116 (movhi_insn): Update store instruction constraint which are saving
11117 6-bit short immediates.
11118 (movsi_insn): Consider also short scaled load operations.
11119 (zero_extendhisi2_i): Use Usd constraint instead of T.
11120 (extendhisi2_i): Add q constraint.
11121 (arc_clzsi2): Add type and length attributes.
11122 (arc_ctzsi2): Likewise.
11123 * config/arc/constraints.md (Usc): Update constraint, the
11124 assembler can parse two relocations for a single instruction.
11126 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
11128 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
11129 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
11131 2017-08-31 Olivier Hainque <hainque@adacore.com>
11133 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
11134 match as powerpc-wrs-vxworks*.
11136 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
11138 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
11139 register constraint for by-element operand.
11140 (aarch64_mls_elt_merge<mode>): Likewise.
11142 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11144 * config/arc/arc.c (arc_can_follow_jump): Check for short
11147 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11149 * config.gcc: Use g.opt for arc.
11150 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
11151 functionality moved to ...
11152 (legitimate_scaled_address_p): New function, ...here.
11153 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
11154 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
11155 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
11157 (arc_override_options): Handle G option.
11158 (arc_output_pic_addr_const): Correct function definition.
11159 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
11160 (arc_decl_anon_ns_mem_p): Delete.
11161 (arc_in_small_data_p): Overhaul this function to take into
11162 consideration the value given via G option.
11163 (arc_rewrite_small_data_1): Renamed and corrected old
11164 arc_rewrite_small_data function.
11165 (arc_rewrite_small_data): New function.
11166 (small_data_pattern): Don't use pic_offset_table_rtx.
11167 * config/arc/arc.h (CC1_SPEC): Recognize G option.
11168 * config/arc/simdext.md (movmisalignv2hi): Use
11169 prepare_move_operands function.
11171 (movmisalign*): Likewise.
11172 * doc/invoke.texi (ARC options): Document -G option.
11174 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
11176 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
11178 * config/arc/arc.c (arc_print_operand): Output scalled address for
11179 sdata whenever is possible.
11180 (arc_in_small_data_p): Allow sdata for 64bit datum when double
11181 load/stores are available.
11182 (compact_sda_memory_operand): Check for the alignment required by
11183 code density instructions.
11184 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
11186 * config/arc/constraints.md (Usd): Update constraint.
11187 (Us0): New constraint.
11188 (Usc): Update constraint.
11190 2017-08-31 Richard Biener <rguenther@suse.de>
11192 PR middle-end/82054
11193 * dwarf2out.c (dwarf2out_early_global_decl): Process each
11194 function only once.
11196 2017-08-31 Tamar Christina <tamar.christina@arm.com>
11198 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
11199 Resize type_signature.
11201 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11202 Alan Hayward <alan.hayward@arm.com>
11203 David Sherwood <david.sherwood@arm.com>
11205 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
11206 subregs whose inner modes can be stored in GPRs.
11207 (aarch64_classify_index): Likewise.
11209 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11210 Alan Hayward <alan.hayward@arm.com>
11211 David Sherwood <david.sherwood@arm.com>
11213 * config/aarch64/iterators.md (V_cmp_result): Rename to...
11214 (V_INT_EQUIV): ...this.
11215 (v_cmp_result): Rename to...
11216 (v_int_equiv): ...this.
11217 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
11218 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
11219 (copysign<mode>3): Likewise.
11220 (aarch64_simd_bsl<mode>_internal): Likewise.
11221 (aarch64_simd_bsl<mode>): Likewise.
11222 (vec_cmp<mode><mode>): Likewise.
11223 (vcond<mode><mode>): Likewise.
11224 (vcond<v_cmp_mixed><mode>): Likewise.
11225 (vcondu<mode><v_cmp_mixed>): Likewise.
11226 (aarch64_cm<optab><mode>): Likewise.
11227 (aarch64_cmtst<mode>): Likewise.
11228 (aarch64_fac<optab><mode>): Likewise.
11229 (vec_perm_const<mode>): Likewise.
11230 (vcond_mask_<mode><v_cmp_result>): Rename to...
11231 (vcond_mask_<mode><v_int_equiv>): ...this.
11232 (vec_cmp<mode><v_cmp_result>): Rename to...
11233 (vec_cmp<mode><v_int_equiv>): ...this.
11235 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
11236 Alan Hayward <alan.hayward@arm.com>
11237 David Sherwood <david.sherwood@arm.com>
11239 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
11241 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
11242 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
11243 (UNSPEC_LD4_DREG): New unspecs.
11244 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
11245 (aarch64_ld2<mode>_dreg_be): Replace with...
11246 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
11248 (aarch64_ld3<mode>_dreg_le)
11249 (aarch64_ld3<mode>_dreg_be): Replace with...
11250 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
11252 (aarch64_ld4<mode>_dreg_le)
11253 (aarch64_ld4<mode>_dreg_be): Replace with...
11254 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
11257 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11259 PR tree-optimization/81987
11260 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
11261 insert an initializer in a location not dominated by the stride
11264 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11266 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
11267 on the entire header of the finally block in the fallthru case.
11269 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
11271 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
11273 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
11275 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
11276 rs6000_emit_move_from_cr and call renamed function.
11277 (rs6000_emit_prologue): Call renamed functions.
11278 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
11279 movesi_from_cr, remove volatile CRs.
11281 2017-08-30 Jon Beniston <jon@beniston.com>
11282 Richard Biener <rguenther@suse.de>
11284 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
11285 of VECTOR_MODE_P check.
11286 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
11287 element vector types.
11289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11291 * df.h (df_read_modify_subreg_p): Remove in favor of...
11292 * rtl.h (read_modify_subreg_p): ...this new function. Take a
11293 const_rtx instead of an rtx.
11294 * cprop.c (local_cprop_find_used_regs): Update accordingly.
11295 * df-problems.c (df_word_lr_mark_ref): Likewise.
11296 * ira-lives.c (mark_pseudo_reg_live): Likewise.
11297 (mark_pseudo_reg_dead): Likewise.
11298 (mark_ref_dead): Likewise.
11299 * reginfo.c (init_subregs_of_mode): Likewise.
11300 * sched-deps.c (sched_analyze_1): Likewise.
11301 * df-scan.c (df_def_record_1): Likewise.
11302 (df_uses_record): Likewise.
11303 (df_read_modify_subreg_p): Remove in favor of...
11304 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
11305 const_rtx instead of an rtx.
11307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11308 Alan Hayward <alan.hayward@arm.com>
11309 David Sherwood <david.sherwood@arm.com>
11311 * rtl.h (partial_subreg_p): New function.
11312 * caller-save.c (save_call_clobbered_regs): Use it.
11313 * calls.c (expand_call): Likewise.
11314 * combine.c (combinable_i3pat): Likewise.
11315 (simplify_set): Likewise.
11316 (make_extraction): Likewise.
11317 (make_compound_operation_int): Likewise.
11318 (gen_lowpart_or_truncate): Likewise.
11319 (force_to_mode): Likewise.
11320 (make_field_assignment): Likewise.
11321 (reg_truncated_to_mode): Likewise.
11322 (record_truncated_value): Likewise.
11323 (move_deaths): Likewise.
11324 * cse.c (record_jump_cond): Likewise.
11325 (cse_insn): Likewise.
11326 * cselib.c (cselib_lookup_1): Likewise.
11327 * expmed.c (extract_bit_field_using_extv): Likewise.
11328 * function.c (assign_parm_setup_reg): Likewise.
11329 * ifcvt.c (noce_convert_multiple_sets): Likewise.
11330 * ira-build.c (create_insn_allocnos): Likewise.
11331 * lra-coalesce.c (merge_pseudos): Likewise.
11332 * lra-constraints.c (match_reload): Likewise.
11333 (simplify_operand_subreg): Likewise.
11334 (curr_insn_transform): Likewise.
11335 * lra-lives.c (process_bb_lives): Likewise.
11336 * lra.c (new_insn_reg): Likewise.
11337 (lra_substitute_pseudo): Likewise.
11338 * regcprop.c (mode_change_ok): Likewise.
11339 (maybe_mode_change): Likewise.
11340 (copyprop_hardreg_forward_1): Likewise.
11341 * reload.c (push_reload): Likewise.
11342 (find_reloads): Likewise.
11343 (find_reloads_subreg_address): Likewise.
11344 * reload1.c (alter_reg): Likewise.
11345 (eliminate_regs_1): Likewise.
11346 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11348 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
11350 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
11351 back to if statements, including unpack.
11353 2017-08-30 Martin Liska <mliska@suse.cz>
11355 PR inline-asm/82001
11356 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
11358 (func_checker::compare_asm_inputs_outputs): ... this function.
11359 (func_checker::compare_gimple_asm): Use the function to compare
11360 also ASM constrains.
11361 * ipa-icf-gimple.h: Rename the function.
11363 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11364 Alan Hayward <alan.hayward@arm.com>
11365 David Sherwood <david.sherwood@arm.com>
11367 * coretypes.h (complex_mode): New type.
11368 * gdbhooks.py (build_pretty_printer): Handle it.
11369 * machmode.h (complex_mode): New class.
11370 (complex_mode::includes_p): New function.
11371 (is_complex_int_mode): Likewise.
11372 (is_complex_float_mode): Likewise.
11373 * genmodes.c (get_mode_class): Handle complex mode classes.
11374 * function.c (expand_function_end): Use is_complex_int_mode.
11376 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11377 Alan Hayward <alan.hayward@arm.com>
11378 David Sherwood <david.sherwood@arm.com>
11380 * coretypes.h (scalar_mode_pod): New typedef.
11381 * gdbhooks.py (build_pretty_printer): Handle it.
11382 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
11383 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
11384 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
11385 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
11386 as_a <scalar_mode>.
11388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11389 Alan Hayward <alan.hayward@arm.com>
11390 David Sherwood <david.sherwood@arm.com>
11392 * machmode.h (mode_for_vector): Take a scalar_mode instead
11394 * stor-layout.c (mode_for_vector): Likewise.
11395 * explow.c (promote_mode): Use as_a <scalar_mode>.
11396 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
11398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11399 Alan Hayward <alan.hayward@arm.com>
11400 David Sherwood <david.sherwood@arm.com>
11402 * target.def (preferred_simd_mode): Take a scalar_mode
11403 instead of a machine_mode.
11404 * targhooks.h (default_preferred_simd_mode): Likewise.
11405 * targhooks.c (default_preferred_simd_mode): Likewise.
11406 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
11407 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
11408 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
11409 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
11410 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
11411 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
11412 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
11413 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
11415 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
11416 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
11417 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
11418 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
11419 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
11420 * doc/tm.texi: Regenerate.
11421 * optabs-query.c (can_vec_mask_load_store_p): Return false for
11424 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11425 Alan Hayward <alan.hayward@arm.com>
11426 David Sherwood <david.sherwood@arm.com>
11428 * target.def (scalar_mode_supported_p): Take a scalar_mode
11429 instead of a machine_mode.
11430 * targhooks.h (default_scalar_mode_supported_p): Likewise.
11431 * targhooks.c (default_scalar_mode_supported_p): Likewise.
11432 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
11433 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11434 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
11435 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
11436 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
11437 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
11438 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
11439 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
11440 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
11441 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
11442 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
11443 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
11445 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
11446 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
11447 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11448 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11449 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
11451 * doc/tm.texi: Regenerate.
11453 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11454 Alan Hayward <alan.hayward@arm.com>
11455 David Sherwood <david.sherwood@arm.com>
11457 * coretypes.h (opt_scalar_mode): New typedef.
11458 * gdbhooks.py (build_pretty_printers): Handle it.
11459 * machmode.h (mode_iterator::get_2xwider): Add overload for
11461 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
11463 * expr.c (convert_mode_scalar): Likewise.
11464 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11465 * optabs.c (expand_float): Likewise.
11466 (expand_fix): Likewise.
11467 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11469 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11470 Alan Hayward <alan.hayward@arm.com>
11471 David Sherwood <david.sherwood@arm.com>
11473 * optabs.c (expand_float): Explicitly check for scalars before
11474 using a branching expansion.
11475 (expand_fix): Likewise.
11477 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11478 Alan Hayward <alan.hayward@arm.com>
11479 David Sherwood <david.sherwood@arm.com>
11481 * expr.c (convert_mode): Split scalar handling out into...
11482 (convert_mode_scalar): ...this new function. Treat the modes
11485 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11486 Alan Hayward <alan.hayward@arm.com>
11487 David Sherwood <david.sherwood@arm.com>
11489 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
11491 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
11493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11494 Alan Hayward <alan.hayward@arm.com>
11495 David Sherwood <david.sherwood@arm.com>
11497 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
11498 rather than a machine_mode.
11499 (fixed_from_string): Likewise.
11500 (fixed_convert): Likewise.
11501 (fixed_convert_from_int): Likewise.
11502 (fixed_convert_from_real): Likewise.
11503 (real_convert_from_fixed): Likewise.
11504 * fixed-value.c (fixed_from_double_int): Likewise.
11505 (fixed_from_string): Likewise.
11506 (fixed_convert): Likewise.
11507 (fixed_convert_from_int): Likewise.
11508 (fixed_convert_from_real): Likewise.
11509 (real_convert_from_fixed): Likewise.
11510 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
11512 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11513 Alan Hayward <alan.hayward@arm.com>
11514 David Sherwood <david.sherwood@arm.com>
11516 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
11517 of separate mode class checks. Do not allow vector modes here.
11518 (immed_wide_int_const): Use as_a <scalar_mode>.
11519 * explow.c (trunc_int_for_mode): Likewise.
11520 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
11521 (wi::shwi): Likewise.
11522 (wi::min_value): Likewise.
11523 (wi::max_value): Likewise.
11524 * dwarf2out.c (loc_descriptor): Likewise.
11525 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
11526 for CONST_WIDE_INT.
11528 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11529 Alan Hayward <alan.hayward@arm.com>
11530 David Sherwood <david.sherwood@arm.com>
11532 * tree.h (SCALAR_TYPE_MODE): New macro.
11533 * expr.c (expand_expr_addr_expr_1): Use it.
11534 (expand_expr_real_2): Likewise.
11535 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
11536 (fold_convert_const_fixed_from_int): Likewise.
11537 (fold_convert_const_fixed_from_real): Likewise.
11538 (native_encode_fixed): Likewise
11539 (native_encode_complex): Likewise
11540 (native_encode_vector): Likewise.
11541 (native_interpret_fixed): Likewise.
11542 (native_interpret_real): Likewise.
11543 (native_interpret_complex): Likewise.
11544 (native_interpret_vector): Likewise.
11545 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
11546 (simd_clone_adjust_argument_types): Likewise.
11547 (simd_clone_init_simd_arrays): Likewise.
11548 (simd_clone_adjust): Likewise.
11549 * stor-layout.c (layout_type): Likewise.
11550 * tree.c (build_minus_one_cst): Likewise.
11551 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11552 * tree-inline.c (estimate_move_cost): Likewise.
11553 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
11554 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
11555 (vectorizable_reduction): Likewise.
11556 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
11557 (vect_recog_mixed_size_cond_pattern): Likewise.
11558 (check_bool_pattern): Likewise.
11559 (adjust_bool_pattern): Likewise.
11560 (search_type_for_mask_1): Likewise.
11561 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11562 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11563 (vectorizable_load): Likewise.
11564 (vectorizable_store): Likewise.
11565 * ubsan.c (ubsan_encode_value): Likewise.
11566 * varasm.c (output_constant): Likewise.
11568 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11569 Alan Hayward <alan.hayward@arm.com>
11570 David Sherwood <david.sherwood@arm.com>
11572 * coretypes.h (scalar_mode): New class.
11573 * machmode.h (scalar_mode): Likewise.
11574 (scalar_mode::includes_p): New function.
11575 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
11576 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
11577 * genmodes.c (get_mode_class): Handle remaining scalar modes.
11578 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
11579 * expmed.c (store_bit_field_1): Likewise.
11580 (extract_bit_field_1): Likewise.
11581 * expr.c (write_complex_part): Likewise.
11582 (read_complex_part): Likewise.
11583 (emit_move_complex_push): Likewise.
11584 (expand_expr_real_2): Likewise.
11585 * function.c (assign_parm_setup_reg): Likewise.
11586 (assign_parms_unsplit_complex): Likewise.
11587 * optabs.c (expand_binop): Likewise.
11588 * rtlanal.c (subreg_get_info): Likewise.
11589 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11590 * varasm.c (output_constant_pool_2): Likewise.
11592 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11593 Alan Hayward <alan.hayward@arm.com>
11594 David Sherwood <david.sherwood@arm.com>
11596 * expmed.c (extract_high_half): Use scalar_int_mode and remove
11598 (expmed_mult_highpart_optab): Likewise.
11599 (expmed_mult_highpart): Likewise.
11601 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11602 Alan Hayward <alan.hayward@arm.com>
11603 David Sherwood <david.sherwood@arm.com>
11605 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
11606 instead of a machine_mode.
11607 (builtin_memset_read_str): Likewise.
11608 * builtins.c (c_readstr): Likewise.
11609 (builtin_memcpy_read_str): Likewise.
11610 (builtin_strncpy_read_str): Likewise.
11611 (builtin_memset_read_str): Likewise.
11612 (builtin_memset_gen_str): Likewise.
11613 (expand_builtin_signbit): Use scalar_int_mode for local variables.
11614 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
11615 instead of a machine_mode.
11616 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
11618 (make_extraction): Likewise.
11619 (try_widen_shift_mode): Take and return scalar_int_modes instead
11621 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
11622 a scalar_int_mode instead of a machine_mode.
11623 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
11624 (avr_addr_space_pointer_mode): Likewise.
11625 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
11626 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
11627 (msp430_unwind_word_mode): Likewise.
11628 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
11629 (spu_addr_space_pointer_mode): Likewise.
11630 (spu_addr_space_address_mode): Likewise.
11631 (spu_libgcc_cmp_return_mode): Likewise.
11632 (spu_libgcc_shift_count_mode): Likewise.
11633 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
11634 (rl78_addr_space_pointer_mode): Likewise.
11635 (fl78_unwind_word_mode): Likewise.
11636 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
11638 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
11639 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
11640 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
11641 (mips_valid_pointer_mode): Likewise.
11642 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
11643 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
11644 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
11646 (ft32_addr_space_address_mode): Likewise.
11647 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
11648 scalar_int_mode instead of a machine_mode.
11649 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
11651 (m32c_addr_space_address_mode): Likewise.
11652 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
11653 (rs6000_eh_return_filter_mode): Likewise.
11654 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
11655 (rs6000_eh_return_filter_mode): Likewise.
11656 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
11657 (s390_libgcc_shift_count_mode): Likewise.
11658 (s390_unwind_word_mode): Likewise.
11659 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
11661 * target.def (mode_rep_extended): Likewise.
11662 (valid_pointer_mode): Likewise.
11663 (addr_space.valid_pointer_mode): Likewise.
11664 (eh_return_filter_mode): Return a scalar_int_mode rather than
11666 (libgcc_cmp_return_mode): Likewise.
11667 (libgcc_shift_count_mode): Likewise.
11668 (unwind_word_mode): Likewise.
11669 (addr_space.pointer_mode): Likewise.
11670 (addr_space.address_mode): Likewise.
11671 * doc/tm.texi: Regenerate.
11672 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
11674 (do_jump): Use scalar_int_mode for local variables.
11675 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
11676 rather than a machine_mode.
11677 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
11678 (scompare_loc_descriptor_wide): Likewise.
11679 (scompare_loc_descriptor_narrow): Likewise.
11680 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
11682 * except.c (sjlj_emit_dispatch_table): Likewise.
11683 (expand_builtin_eh_copy_values): Likewise.
11684 * explow.c (convert_memory_address_addr_space_1): Likewise.
11685 Take a scalar_int_mode rather than a machine_mode.
11686 (convert_memory_address_addr_space): Take a scalar_int_mode rather
11687 than a machine_mode.
11688 (memory_address_addr_space): Use scalar_int_mode for local variables.
11689 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
11690 rather than a machine_mode.
11691 * expmed.c (mask_rtx): Likewise.
11692 (init_expmed_one_conv): Likewise.
11693 (expand_mult_highpart_adjust): Likewise.
11694 (extract_high_half): Likewise.
11695 (expmed_mult_highpart_optab): Likewise.
11696 (expmed_mult_highpart): Likewise.
11697 (expand_smod_pow2): Likewise.
11698 (expand_sdiv_pow2): Likewise.
11699 (emit_store_flag_int): Likewise.
11700 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
11702 (extract_low_bits): Likewise.
11703 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
11705 * expr.c (pieces_addr::adjust): Likewise.
11706 (can_store_by_pieces): Likewise.
11707 (store_by_pieces): Likewise.
11708 (clear_by_pieces_1): Likewise.
11709 (expand_expr_addr_expr_1): Likewise.
11710 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
11711 (expand_expr_real_1): Likewise.
11712 (try_casesi): Likewise.
11713 * final.c (shorten_branches): Likewise.
11714 * fold-const.c (fold_convert_const_int_from_fixed): Change the
11715 type of "mode" to machine_mode.
11716 * internal-fn.c (expand_arith_overflow_result_store): Take a
11717 scalar_int_mode rather than a machine_mode.
11718 (expand_mul_overflow): Use scalar_int_mode for local variables.
11719 * loop-doloop.c (doloop_modify): Likewise.
11720 (doloop_optimize): Likewise.
11721 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
11722 than a machine_mode.
11723 (expand_doubleword_shift_condmove): Likewise.
11724 (expand_doubleword_shift): Likewise.
11725 (expand_doubleword_clz): Likewise.
11726 (expand_doubleword_popcount): Likewise.
11727 (expand_doubleword_parity): Likewise.
11728 (expand_absneg_bit): Use scalar_int_mode for local variables.
11729 (prepare_float_lib_cmp): Likewise.
11730 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
11731 rather than a machine_mode.
11732 (convert_memory_address_addr_space): Likewise.
11733 (get_mode_bounds): Likewise.
11734 (get_address_mode): Return a scalar_int_mode rather than a
11736 * rtlanal.c (get_address_mode): Likewise.
11737 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
11738 than a machine_mode.
11739 * targhooks.c (default_mode_rep_extended): Likewise.
11740 (default_valid_pointer_mode): Likewise.
11741 (default_addr_space_valid_pointer_mode): Likewise.
11742 (default_eh_return_filter_mode): Return a scalar_int_mode rather
11743 than a machine_mode.
11744 (default_libgcc_cmp_return_mode): Likewise.
11745 (default_libgcc_shift_count_mode): Likewise.
11746 (default_unwind_word_mode): Likewise.
11747 (default_addr_space_pointer_mode): Likewise.
11748 (default_addr_space_address_mode): Likewise.
11749 * targhooks.h (default_eh_return_filter_mode): Likewise.
11750 (default_libgcc_cmp_return_mode): Likewise.
11751 (default_libgcc_shift_count_mode): Likewise.
11752 (default_unwind_word_mode): Likewise.
11753 (default_addr_space_pointer_mode): Likewise.
11754 (default_addr_space_address_mode): Likewise.
11755 (default_mode_rep_extended): Take a scalar_int_mode rather than
11757 (default_valid_pointer_mode): Likewise.
11758 (default_addr_space_valid_pointer_mode): Likewise.
11759 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
11761 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
11762 rather than a machine_mode.
11763 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
11764 for local variables.
11765 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11766 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
11767 than a machine_mode.
11769 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11770 Alan Hayward <alan.hayward@arm.com>
11771 David Sherwood <david.sherwood@arm.com>
11773 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
11774 the mode argument to scalar_int_mode.
11775 (do_jump_by_parts_zero_rtx): Likewise.
11776 (do_jump_by_parts_equality_rtx): Likewise.
11777 (do_jump_by_parts_greater): Take a mode argument.
11778 (do_jump_by_parts_equality): Likewise.
11779 (do_jump_1): Update calls accordingly.
11781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11782 Alan Hayward <alan.hayward@arm.com>
11783 David Sherwood <david.sherwood@arm.com>
11785 * is-a.h (safe_dyn_cast): New function.
11786 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
11787 (jump_table_for_label): Likewise.
11788 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
11789 instead of an rtx_insn *.
11790 (shorten_branches): Use dyn_cast instead of LABEL_P and
11791 JUMP_TABLE_DATA_P. Use jump_table_for_label and
11792 rtx_jump_table_data::get_data_mode.
11793 (final_scan_insn): Likewise.
11795 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11796 Alan Hayward <alan.hayward@arm.com>
11797 David Sherwood <david.sherwood@arm.com>
11799 * combine.c (try_combine): Use is_a <scalar_int_mode> when
11800 trying to combine a full-register integer set with a subreg
11803 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11804 Alan Hayward <alan.hayward@arm.com>
11805 David Sherwood <david.sherwood@arm.com>
11807 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
11808 that is always either address_mode or pointer_mode.
11810 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11811 Alan Hayward <alan.hayward@arm.com>
11812 David Sherwood <david.sherwood@arm.com>
11814 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
11815 when the two are known to be equal.
11817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11818 Alan Hayward <alan.hayward@arm.com>
11819 David Sherwood <david.sherwood@arm.com>
11821 * simplify-rtx.c (simplify_const_unary_operation): Use
11822 is_a <scalar_int_mode> instead of checking for a nonzero
11823 precision. Forcibly convert op_mode to a scalar_int_mode
11824 in that case. More clearly differentiate the operand and
11825 result modes and use the former when deciding what the value
11826 of a count-bits operation should be. Use is_int_mode instead
11827 of checking for a MODE_INT. Remove redundant check for whether
11828 this mode has a zero precision.
11830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11831 Alan Hayward <alan.hayward@arm.com>
11832 David Sherwood <david.sherwood@arm.com>
11834 * optabs.c (widen_leading): Change the type of the mode argument
11835 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
11836 (widen_bswap): Likewise.
11837 (expand_parity): Likewise.
11838 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
11839 (expand_ffs): Likewise.
11840 (epand_unop): Check for scalar integer modes before calling the
11843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11844 Alan Hayward <alan.hayward@arm.com>
11845 David Sherwood <david.sherwood@arm.com>
11847 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
11849 (expand_expr_real_1): Update call accordingly.
11851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11852 Alan Hayward <alan.hayward@arm.com>
11853 David Sherwood <david.sherwood@arm.com>
11855 * expmed.c (store_bit_field_using_insv): Add op0_mode and
11856 value_mode arguments. Use scalar_int_mode internally.
11857 (store_bit_field_1): Rename the new integer mode from imode
11858 to op0_mode and use it instead of GET_MODE (op0). Update calls
11859 to store_split_bit_field, store_bit_field_using_insv and
11860 store_fixed_bit_field.
11861 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
11862 Use scalar_int_mode internally. Use a bit count rather than a mode
11863 when calculating the largest bit size for get_best_mode.
11864 Update calls to store_split_bit_field and store_fixed_bit_field_1.
11865 (store_fixed_bit_field_1): Add mode and value_mode arguments.
11866 Remove assertion that OP0 has a scalar integer mode.
11867 (store_split_bit_field): Add op0_mode and value_mode arguments.
11868 Update calls to extract_fixed_bit_field.
11869 (extract_bit_field_using_extv): Add an op0_mode argument.
11870 Use scalar_int_mode internally.
11871 (extract_bit_field_1): Rename the new integer mode from imode to
11872 op0_mode and use it instead of GET_MODE (op0). Update calls to
11873 extract_split_bit_field, extract_bit_field_using_extv and
11874 extract_fixed_bit_field.
11875 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
11876 to extract_split_bit_field and extract_fixed_bit_field_1.
11877 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
11878 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
11879 on the target mode.
11880 (extract_split_bit_field): Add an op0_mode argument. Update call
11881 to extract_fixed_bit_field.
11883 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11884 Alan Hayward <alan.hayward@arm.com>
11885 David Sherwood <david.sherwood@arm.com>
11887 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
11888 * explow.c (hard_function_value): Likewise.
11889 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
11890 convert_to_mode call outside the loop.
11891 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
11892 for the mode iterator. Require the mode specified by max_pieces
11894 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
11896 (copy_blkmode_to_reg): Likewise.
11897 (set_storage_via_setmem): Likewise.
11898 * optabs.c (prepare_cmp_insn): Likewise.
11899 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11900 * stor-layout.c (finish_bitfield_representative): Likewise.
11902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11903 Alan Hayward <alan.hayward@arm.com>
11904 David Sherwood <david.sherwood@arm.com>
11906 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
11907 * expr.c (convert_move): Use them.
11908 (convert_modes): Likewise.
11909 (store_expr_with_bounds): Likewise.
11911 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11912 Alan Hayward <alan.hayward@arm.com>
11913 David Sherwood <david.sherwood@arm.com>
11915 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
11916 parameter for the mode of "x". Remove the "known_x", "known_mode"
11917 and "known_ret" arguments. Change the type of the mode argument
11918 to scalar_int_mode.
11919 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
11920 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
11921 (reg_num_sign_bit_copies_for_combine): Likewise.
11922 * rtlanal.c (nonzero_bits1): Likewise.
11923 (num_sign_bit_copies1): Likewise.
11924 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
11925 (reg_num_sign_bit_copies_general): Likewise.
11926 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
11927 (reg_nonzero_bits_general): Likewise.
11929 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11930 Alan Hayward <alan.hayward@arm.com>
11931 David Sherwood <david.sherwood@arm.com>
11933 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
11934 than in subroutines. Return 1 for non-integer modes.
11935 (cached_num_sign_bit_copies): Change the type of the mode parameter
11936 to scalar_int_mode.
11937 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
11938 classes. Handle CONST_INT_P first and then check whether X also
11939 has a scalar integer mode. Check the same thing for inner registers
11940 of a SUBREG and for values that are being extended or truncated.
11942 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11943 Alan Hayward <alan.hayward@arm.com>
11944 David Sherwood <david.sherwood@arm.com>
11946 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
11947 in subroutines. Return the mode mask for non-integer modes.
11948 (cached_nonzero_bits): Change the type of the mode parameter
11949 to scalar_int_mode.
11950 (nonzero_bits1): Likewise. Remove early exit for other mode
11951 classes. Handle CONST_INT_P first and then check whether X
11952 also has a scalar integer mode.
11954 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11955 Alan Hayward <alan.hayward@arm.com>
11956 David Sherwood <david.sherwood@arm.com>
11958 * expr.c (widest_int_mode_for_size): Make the comment match the code.
11959 Return a scalar_int_mode and assert that the size is greater than
11961 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
11962 (op_by_pieces_d::op_by_pieces_d): Likewise.
11963 (op_by_pieces_d::run): Likewise.
11964 (can_store_by_pieces): Likewise.
11966 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11967 Alan Hayward <alan.hayward@arm.com>
11968 David Sherwood <david.sherwood@arm.com>
11970 * combine.c (extract_left_shift): Add a mode argument and update
11972 (make_compound_operation_int): Change the type of the mode parameter
11973 to scalar_int_mode and update the call to extract_left_shift.
11975 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11976 Alan Hayward <alan.hayward@arm.com>
11977 David Sherwood <david.sherwood@arm.com>
11979 * combine.c (simplify_and_const_int): Change the type of the mode
11980 parameter to scalar_int_mode.
11981 (simplify_and_const_int_1): Likewise. Update recursive call.
11983 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11984 Alan Hayward <alan.hayward@arm.com>
11985 David Sherwood <david.sherwood@arm.com>
11987 * combine.c (simplify_compare_const): Check that the mode is a
11988 scalar_int_mode (rather than VOIDmode) before testing its
11990 (simplify_comparison): Move COMPARISON_P handling out of the
11991 loop and restrict the latter part of the loop to scalar_int_modes.
11992 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
11993 and when considering SUBREG_REGs. Use is_int_mode instead of
11994 checking GET_MODE_CLASS against MODE_INT.
11996 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11997 Alan Hayward <alan.hayward@arm.com>
11998 David Sherwood <david.sherwood@arm.com>
12000 * combine.c (try_widen_shift_mode): Move check for equal modes to...
12001 (simplify_shift_const_1): ...here. Use scalar_int_mode for
12002 shift_unit_mode and for modes involved in scalar shifts.
12004 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12005 Alan Hayward <alan.hayward@arm.com>
12006 David Sherwood <david.sherwood@arm.com>
12008 * combine.c (force_int_to_mode): New function, split out from...
12009 (force_to_mode): ...here. Keep xmode up-to-date and use it
12010 instead of GET_MODE (x).
12012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12013 Alan Hayward <alan.hayward@arm.com>
12014 David Sherwood <david.sherwood@arm.com>
12016 * optabs-query.h (extraction_insn::struct_mode): Change type to
12017 opt_scalar_int_mode and update comment.
12018 (extraction_insn::field_mode): Change type to scalar_int_mode.
12019 (extraction_insn::pos_mode): Likewise.
12020 * combine.c (make_extraction): Update accordingly.
12021 * optabs-query.c (get_traditional_extraction_insn): Likewise.
12022 (get_optab_extraction_insn): Likewise.
12023 * recog.c (simplify_while_replacing): Likewise.
12024 * expmed.c (narrow_bit_field_mem): Change the type of the mode
12025 parameter to opt_scalar_int_mode.
12027 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12028 Alan Hayward <alan.hayward@arm.com>
12029 David Sherwood <david.sherwood@arm.com>
12031 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
12032 to a scalar_int_mode instead of a machine_mode.
12033 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
12034 (get_best_mode): Return a boolean and use a pointer argument to store
12035 the selected mode. Replace the limit mode parameter with a bit limit.
12036 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
12037 for the values returned by bit_field_mode_iterator::next_mode.
12038 (store_bit_field): Update call to get_best_mode.
12039 (store_fixed_bit_field): Likewise.
12040 (extract_fixed_bit_field): Likewise.
12041 * expr.c (optimize_bitfield_assignment_op): Likewise.
12042 * fold-const.c (optimize_bit_field_compare): Likewise.
12043 (fold_truth_andor_1): Likewise.
12044 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
12045 Update for new type of m_mode.
12046 (get_best_mode): As above.
12048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12049 Alan Hayward <alan.hayward@arm.com>
12050 David Sherwood <david.sherwood@arm.com>
12052 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
12053 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
12054 (store_bit_field): Check is_a <scalar_int_mode> before calling
12055 strict_volatile_bitfield_p.
12056 (extract_bit_field): Likewise.
12058 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12059 Alan Hayward <alan.hayward@arm.com>
12060 David Sherwood <david.sherwood@arm.com>
12062 * target.def (cstore_mode): Return a scalar_int_mode.
12063 * doc/tm.texi: Regenerate.
12064 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
12065 * targhooks.h (default_cstore_mode): Likewise.
12066 * targhooks.c (default_cstore_mode): Likewise, using a forced
12068 * expmed.c (emit_cstore): Expect the target of the cstore to be
12071 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12072 Alan Hayward <alan.hayward@arm.com>
12073 David Sherwood <david.sherwood@arm.com>
12075 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
12077 (niter_desc): Likewise mode.
12078 (iv_analyze): Add a mode parameter.
12080 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
12081 and change its type to scalar_int_mode.
12082 * loop-iv.c: Update commentary at head of file.
12083 (iv_constant): Pass the mode paraeter before the rtx it describes
12084 and change its type to scalar_int_mode. Remove VOIDmode handling.
12085 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
12086 (iv_extend): Likewise.
12087 (shorten_into_mode): Likewise.
12088 (iv_add): Use scalar_int_mode.
12089 (iv_mult): Likewise.
12090 (iv_shift): Likewise.
12091 (canonicalize_iv_subregs): Likewise.
12092 (get_biv_step_1): Pass the outer_mode parameter before the rtx
12093 it describes and change its mode to scalar_int_mode. Also change
12094 the type of the returned inner_mode to scalar_int_mode.
12095 (get_biv_step): Likewise, turning outer_mode from a pointer
12096 into a direct parameter. Update call to get_biv_step_1.
12097 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
12098 iv_constant and get_biv_step.
12099 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
12100 and change its type to scalar_int_mode. Don't initialise iv->mode
12101 to VOIDmode and remove later checks for its still being VOIDmode.
12102 Update calls to iv_analyze_op and iv_analyze_expr. Check
12103 is_a <scalar_int_mode> when changing the mode under consideration.
12104 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
12105 Update call to iv_analyze_expr.
12106 (iv_analyze_op): Add a mode parameter. Reject subregs whose
12107 inner register is not also a scalar_int_mode. Update call to
12109 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
12110 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
12111 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
12112 separate mode class checks. Update calls to iv_analyze. Remove
12113 fix-up of VOIDmodes after iv_analyze_biv.
12114 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
12115 don't have a scalar_int_mode. Update call to biv_p.
12117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12118 Alan Hayward <alan.hayward@arm.com>
12119 David Sherwood <david.sherwood@arm.com>
12121 * cfgexpand.c (convert_debug_memory_address): Use
12122 as_a <scalar_int_mode>.
12123 * combine.c (expand_compound_operation): Likewise.
12124 (make_extraction): Likewise.
12125 (change_zero_ext): Likewise.
12126 (simplify_comparison): Likewise.
12127 * cse.c (cse_insn): Likewise.
12128 * dwarf2out.c (minmax_loc_descriptor): Likewise.
12129 (mem_loc_descriptor): Likewise.
12130 (loc_descriptor): Likewise.
12131 * expmed.c (init_expmed_one_mode): Likewise.
12132 (synth_mult): Likewise.
12133 (emit_store_flag_1): Likewise.
12134 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
12135 of a comparison with size.
12136 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
12137 (reduce_to_bit_field_precision): Likewise.
12138 * function.c (expand_function_end): Likewise.
12139 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
12140 * loop-doloop.c (doloop_modify): Likewise.
12141 * optabs.c (expand_binop): Likewise.
12142 (expand_unop): Likewise.
12143 (expand_copysign_absneg): Likewise.
12144 (prepare_cmp_insn): Likewise.
12145 (maybe_legitimize_operand): Likewise.
12146 * recog.c (const_scalar_int_operand): Likewise.
12147 * rtlanal.c (get_address_mode): Likewise.
12148 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12149 (simplify_cond_clz_ctz): Likewise.
12150 * tree-nested.c (get_nl_goto_field): Likewise.
12151 * tree.c (build_vector_type_for_mode): Likewise.
12152 * var-tracking.c (use_narrower_mode): Likewise.
12154 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12155 Alan Hayward <alan.hayward@arm.com>
12156 David Sherwood <david.sherwood@arm.com>
12158 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
12159 * builtins.c (expand_builtin_signbit): Use it.
12160 * cfgexpand.c (expand_debug_expr): Likewise.
12161 * dojump.c (do_jump): Likewise.
12162 (do_compare_and_jump): Likewise.
12163 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
12164 * expmed.c (make_tree): Likewise.
12165 * expr.c (expand_expr_real_2): Likewise.
12166 (expand_expr_real_1): Likewise.
12167 (try_casesi): Likewise.
12168 * fold-const-call.c (fold_const_call_ss): Likewise.
12169 * fold-const.c (unextend): Likewise.
12170 (extract_muldiv_1): Likewise.
12171 (fold_single_bit_test): Likewise.
12172 (native_encode_int): Likewise.
12173 (native_encode_string): Likewise.
12174 (native_interpret_int): Likewise.
12175 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
12176 * internal-fn.c (expand_addsub_overflow): Likewise.
12177 (expand_neg_overflow): Likewise.
12178 (expand_mul_overflow): Likewise.
12179 (expand_arith_overflow): Likewise.
12180 * match.pd: Likewise.
12181 * stor-layout.c (layout_type): Likewise.
12182 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12183 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
12184 * tree-ssanames.c (get_range_info): Likewise.
12185 * tree-switch-conversion.c (array_value_type) Likewise.
12186 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
12187 (vect_recog_divmod_pattern): Likewise.
12188 (vect_recog_mixed_size_cond_pattern): Likewise.
12189 * tree-vrp.c (extract_range_basic): Likewise.
12190 (simplify_float_conversion_using_ranges): Likewise.
12191 * tree.c (int_fits_type_p): Likewise.
12192 * ubsan.c (instrument_bool_enum_load): Likewise.
12193 * varasm.c (mergeable_string_section): Likewise.
12194 (narrowing_initializer_constant_valid_p): Likewise.
12195 (output_constant): Likewise.
12197 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12198 Alan Hayward <alan.hayward@arm.com>
12199 David Sherwood <david.sherwood@arm.com>
12201 * machmode.h (NARROWEST_INT_MODE): New macro.
12202 * expr.c (alignment_for_piecewise_move): Use it instead of
12203 GET_CLASS_NARROWEST_MODE (MODE_INT).
12204 (push_block): Likewise.
12205 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
12207 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12209 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12210 Alan Hayward <alan.hayward@arm.com>
12211 David Sherwood <david.sherwood@arm.com>
12213 * postreload.c (move2add_valid_value_p): Change the type of the
12214 mode parameter to scalar_int_mode.
12215 (move2add_use_add2_insn): Add a mode parameter and use it instead
12217 (move2add_use_add3_insn): Likewise.
12218 (reload_cse_move2add): Update accordingly.
12220 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12221 Alan Hayward <alan.hayward@arm.com>
12222 David Sherwood <david.sherwood@arm.com>
12224 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
12226 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
12227 * optabs.c (expand_unop): Likewise.
12229 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12230 Alan Hayward <alan.hayward@arm.com>
12231 David Sherwood <david.sherwood@arm.com>
12233 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
12234 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
12235 (popcount_loc_descriptor): Likewise.
12236 (bswap_loc_descriptor): Likewise.
12237 (rotate_loc_descriptor): Likewise.
12238 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
12239 calling the functions above.
12241 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12242 Alan Hayward <alan.hayward@arm.com>
12243 David Sherwood <david.sherwood@arm.com>
12245 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
12247 (try_combine): Likewise.
12248 (simplify_if_then_else): Likewise.
12249 * cse.c (cse_insn): Likewise.
12250 * dwarf2out.c (mem_loc_descriptor): Likewise.
12251 * emit-rtl.c (gen_lowpart_common): Likewise.
12252 * simplify-rtx.c (simplify_truncation): Likewise.
12253 (simplify_binary_operation_1): Likewise.
12254 (simplify_const_relational_operation): Likewise.
12255 (simplify_ternary_operation): Likewise.
12256 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
12258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12259 Alan Hayward <alan.hayward@arm.com>
12260 David Sherwood <david.sherwood@arm.com>
12262 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
12263 * reload.c (push_reload): Likewise.
12264 (find_reloads): Likewise.
12266 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12267 Alan Hayward <alan.hayward@arm.com>
12268 David Sherwood <david.sherwood@arm.com>
12270 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
12271 (make_compound_operation_int): Likewise.
12272 (change_zero_ext): Likewise.
12273 * expr.c (convert_move): Likewise.
12274 (convert_modes): Likewise.
12275 * fwprop.c (forward_propagate_subreg): Likewise.
12276 * loop-iv.c (get_biv_step_1): Likewise.
12277 * optabs.c (widen_operand): Likewise.
12278 * postreload.c (move2add_valid_value_p): Likewise.
12279 * recog.c (simplify_while_replacing): Likewise.
12280 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12281 (simplify_binary_operation_1): Likewise. Remove redundant
12282 mode equality check.
12284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12285 Alan Hayward <alan.hayward@arm.com>
12286 David Sherwood <david.sherwood@arm.com>
12288 * combine.c (combine_simplify_rtx): Add checks for
12289 is_a <scalar_int_mode>.
12290 (simplify_if_then_else): Likewise.
12291 (make_field_assignment): Likewise.
12292 (simplify_comparison): Likewise.
12293 * ifcvt.c (noce_try_bitop): Likewise.
12294 * loop-invariant.c (canonicalize_address_mult): Likewise.
12295 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12297 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12298 Alan Hayward <alan.hayward@arm.com>
12299 David Sherwood <david.sherwood@arm.com>
12301 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
12302 is_a <scalar_int_mode> instead of != BLKmode.
12304 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12305 Alan Hayward <alan.hayward@arm.com>
12306 David Sherwood <david.sherwood@arm.com>
12308 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
12309 instead of != VOIDmode.
12310 * combine.c (if_then_else_cond): Likewise.
12311 (change_zero_ext): Likewise.
12312 * dwarf2out.c (mem_loc_descriptor): Likewise.
12313 (loc_descriptor): Likewise.
12314 * rtlanal.c (canonicalize_condition): Likewise.
12315 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
12317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12318 Alan Hayward <alan.hayward@arm.com>
12319 David Sherwood <david.sherwood@arm.com>
12321 * simplify-rtx.c (simplify_binary_operation_1): Use
12322 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
12324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12325 Alan Hayward <alan.hayward@arm.com>
12326 David Sherwood <david.sherwood@arm.com>
12328 * wide-int.h (int_traits<unsigned char>) New class.
12329 (int_traits<unsigned short>) Likewise.
12330 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
12331 Use GET_MODE_UNIT_PRECISION and remove redundant test for
12333 * combine.c (set_nonzero_bits_and_sign_copies): Use
12334 is_a <scalar_int_mode>.
12335 (find_split_point): Likewise.
12336 (combine_simplify_rtx): Likewise.
12337 (simplify_logical): Likewise.
12338 (expand_compound_operation): Likewise.
12339 (expand_field_assignment): Likewise.
12340 (make_compound_operation): Likewise.
12341 (extended_count): Likewise.
12342 (change_zero_ext): Likewise.
12343 (simplify_comparison): Likewise.
12344 * dwarf2out.c (scompare_loc_descriptor): Likewise.
12345 (ucompare_loc_descriptor): Likewise.
12346 (minmax_loc_descriptor): Likewise.
12347 (mem_loc_descriptor): Likewise.
12348 (loc_descriptor): Likewise.
12349 * expmed.c (init_expmed_one_mode): Likewise.
12350 * lra-constraints.c (lra_constraint_offset): Likewise.
12351 * optabs.c (prepare_libcall_arg): Likewise.
12352 * postreload.c (move2add_note_store): Likewise.
12353 * reload.c (operands_match_p): Likewise.
12354 * rtl.h (load_extend_op): Likewise.
12355 * rtlhooks.c (gen_lowpart_general): Likewise.
12356 * simplify-rtx.c (simplify_truncation): Likewise.
12357 (simplify_unary_operation_1): Likewise.
12358 (simplify_binary_operation_1): Likewise.
12359 (simplify_const_binary_operation): Likewise.
12360 (simplify_const_relational_operation): Likewise.
12361 (simplify_subreg): Likewise.
12362 * stor-layout.c (bitwise_mode_for_mode): Likewise.
12363 * var-tracking.c (adjust_mems): Likewise.
12364 (prepare_call_arguments): Likewise.
12366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12367 Alan Hayward <alan.hayward@arm.com>
12368 David Sherwood <david.sherwood@arm.com>
12370 * machmode.h (is_int_mode): New fuction.
12371 * combine.c (find_split_point): Use it.
12372 (combine_simplify_rtx): Likewise.
12373 (simplify_if_then_else): Likewise.
12374 (simplify_set): Likewise.
12375 (simplify_shift_const_1): Likewise.
12376 (simplify_comparison): Likewise.
12377 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
12378 * cse.c (notreg_cost): Likewise.
12379 (cse_insn): Likewise.
12380 * cselib.c (cselib_lookup_1): Likewise.
12381 * dojump.c (do_jump_1): Likewise.
12382 (do_compare_rtx_and_jump): Likewise.
12383 * dse.c (get_call_args): Likewise.
12384 * dwarf2out.c (rtl_for_decl_init): Likewise.
12385 (native_encode_initializer): Likewise.
12386 * expmed.c (emit_store_flag_1): Likewise.
12387 (emit_store_flag): Likewise.
12388 * expr.c (convert_modes): Likewise.
12389 (store_field): Likewise.
12390 (expand_expr_real_1): Likewise.
12391 * fold-const.c (fold_read_from_constant_string): Likewise.
12392 * gimple-ssa-sprintf.c (get_format_string): Likewise.
12393 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
12394 * optabs.c (expand_binop): Likewise.
12395 (expand_unop): Likewise.
12396 (expand_abs_nojump): Likewise.
12397 (expand_one_cmpl_abs_nojump): Likewise.
12398 * simplify-rtx.c (mode_signbit_p): Likewise.
12399 (val_signbit_p): Likewise.
12400 (val_signbit_known_set_p): Likewise.
12401 (val_signbit_known_clear_p): Likewise.
12402 (simplify_relational_operation_1): Likewise.
12403 * tree.c (vector_type_mode): Likewise.
12405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12406 Alan Hayward <alan.hayward@arm.com>
12407 David Sherwood <david.sherwood@arm.com>
12409 * machmode.h (smallest_mode_for_size): Fix formatting.
12410 (smallest_int_mode_for_size): New function.
12411 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
12412 instead of smallest_mode_for_size.
12413 * combine.c (make_extraction): Likewise.
12414 * config/arc/arc.c (arc_expand_movmem): Likewise.
12415 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
12416 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
12417 * config/s390/s390.c (s390_expand_insv): Likewise.
12418 * config/sparc/sparc.c (assign_int_registers): Likewise.
12419 * config/spu/spu.c (spu_function_value): Likewise.
12420 (spu_function_arg): Likewise.
12421 * coverage.c (get_gcov_type): Likewise.
12422 (get_gcov_unsigned_t): Likewise.
12423 * dse.c (find_shift_sequence): Likewise.
12424 * expmed.c (store_bit_field_1): Likewise.
12425 * expr.c (convert_move): Likewise.
12426 (store_field): Likewise.
12427 * internal-fn.c (expand_arith_overflow): Likewise.
12428 * optabs-query.c (get_best_extraction_insn): Likewise.
12429 * optabs.c (expand_twoval_binop_libfunc): Likewise.
12430 * stor-layout.c (layout_type): Likewise.
12431 (initialize_sizetypes): Likewise.
12432 * targhooks.c (default_get_mask_mode): Likewise.
12433 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
12435 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12436 Alan Hayward <alan.hayward@arm.com>
12437 David Sherwood <david.sherwood@arm.com>
12439 * machmode.h (opt_mode::else_blk): New function.
12440 (int_mode_for_mode): Declare.
12441 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
12442 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
12444 * cfgexpand.c (expand_debug_expr): Likewise.
12445 * combine.c (gen_lowpart_or_truncate): Likewise.
12446 (gen_lowpart_for_combine): Likewise.
12447 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
12448 * config/avr/avr.c (avr_to_int_mode): Likewise.
12449 (avr_out_plus_1): Likewise.
12450 (avr_out_plus): Likewise.
12451 (avr_out_round): Likewise.
12452 * config/i386/i386.c (ix86_split_to_parts): Likewise.
12453 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
12454 (s390_expand_vcond): Likewise.
12455 * config/spu/spu.c (spu_split_immediate): Likewise.
12456 (spu_expand_mov): Likewise.
12457 * dse.c (get_stored_val): Likewise.
12458 * expmed.c (store_bit_field_1): Likewise.
12459 (convert_extracted_bit_field): Use int_mode_for_mode instead of
12461 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
12462 (extract_low_bits): Likewise.
12463 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
12464 handling rather than repeating the check.
12465 (emit_group_store): Likewise.
12466 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
12467 * optabs.c (expand_absneg_bit): Likewise.
12468 (expand_copysign_absneg): Likewise.
12469 (expand_copysign_bit): Likewise.
12470 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
12471 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
12472 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
12473 * var-tracking.c (prepare_call_arguments): Likewise.
12474 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
12475 int_mode_for_mode instead of mode_for_size.
12476 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
12478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12479 Alan Hayward <alan.hayward@arm.com>
12480 David Sherwood <david.sherwood@arm.com>
12482 * machmode.h (int_mode_for_size): New function.
12483 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
12484 instead of mode_for_size.
12485 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
12487 * combine.c (expand_field_assignment): Use int_mode_for_size
12488 instead of mode_for_size.
12489 (make_extraction): Likewise.
12490 (simplify_shift_const_1): Likewise.
12491 (simplify_comparison): Likewise.
12492 * dojump.c (do_jump): Likewise.
12493 * dwarf2out.c (mem_loc_descriptor): Likewise.
12494 * emit-rtl.c (init_derived_machine_modes): Likewise.
12495 * expmed.c (flip_storage_order): Likewise.
12496 (convert_extracted_bit_field): Likewise.
12497 * expr.c (copy_blkmode_from_reg): Likewise.
12498 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
12499 * internal-fn.c (expand_mul_overflow): Likewise.
12500 * lower-subreg.c (simple_move): Likewise.
12501 * optabs-libfuncs.c (init_optabs): Likewise.
12502 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12503 * tree.c (vector_type_mode): Likewise.
12504 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
12505 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
12506 * tree-vect-generic.c (expand_vector_parallel): Likewise.
12507 * tree-vect-stmts.c (vectorizable_load): Likewise.
12508 (vectorizable_store): Likewise.
12510 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12511 Alan Hayward <alan.hayward@arm.com>
12512 David Sherwood <david.sherwood@arm.com>
12514 * coretypes.h (pod_mode): New type.
12515 (scalar_int_mode_pod): New typedef.
12516 * machmode.h (pod_mode): New class.
12517 (int_n_data_t::m): Change type to scalar_int_mode_pod.
12518 * genmodes.c (emit_mode_int_n): Update accordingly.
12519 * lower-subreg.h (target_lower_subreg): Change type to
12520 scalar_int_mode_pod.
12521 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
12522 scalar_int_mode_pod.
12524 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12525 Alan Hayward <alan.hayward@arm.com>
12526 David Sherwood <david.sherwood@arm.com>
12528 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
12529 machine_mode to scalar_int_mode.
12530 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
12531 (rs6000_option_override_internal): Remove cast to int.
12532 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
12533 machine_mode to scalar_int_mode.
12534 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
12535 (rs6000_option_override_internal): Remove cast to int.
12536 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
12537 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
12539 * config/s390/s390.c (s390_expand_builtin): Likewise.
12540 * coretypes.h (scalar_int_mode): New type.
12541 (opt_scalar_int_mode): New typedef.
12542 * machmode.h (scalar_int_mode): New class.
12543 (scalar_int_mode::includes_p): New function.
12544 (byte_mode): Change type to scalar_int_mode.
12545 (word_mode): Likewise.
12546 (ptr_mode): Likewise.
12547 * emit-rtl.c (byte_mode): Likewise.
12548 (word_mode): Likewise.
12549 (ptr_mode): Likewise.
12550 (init_derived_machine_modes): Update accordingly.
12551 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
12552 and MODE_PARTIAL_INT.
12553 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
12554 opt_scalar_int_mode.
12556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12557 Alan Hayward <alan.hayward@arm.com>
12558 David Sherwood <david.sherwood@arm.com>
12560 * target.def (libgcc_floating_mode_supported_p): Take a
12562 * doc/tm.texi: Regenerate.
12563 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
12565 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12566 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
12569 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12570 Alan Hayward <alan.hayward@arm.com>
12571 David Sherwood <david.sherwood@arm.com>
12573 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
12574 * doc/tm.texi: Regenerate.
12575 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
12576 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
12577 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
12578 * targhooks.h (default_floatn_mode): Likewise.
12579 * targhooks.c (default_floatn_mode): Likewise.
12580 * tree.c (build_common_tree_nodes): Update accordingly.
12582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12583 Alan Hayward <alan.hayward@arm.com>
12584 David Sherwood <david.sherwood@arm.com>
12586 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
12587 (mode_iterator::iterate_p): Likewise.
12588 (mode_iterator::get_wider): Likewise.
12589 * expr.c (init_expr_target): Use opt_scalar_float_mode.
12591 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12592 Alan Hayward <alan.hayward@arm.com>
12593 David Sherwood <david.sherwood@arm.com>
12595 * coretypes.h (opt_scalar_float_mode): New typedef.
12596 * machmode.h (float_mode_for_size): New function.
12597 * emit-rtl.c (double_mode): Delete.
12598 (init_emit_once): Use float_mode_for_size.
12599 * stor-layout.c (layout_type): Likewise.
12600 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
12602 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12603 Alan Hayward <alan.hayward@arm.com>
12604 David Sherwood <david.sherwood@arm.com>
12606 * output.h (assemble_real): Take a scalar_float_mode.
12607 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
12608 * config/arm/arm.md (consttable_4): Likewise.
12609 (consttable_8): Likewise.
12610 (consttable_16): Likewise.
12611 * config/mips/mips.md (consttable_float): Likewise.
12612 * config/s390/s390.c (s390_output_pool_entry): Likewise.
12613 * varasm.c (assemble_real): Take a scalar_float_mode.
12614 (output_constant_pool_2): Update accordingly.
12615 (output_constant): Likewise.
12617 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12618 Alan Hayward <alan.hayward@arm.com>
12619 David Sherwood <david.sherwood@arm.com>
12621 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
12622 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
12623 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
12624 (native_encode_real): Likewise.
12625 (native_interpret_real): Likewise.
12626 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
12627 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12630 Alan Hayward <alan.hayward@arm.com>
12631 David Sherwood <david.sherwood@arm.com>
12633 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
12634 <scalar_float_mode>. Simplify.
12635 (gen_extend_conv_libfunc): Likewise.
12637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12638 Alan Hayward <alan.hayward@arm.com>
12639 David Sherwood <david.sherwood@arm.com>
12641 * coretypes.h (scalar_float_mode): New type.
12642 * machmode.h (mode_traits::from_int): Use machine_mode if
12643 USE_ENUM_MODES is defined.
12644 (is_a): New function.
12646 (dyn_cast): Likewise.
12647 (scalar_float_mode): New class.
12648 (scalar_float_mode::includes_p): New function.
12649 (is_float_mode): Likewise.
12650 * gdbhooks.py (MachineModePrinter): New class.
12651 (build_pretty_printer): Use it for scalar_float_mode.
12652 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
12653 (format_helper::format_helper): Turn into a template.
12654 * genmodes.c (get_mode_class): New function.
12655 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
12656 or machine_mode if none.
12657 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
12658 as_a <scalar_float_mode>.
12659 * dwarf2out.c (mem_loc_descriptor): Likewise.
12660 (insert_float): Likewise.
12661 (add_const_value_attribute): Likewise.
12662 * simplify-rtx.c (simplify_immed_subreg): Likewise.
12663 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
12664 (expand_unop): Update accordingly.
12665 (expand_abs_nojump): Likewise.
12666 (expand_copysign_absneg): Take a scalar_float_mode.
12667 (expand_copysign_bit): Likewise.
12668 (expand_copysign): Update accordingly.
12670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12671 Alan Hayward <alan.hayward@arm.com>
12672 David Sherwood <david.sherwood@arm.com>
12674 * coretypes.h (opt_mode): New class.
12675 * machmode.h (opt_mode): Likewise.
12676 (opt_mode::else_void): New function.
12677 (opt_mode::require): Likewise.
12678 (opt_mode::exists): Likewise.
12679 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
12680 (GET_MODE_2XWIDER_MODE): Likewise.
12681 (mode_iterator::get_wider): Update accordingly.
12682 (mode_iterator::get_2xwider): Likewise.
12683 (mode_iterator::get_known_wider): Likewise, turning into a template.
12684 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
12685 forcing a wider mode to exist.
12686 * config/cr16/cr16.h (LONG_REG_P): Likewise.
12687 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12688 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
12689 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
12690 * lower-subreg.c (init_lower_subreg): Likewise.
12691 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
12692 on the final iteration.
12693 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
12694 a wider mode exists before asking for a move pattern.
12695 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
12696 forcing a wider mode to exist.
12697 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
12698 returning false if no such mode exists.
12699 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
12700 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
12701 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
12702 Avoid checking for a MODE_INT if we already know the mode is not a
12704 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
12705 forcing a wider mode to exist.
12706 (expmed_mult_highpart_optab): Likewise.
12707 (expmed_mult_highpart): Likewise.
12708 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
12710 * lto-streamer-in.c (lto_input_mode_table): Likewise.
12711 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
12712 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
12713 * internal-fn.c (expand_mul_overflow): Update use of
12714 GET_MODE_2XWIDER_MODE.
12715 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12716 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
12717 GET_MODE_WIDER_MODE.
12718 (convert_plusminus_to_widen): Likewise.
12719 * tree-switch-conversion.c (array_value_type): Likewise.
12720 * var-tracking.c (emit_note_insn_var_location): Likewise.
12721 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
12722 Return false inside rather than outside the loop if no wider mode
12724 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
12725 and GET_MODE_2XWIDER_MODE
12726 (can_compare_p): Use else_void.
12727 * gdbhooks.py (OptMachineModePrinter): New class.
12728 (build_pretty_printer): Use it for opt_mode.
12730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12731 Alan Hayward <alan.hayward@arm.com>
12732 David Sherwood <david.sherwood@arm.com>
12734 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
12735 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
12737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12738 Alan Hayward <alan.hayward@arm.com>
12739 David Sherwood <david.sherwood@arm.com>
12741 * machmode.h (mode_traits): New structure.
12742 (get_narrowest_mode): New function.
12743 (mode_iterator::start): Likewise.
12744 (mode_iterator::iterate_p): Likewise.
12745 (mode_iterator::get_wider): Likewise.
12746 (mode_iterator::get_known_wider): Likewise.
12747 (mode_iterator::get_2xwider): Likewise.
12748 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
12749 (FOR_EACH_MODE): Likewise.
12750 (FOR_EACH_MODE_FROM): Likewise.
12751 (FOR_EACH_MODE_UNTIL): Likewise.
12752 (FOR_EACH_WIDER_MODE): Likewise.
12753 (FOR_EACH_2XWIDER_MODE): Likewise.
12754 * builtins.c (expand_builtin_strlen): Use new mode iterators.
12755 * combine.c (simplify_comparison): Likewise
12756 * config/i386/i386.c (type_natural_mode): Likewise.
12757 * cse.c (cse_insn): Likewise.
12758 * dse.c (find_shift_sequence): Likewise.
12759 * emit-rtl.c (init_derived_machine_modes): Likewise.
12760 (init_emit_once): Likewise.
12761 * explow.c (hard_function_value): Likewise.
12762 * expmed.c (extract_fixed_bit_field_1): Likewise.
12763 (extract_bit_field_1): Likewise.
12764 (expand_divmod): Likewise.
12765 (emit_store_flag_1): Likewise.
12766 * expr.c (init_expr_target): Likewise.
12767 (convert_move): Likewise.
12768 (alignment_for_piecewise_move): Likewise.
12769 (widest_int_mode_for_size): Likewise.
12770 (emit_block_move_via_movmem): Likewise.
12771 (copy_blkmode_to_reg): Likewise.
12772 (set_storage_via_setmem): Likewise.
12773 (compress_float_constant): Likewise.
12774 * omp-low.c (omp_clause_aligned_alignment): Likewise.
12775 * optabs-query.c (get_best_extraction_insn): Likewise.
12776 * optabs.c (expand_binop): Likewise.
12777 (expand_twoval_unop): Likewise.
12778 (expand_twoval_binop): Likewise.
12779 (widen_leading): Likewise.
12780 (widen_bswap): Likewise.
12781 (expand_parity): Likewise.
12782 (expand_unop): Likewise.
12783 (prepare_cmp_insn): Likewise.
12784 (prepare_float_lib_cmp): Likewise.
12785 (expand_float): Likewise.
12786 (expand_fix): Likewise.
12787 (expand_sfix_optab): Likewise.
12788 * postreload.c (move2add_use_add2_insn): Likewise.
12789 * reg-stack.c (reg_to_stack): Likewise.
12790 * reginfo.c (choose_hard_reg_mode): Likewise.
12791 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
12792 * stor-layout.c (mode_for_size): Likewise.
12793 (smallest_mode_for_size): Likewise.
12794 (mode_for_vector): Likewise.
12795 (finish_bitfield_representative): Likewise.
12796 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
12797 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
12798 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12799 * var-tracking.c (prepare_call_arguments): Likewise.
12801 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12802 Alan Hayward <alan.hayward@arm.com>
12803 David Sherwood <david.sherwood@arm.com>
12805 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
12806 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
12807 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
12808 * machmode.h (mode_size): Move earlier in file.
12809 (mode_precision): Likewise.
12810 (mode_inner): Likewise.
12811 (mode_nunits): Likewise.
12812 (mode_unit_size): Likewise.
12813 (unit_unit_precision): Likewise.
12814 (mode_wider): Likewise.
12815 (mode_2xwider): Likewise.
12816 (machine_mode): New class.
12817 (mode_to_bytes): New function.
12818 (mode_to_bits): Likewise.
12819 (mode_to_precision): Likewise.
12820 (mode_to_inner): Likewise.
12821 (mode_to_unit_size): Likewise.
12822 (mode_to_unit_precision): Likewise.
12823 (mode_to_nunits): Likewise.
12824 (GET_MODE_SIZE): Use mode_to_bytes.
12825 (GET_MODE_BITSIZE): Use mode_to_bits.
12826 (GET_MODE_PRECISION): Use mode_to_precision.
12827 (GET_MODE_INNER): Use mode_to_inner.
12828 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
12829 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
12830 (GET_MODE_NUNITS): Use mode_to_nunits.
12831 * system.h (ALWAYS_INLINE): New macro.
12832 * config/powerpcspe/powerpcspe-c.c
12833 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
12834 int for arg1_mode and arg2_mode.
12836 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12837 Alan Hayward <alan.hayward@arm.com>
12838 David Sherwood <david.sherwood@arm.com>
12840 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
12841 Prefix mode names with E_ in case statements.
12842 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12843 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
12844 (aarch64_split_simd_move): Likewise.
12845 (aarch64_gen_storewb_pair): Likewise.
12846 (aarch64_gen_loadwb_pair): Likewise.
12847 (aarch64_gen_store_pair): Likewise.
12848 (aarch64_gen_load_pair): Likewise.
12849 (aarch64_get_condition_code_1): Likewise.
12850 (aarch64_constant_pool_reload_icode): Likewise.
12851 (get_rsqrte_type): Likewise.
12852 (get_rsqrts_type): Likewise.
12853 (get_recpe_type): Likewise.
12854 (get_recps_type): Likewise.
12855 (aarch64_gimplify_va_arg_expr): Likewise.
12856 (aarch64_simd_container_mode): Likewise.
12857 (aarch64_emit_load_exclusive): Likewise.
12858 (aarch64_emit_store_exclusive): Likewise.
12859 (aarch64_expand_compare_and_swap): Likewise.
12860 (aarch64_gen_atomic_cas): Likewise.
12861 (aarch64_emit_bic): Likewise.
12862 (aarch64_emit_atomic_swap): Likewise.
12863 (aarch64_emit_atomic_load_op): Likewise.
12864 (aarch64_evpc_trn): Likewise.
12865 (aarch64_evpc_uzp): Likewise.
12866 (aarch64_evpc_zip): Likewise.
12867 (aarch64_evpc_ext): Likewise.
12868 (aarch64_evpc_rev): Likewise.
12869 (aarch64_evpc_dup): Likewise.
12870 (aarch64_gen_ccmp_first): Likewise.
12871 (aarch64_gen_ccmp_next): Likewise.
12872 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
12873 (alpha_emit_xfloating_libcall): Likewise.
12874 (emit_insxl): Likewise.
12875 (alpha_arg_type): Likewise.
12876 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
12877 (arc_preferred_simd_mode): Likewise.
12878 (arc_secondary_reload): Likewise.
12879 (get_arc_condition_code): Likewise.
12880 (arc_print_operand): Likewise.
12881 (arc_legitimate_constant_p): Likewise.
12882 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12883 * config/arc/arc.md (casesi_load): Likewise.
12884 (casesi_compact_jump): Likewise.
12885 * config/arc/predicates.md (proper_comparison_operator): Likewise.
12886 (cc_use_register): Likewise.
12887 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12888 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
12889 (arm_init_iwmmxt_builtins): Likewise.
12890 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
12891 (neon_expand_vector_init): Likewise.
12892 (arm_attr_length_move_neon): Likewise.
12893 (maybe_get_arm_condition_code): Likewise.
12894 (arm_emit_vector_const): Likewise.
12895 (arm_preferred_simd_mode): Likewise.
12896 (arm_output_iwmmxt_tinsr): Likewise.
12897 (thumb1_output_casesi): Likewise.
12898 (thumb2_output_casesi): Likewise.
12899 (arm_emit_load_exclusive): Likewise.
12900 (arm_emit_store_exclusive): Likewise.
12901 (arm_expand_compare_and_swap): Likewise.
12902 (arm_evpc_neon_vuzp): Likewise.
12903 (arm_evpc_neon_vzip): Likewise.
12904 (arm_evpc_neon_vrev): Likewise.
12905 (arm_evpc_neon_vtrn): Likewise.
12906 (arm_evpc_neon_vext): Likewise.
12907 (arm_validize_comparison): Likewise.
12908 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
12909 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
12910 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
12911 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
12912 (c6x_preferred_simd_mode): Likewise.
12913 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
12914 (epiphany_rtx_costs): Likewise.
12915 * config/epiphany/predicates.md (proper_comparison_operator):
12917 * config/frv/frv.c (condexec_memory_operand): Likewise.
12918 (frv_emit_move): Likewise.
12919 (output_move_single): Likewise.
12920 (output_condmove_single): Likewise.
12921 (frv_hard_regno_mode_ok): Likewise.
12922 (frv_matching_accg_mode): Likewise.
12923 * config/h8300/h8300.c (split_adds_subs): Likewise.
12924 (h8300_rtx_costs): Likewise.
12925 (h8300_print_operand): Likewise.
12926 (compute_mov_length): Likewise.
12927 (output_logical_op): Likewise.
12928 (compute_logical_op_length): Likewise.
12929 (compute_logical_op_cc): Likewise.
12930 (h8300_shift_needs_scratch_p): Likewise.
12931 (output_a_shift): Likewise.
12932 (compute_a_shift_length): Likewise.
12933 (compute_a_shift_cc): Likewise.
12934 (expand_a_rotate): Likewise.
12935 (output_a_rotate): Likewise.
12936 * config/i386/i386.c (classify_argument): Likewise.
12937 (function_arg_advance_32): Likewise.
12938 (function_arg_32): Likewise.
12939 (function_arg_64): Likewise.
12940 (function_value_64): Likewise.
12941 (ix86_gimplify_va_arg): Likewise.
12942 (ix86_legitimate_constant_p): Likewise.
12943 (put_condition_code): Likewise.
12944 (split_double_mode): Likewise.
12945 (ix86_avx256_split_vector_move_misalign): Likewise.
12946 (ix86_expand_vector_logical_operator): Likewise.
12947 (ix86_split_idivmod): Likewise.
12948 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
12949 (ix86_build_const_vector): Likewise.
12950 (ix86_build_signbit_mask): Likewise.
12951 (ix86_match_ccmode): Likewise.
12952 (ix86_cc_modes_compatible): Likewise.
12953 (ix86_expand_branch): Likewise.
12954 (ix86_expand_sse_cmp): Likewise.
12955 (ix86_expand_sse_movcc): Likewise.
12956 (ix86_expand_int_sse_cmp): Likewise.
12957 (ix86_expand_vec_perm_vpermi2): Likewise.
12958 (ix86_expand_vec_perm): Likewise.
12959 (ix86_expand_sse_unpack): Likewise.
12960 (ix86_expand_int_addcc): Likewise.
12961 (ix86_split_to_parts): Likewise.
12962 (ix86_vectorize_builtin_gather): Likewise.
12963 (ix86_vectorize_builtin_scatter): Likewise.
12964 (avx_vpermilp_parallel): Likewise.
12965 (inline_memory_move_cost): Likewise.
12966 (ix86_tieable_integer_mode_p): Likewise.
12967 (x86_maybe_negate_const_int): Likewise.
12968 (ix86_expand_vector_init_duplicate): Likewise.
12969 (ix86_expand_vector_init_one_nonzero): Likewise.
12970 (ix86_expand_vector_init_one_var): Likewise.
12971 (ix86_expand_vector_init_concat): Likewise.
12972 (ix86_expand_vector_init_interleave): Likewise.
12973 (ix86_expand_vector_init_general): Likewise.
12974 (ix86_expand_vector_set): Likewise.
12975 (ix86_expand_vector_extract): Likewise.
12976 (emit_reduc_half): Likewise.
12977 (ix86_emit_i387_round): Likewise.
12978 (ix86_mangle_type): Likewise.
12979 (ix86_expand_round_sse4): Likewise.
12980 (expand_vec_perm_blend): Likewise.
12981 (canonicalize_vector_int_perm): Likewise.
12982 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
12983 (expand_vec_perm_1): Likewise.
12984 (expand_vec_perm_interleave3): Likewise.
12985 (expand_vec_perm_even_odd_pack): Likewise.
12986 (expand_vec_perm_even_odd_1): Likewise.
12987 (expand_vec_perm_broadcast_1): Likewise.
12988 (ix86_vectorize_vec_perm_const_ok): Likewise.
12989 (ix86_expand_vecop_qihi): Likewise.
12990 (ix86_expand_mul_widen_hilo): Likewise.
12991 (ix86_expand_sse2_abs): Likewise.
12992 (ix86_expand_pextr): Likewise.
12993 (ix86_expand_pinsr): Likewise.
12994 (ix86_preferred_simd_mode): Likewise.
12995 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
12996 * config/i386/sse.md (*andnot<mode>3): Likewise.
12997 (<mask_codefor><code><mode>3<mask_name>): Likewise.
12998 (*<code><mode>3): Likewise.
12999 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
13000 (ia64_expand_atomic_op): Likewise.
13001 (ia64_arg_type): Likewise.
13002 (ia64_mode_to_int): Likewise.
13003 (ia64_scalar_mode_supported_p): Likewise.
13004 (ia64_vector_mode_supported_p): Likewise.
13005 (expand_vec_perm_broadcast): Likewise.
13006 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
13007 (iq2000_function_arg_advance): Likewise.
13008 (iq2000_function_arg): Likewise.
13009 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
13010 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
13011 (m68k_libcall_value): Likewise.
13012 (m68k_function_value): Likewise.
13013 (sched_attr_op_type): Likewise.
13014 * config/mcore/mcore.c (mcore_output_move): Likewise.
13015 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
13017 (microblaze_function_arg): Likewise.
13018 * config/mips/mips.c (mips16_build_call_stub): Likewise.
13019 (mips_print_operand): Likewise.
13020 (mips_mode_ok_for_mov_fmt_p): Likewise.
13021 (mips_vector_mode_supported_p): Likewise.
13022 (mips_preferred_simd_mode): Likewise.
13023 (mips_expand_vpc_loongson_even_odd): Likewise.
13024 (mips_expand_vec_unpack): Likewise.
13025 (mips_expand_vi_broadcast): Likewise.
13026 (mips_expand_vector_init): Likewise.
13027 (mips_expand_vec_reduc): Likewise.
13028 (mips_expand_msa_cmp): Likewise.
13029 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
13030 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
13031 (cc_flags_for_mode): Likewise.
13032 * config/msp430/msp430.c (msp430_print_operand): Likewise.
13033 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
13034 (nds32_output_casesi_pc_relative): Likewise.
13035 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13036 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
13037 (nvptx_gen_unpack): Likewise.
13038 (nvptx_gen_pack): Likewise.
13039 (nvptx_gen_shuffle): Likewise.
13040 (nvptx_gen_wcast): Likewise.
13041 (nvptx_preferred_simd_mode): Likewise.
13042 * config/pa/pa.c (pa_secondary_reload): Likewise.
13043 * config/pa/predicates.md (base14_operand): Likewise.
13044 * config/powerpcspe/powerpcspe-c.c
13045 (altivec_resolve_overloaded_builtin): Likewise.
13046 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
13048 (rs6000_preferred_simd_mode): Likewise.
13049 (output_vec_const_move): Likewise.
13050 (rs6000_expand_vector_extract): Likewise.
13051 (rs6000_split_vec_extract_var): Likewise.
13052 (reg_offset_addressing_ok_p): Likewise.
13053 (rs6000_legitimate_offset_address_p): Likewise.
13054 (rs6000_legitimize_address): Likewise.
13055 (rs6000_emit_set_const): Likewise.
13056 (rs6000_const_vec): Likewise.
13057 (rs6000_emit_move): Likewise.
13058 (spe_build_register_parallel): Likewise.
13059 (rs6000_darwin64_record_arg_recurse): Likewise.
13060 (swap_selector_for_mode): Likewise.
13061 (spe_init_builtins): Likewise.
13062 (paired_init_builtins): Likewise.
13063 (altivec_init_builtins): Likewise.
13064 (do_load_for_compare): Likewise.
13065 (rs6000_generate_compare): Likewise.
13066 (rs6000_expand_float128_convert): Likewise.
13067 (emit_load_locked): Likewise.
13068 (emit_store_conditional): Likewise.
13069 (rs6000_output_function_epilogue): Likewise.
13070 (rs6000_handle_altivec_attribute): Likewise.
13071 (rs6000_function_value): Likewise.
13072 (emit_fusion_gpr_load): Likewise.
13073 (emit_fusion_p9_load): Likewise.
13074 (emit_fusion_p9_store): Likewise.
13075 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
13076 (fusion_gpr_mem_load): Likewise.
13077 (fusion_addis_mem_combo_load): Likewise.
13078 (fusion_addis_mem_combo_store): Likewise.
13079 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
13080 (fusion_gpr_mem_load): Likewise.
13081 (fusion_addis_mem_combo_load): Likewise.
13082 (fusion_addis_mem_combo_store): Likewise.
13083 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13085 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
13086 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
13087 (rs6000_preferred_simd_mode): Likewise.
13088 (output_vec_const_move): Likewise.
13089 (rs6000_expand_vector_extract): Likewise.
13090 (rs6000_split_vec_extract_var): Likewise.
13091 (reg_offset_addressing_ok_p): Likewise.
13092 (rs6000_legitimate_offset_address_p): Likewise.
13093 (rs6000_legitimize_address): Likewise.
13094 (rs6000_emit_set_const): Likewise.
13095 (rs6000_const_vec): Likewise.
13096 (rs6000_emit_move): Likewise.
13097 (rs6000_darwin64_record_arg_recurse): Likewise.
13098 (swap_selector_for_mode): Likewise.
13099 (paired_init_builtins): Likewise.
13100 (altivec_init_builtins): Likewise.
13101 (rs6000_expand_float128_convert): Likewise.
13102 (emit_load_locked): Likewise.
13103 (emit_store_conditional): Likewise.
13104 (rs6000_output_function_epilogue): Likewise.
13105 (rs6000_handle_altivec_attribute): Likewise.
13106 (rs6000_function_value): Likewise.
13107 (emit_fusion_gpr_load): Likewise.
13108 (emit_fusion_p9_load): Likewise.
13109 (emit_fusion_p9_store): Likewise.
13110 * config/rx/rx.c (rx_gen_move_template): Likewise.
13111 (flags_from_mode): Likewise.
13112 * config/s390/predicates.md (s390_alc_comparison): Likewise.
13113 (s390_slb_comparison): Likewise.
13114 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
13115 (s390_vector_mode_supported_p): Likewise.
13116 (s390_cc_modes_compatible): Likewise.
13117 (s390_match_ccmode_set): Likewise.
13118 (s390_canonicalize_comparison): Likewise.
13119 (s390_emit_compare_and_swap): Likewise.
13120 (s390_branch_condition_mask): Likewise.
13121 (s390_rtx_costs): Likewise.
13122 (s390_secondary_reload): Likewise.
13123 (__SECONDARY_RELOAD_CASE): Likewise.
13124 (s390_expand_cs): Likewise.
13125 (s390_preferred_simd_mode): Likewise.
13126 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
13127 * config/sh/sh.c (sh_print_operand): Likewise.
13128 (dump_table): Likewise.
13129 (sh_secondary_reload): Likewise.
13130 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13131 * config/sh/sh.md (casesi_worker_1): Likewise.
13132 (casesi_worker_2): Likewise.
13133 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
13134 (fcc_comparison_operator): Likewise.
13135 * config/sparc/sparc.c (sparc_expand_move): Likewise.
13136 (emit_soft_tfmode_cvt): Likewise.
13137 (sparc_preferred_simd_mode): Likewise.
13138 (output_cbranch): Likewise.
13139 (sparc_print_operand): Likewise.
13140 (sparc_expand_vec_perm_bmask): Likewise.
13141 (vector_init_bshuffle): Likewise.
13142 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
13143 (spu_vector_mode_supported_p): Likewise.
13144 (spu_expand_insv): Likewise.
13145 (spu_emit_branch_or_set): Likewise.
13146 (spu_handle_vector_attribute): Likewise.
13147 (spu_builtin_splats): Likewise.
13148 (spu_builtin_extract): Likewise.
13149 (spu_builtin_promote): Likewise.
13150 (spu_expand_sign_extend): Likewise.
13151 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
13152 (tilegx_simd_int): Likewise.
13153 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
13154 (tilepro_simd_int): Likewise.
13155 * config/v850/v850.c (const_double_split): Likewise.
13156 (v850_print_operand): Likewise.
13157 (ep_memory_offset): Likewise.
13158 * config/vax/vax.c (vax_rtx_costs): Likewise.
13159 (vax_output_int_move): Likewise.
13160 (vax_output_int_add): Likewise.
13161 (vax_output_int_subtract): Likewise.
13162 * config/visium/predicates.md (visium_branch_operator): Likewise.
13163 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
13164 (visium_print_operand_address): Likewise.
13165 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
13166 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
13167 (xtensa_expand_conditional_branch): Likewise.
13168 (xtensa_copy_incoming_a7): Likewise.
13169 (xtensa_output_literal): Likewise.
13170 * dfp.c (decimal_real_maxval): Likewise.
13171 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
13173 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13174 Alan Hayward <alan.hayward@arm.com>
13175 David Sherwood <david.sherwood@arm.com>
13177 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
13178 (mode_nunits_inline): Likewise.
13179 (mode_inner_inline): Likewise.
13180 (mode_unit_size_inline): Likewise.
13181 (mode_unit_precision_inline): Likewise.
13182 (emit_insn_modes_h): Likewise. Also emit a #define of the
13184 (emit_mode_wider): Add an E_ prefix to mode names.
13185 (emit_mode_complex): Likewise.
13186 (emit_mode_inner): Likewise.
13187 (emit_mode_adjustments): Likewise.
13188 (emit_mode_int_n): Likewise.
13189 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
13190 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
13191 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
13192 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
13193 (CRC32_BUILTIN, ENTRY): Likewise.
13194 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
13195 (aarch64_pop_regs): Likewise.
13196 (aarch64_process_components): Likewise.
13197 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
13198 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
13199 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
13200 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
13201 * config/arm/arm.c (arm_init_libfuncs): Likewise.
13202 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
13204 * config/i386/i386-builtin.def (pcmpestr): Likewise.
13205 (pcmpistr): Likewise.
13206 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
13207 * config/mmix/mmix.c (mmix_output_condition): Likewise.
13208 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13210 * config/rl78/rl78.c (mduc_regs): Likewise.
13211 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13212 (htm_expand_builtin): Likewise.
13213 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
13214 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
13215 * config/xtensa/xtensa.c (print_operand): Likewise.
13216 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
13217 (NUM_MODE_VECTOR_INT): Likewise.
13218 * genoutput.c (null_operand): Likewise.
13219 (output_operand_data): Likewise.
13220 * genrecog.c (print_parameter_value): Likewise.
13221 * lra.c (debug_operand_data): Likewise.
13223 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13225 * dwarf2out.c (scompare_loc_descriptor_wide)
13226 (scompare_loc_descriptor_narrow): New functions, split out from...
13227 (scompare_loc_descriptor): ...here.
13228 * expmed.c (emit_store_flag_int): New function, split out from...
13229 (emit_store_flag): ...here.
13231 2017-08-30 Richard Biener <rguenther@suse.de>
13233 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
13234 (dwarf2out_early_finish): Move setting of AT_pubnames from
13235 early debug output to early finish.
13237 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13239 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
13240 and -mdata-region to the assembler.
13242 2017-08-30 Richard Biener <rguenther@suse.de>
13244 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
13246 (gen_subprogram_die): Add DW_AT_object_pointer only early.
13247 (dwarf2out_early_global_decl): Only generate a DIE for the
13248 abstract origin if it doesn't already exist or is a declaration DIE.
13249 (resolve_addr): Do not add the linkage name twice when
13250 generating a stub DIE for the DW_TAG_GNU_call_site target.
13252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
13254 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13255 Use machine_mode rather than int for arg1_mode.
13257 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
13260 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
13261 that the second argument of the built-in functions to unpack
13262 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
13263 switch statement instead a lot of if statements.
13264 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
13265 Allow 64-bit values to be in Altivec registers as well as
13266 traditional floating point registers.
13267 (pack<mode>, FMOVE128_VSX iterator): Likewise.
13269 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
13271 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
13272 MAX_REGS_PER_ADDRESS == 1.
13274 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
13276 * config/i386/i386.opt (flag_fentry): Do not init to -1.
13277 * config/i386/i386.c (ix86_option_override_internal): Simplify
13278 setting of opts->x_flag_entry.
13280 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13281 Jakub Jelinek <jakub@redhat.com>
13282 Richard Biener <rguenther@suse.de>
13284 PR tree-optimization/81503
13285 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
13286 folded constant fits in the target type; reorder tests for clarity.
13288 2017-08-29 Martin Liska <mliska@suse.cz>
13290 * passes.def: Include pass_lower_switch.
13291 * stmt.c (dump_case_nodes): Remove and move to
13292 tree-switch-conversion.
13293 (case_values_threshold): Likewise.
13294 (expand_switch_as_decision_tree_p): Likewise.
13295 (emit_case_decision_tree): Likewise.
13296 (expand_case): Likewise.
13297 (balance_case_nodes): Likewise.
13298 (node_has_low_bound): Likewise.
13299 (node_has_high_bound): Likewise.
13300 (node_is_bounded): Likewise.
13301 (emit_case_nodes): Likewise.
13302 (struct simple_case_node): New struct.
13303 (add_case_node): Remove.
13304 (emit_case_dispatch_table): Use vector instead of case_list.
13305 (reset_out_edges_aux): Remove.
13306 (compute_cases_per_edge): Likewise.
13307 (expand_case): Build list of simple_case_node.
13308 (expand_sjlj_dispatch_table): Use it.
13309 * tree-switch-conversion.c (struct case_node): Moved from
13310 stmt.c and adjusted.
13311 (emit_case_nodes): Likewise.
13312 (node_has_low_bound): Likewise.
13313 (node_has_high_bound): Likewise.
13314 (node_is_bounded): Likewise.
13315 (case_values_threshold): Likewise.
13316 (reset_out_edges_aux): Likewise.
13317 (compute_cases_per_edge): Likewise.
13318 (add_case_node): Likewise.
13319 (dump_case_nodes): Likewise.
13320 (balance_case_nodes): Likewise.
13321 (expand_switch_as_decision_tree_p): Likewise.
13322 (emit_jump): Likewise.
13323 (emit_case_decision_tree): Likewise.
13324 (try_switch_expansion): Likewise.
13325 (do_jump_if_equal): Likewise.
13326 (emit_cmp_and_jump_insns): Likewise.
13327 (fix_phi_operands_for_edge): New function.
13328 (record_phi_operand_mapping): Likewise.
13329 (class pass_lower_switch): New pass.
13330 (pass_lower_switch::execute): New function.
13331 (make_pass_lower_switch): Likewise.
13332 (conditional_probability):
13333 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
13334 * tree-pass.h: Add make_pass_lower_switch.
13336 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
13339 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
13342 2017-08-29 Richard Biener <rguenther@suse.de>
13344 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
13345 we do not add a DW_AT_inline attribute twice.
13346 (gen_subprogram_die): Remove code setting DW_AT_inline on
13347 DECL_ABSTRACT_P nodes.
13349 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
13351 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
13352 calls to internal functions.
13353 (gimplify_modify_expr): Likewise.
13354 * tree-call-cdce.c (use_internal_fn): Likewise.
13355 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
13356 (convert_to_divmod): Set the nothrow flag.
13357 * tree-if-conv.c (predicate_mem_writes): Likewise.
13358 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
13359 (vectorizable_call): Likewise.
13360 (vectorizable_store): Likewise.
13361 (vectorizable_load): Likewise.
13362 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
13363 (vect_recog_mask_conversion_pattern): Likewise.
13365 2017-08-29 Martin Liska <mliska@suse.cz>
13368 * gcc.c (driver_handle_option): Add new argument.
13369 * opts-common.c (handle_option): Pass
13370 target_option_override_hook.
13371 * opts-global.c (lang_handle_option): Add new option.
13372 (set_default_handlers): Add new argument.
13373 (decode_options): Likewise.
13374 * opts.c (target_handle_option): Likewise.
13375 (common_handle_option): Call target_option_override_hook.
13376 * opts.h (struct cl_option_handler_func): Add hook for
13377 target option override.
13378 (struct cl_option_handlers): Likewise.
13379 (set_default_handlers): Add new argument.
13380 (decode_options): Likewise.
13381 (common_handle_option): Likewise.
13382 (target_handle_option): Likewise.
13383 * toplev.c (toplev::main): Pass targetm.target_option.override
13386 2017-08-29 Richard Biener <rguenther@suse.de>
13387 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
13389 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
13390 life to the active subtree.
13392 2017-08-28 Jeff Law <law@redhat.com>
13394 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
13395 derive_equivalences.
13396 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
13397 Code moved into....
13398 (edge_info::derive_equivalences): New private member function
13400 * tree-ssa-dom.c (class edge_info): Changed from a struct
13401 to a class. Add ctor/dtor, methods and data members.
13402 (edge_info::edge_info): Renamed from allocate_edge_info.
13403 Initialize additional members.
13404 (edge_info::~edge_info): New.
13405 (free_dom_edge_info): Delete the edge info.
13406 (record_edge_info): Use new class & associated member functions.
13407 Tighten forms for testing for edge equivalences.
13408 (record_temporary_equivalences): Iterate over the simple
13409 equivalences rather than assuming there's only one per edge.
13410 (cprop_into_successor_phis): Iterate over the simple
13411 equivalences rather than assuming there's only one per edge.
13412 (optimize_stmt): Use operand_equal_p rather than pointer
13413 equality for mini-DSE code.
13415 2017-08-28 Nathan Sidwell <nathan@acm.org>
13417 * gcc.c (execute): Fold SIGPIPE handling into switch
13418 statement. Adjust internal error message.
13420 2017-08-28 Richard Biener <rguenther@suse.de>
13423 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
13424 Do nothing for removed DIEs.
13426 2017-08-28 Richard Biener <rguenther@suse.de>
13428 PR tree-optimization/81977
13429 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
13432 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
13435 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
13436 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
13437 using targetm.gen_mem_thread_fence.
13439 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
13442 * config/i386/i386.md (*<btsc><mode>): Change operand 2
13443 predicate to register_operand. Reorder operands.
13444 (*btr<mode>): Ditto.
13445 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
13446 (*btr<mode>_mask): Ditto.
13448 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
13450 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
13451 * config/rs6000/xmmintrin.h: New file.
13452 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
13454 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13457 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
13458 parameter and_insn and return it.
13459 (recombine_lvx_pattern): Insert a copy to ensure availability of
13460 the base register of the copied masking operation at the point of
13461 the instruction replacement.
13462 (recombine_stvx_pattern): Likewise.
13464 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
13466 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
13467 undocumented switches.
13468 (-mpower9-dform-vector): Likewise.
13469 (-mpower9-dform): Likewise.
13470 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13471 comments to delete references to -mpower9-dform* switches.
13472 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
13473 Delete reference to -mpower9-dform* switches, test for
13474 -mpower9-vector instead.
13475 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
13476 (OTHER_P9_VECTOR_MASKS): Likewise.
13477 (POWERPC_MASKS): Likewise.
13478 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
13479 tests against -mpower9-dform* to -mpower9-vector. Delete code
13480 that checked for -mpower9-dform* consistancy with other options.
13481 Add test for -mpower9-misc to enable other power9 switches.
13482 (rs6000_init_hard_regno_mode_ok): Likewise.
13483 (rs6000_option_override_internal): Likewise.
13484 (rs6000_emit_prologue): Likewise.
13485 (rs6000_emit_epilogue): Likewise.
13486 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
13487 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
13488 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
13490 (emit_fusion_p9_store): Likewise.
13491 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
13492 resetting these macros if the assembler does not support ISA 3.0
13494 (TARGET_P9_DFORM_VECTOR): Likewise.
13495 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
13496 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
13498 2017-08-25 Alan Modra <amodra@gmail.com>
13500 PR rtl-optimization/81747
13501 * cse.c (cse_extended_basic_block): Don't attempt to record
13502 equivalences for degenerate conditional branches that branch
13503 to their fall-through.
13505 2017-08-24 Martin Sebor <msebor@redhat.com>
13507 PR middle-end/81908
13508 * gimple-fold.c (size_must_be_zero_p): New function.
13509 (gimple_fold_builtin_memory_op): Call it.
13511 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
13513 * config/rs6000/mm_malloc.h: New file.
13515 2017-08-24 Bin Cheng <bin.cheng@arm.com>
13517 PR tree-optimization/81913
13518 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
13519 analysis when either IVs in condition can wrap.
13521 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
13523 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
13524 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
13526 2017-08-24 Richard Biener <rguenther@suse.de>
13529 * targhooks.c (default_target_can_inline_p): Properly
13530 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
13531 is present and always compare.
13532 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
13533 infer -mfpmath=sse from TARGET_SSE_P.
13534 (ix86_can_inline_p): Properly use target_option_default_node when
13535 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
13537 2017-08-24 Richard Biener <rguenther@suse.de>
13540 * dwarf2out.c (output_die): Handle flag_generate_offload like
13542 (output_comp_unit): Likewise.
13543 (gen_array_type_die): Likewise.
13544 (dwarf2out_early_finish): Likewise.
13545 (note_variable_value_in_expr): Likewise.
13546 (dwarf2out_finish): Likewise. Adjust assert.
13547 * cgraphunit.c (symbol_table::compile): Move setting of
13548 flag_generate_offload earlier ...
13549 (symbol_table::finalize_compilation_unit): ... here, before
13550 early debug finalization.
13552 2017-08-24 Richard Biener <rguenther@suse.de>
13554 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
13555 and ipa-fnsummary.h.
13556 (ix86_can_inline_p): When ix86_fpmath flags do not match
13557 check whether the callee uses FP math at all.
13559 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
13561 PR middle-end/81931
13562 * tree-ssanames.c (get_nonzero_bits): Use element_precision
13563 instead of TYPE_PRECISION.
13565 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
13566 Alan Hayward <alan.hayward@arm.com>
13567 David Sherwood <david.sherwood@arm.com>
13569 * combine.c (make_extraction): Use subreg_offset_from_lsb.
13571 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
13573 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
13575 (ix86_frame::stack_realign_allocate): New field.
13576 (struct machine_frame_state): Modify comments.
13577 (machine_frame_state::sp_realigned_fp_end): New field.
13578 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
13579 layout calculation.
13580 (sp_valid_at): Add assertion to assure no attempt to access invalid
13581 offset of a realigned stack.
13582 (fp_valid_at): Likewise.
13583 (choose_baseaddr): Modify comments.
13584 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
13585 ix86_expand_prologue.
13586 (ix86_expand_prologue): Modify stack realignment and allocation.
13587 (ix86_expand_epilogue): Modify comments.
13588 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
13589 avx2_runtime, avx512f, and avx512f_runtime.
13591 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
13593 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
13594 (mstackrealign): Do not init to -1.
13595 * config/i386/i386.c (ix86_option_override_internal):
13596 Check opts_set, not opts when setting default value of
13597 opts->x_ix86_force_align_arg_pointer.
13599 2017-08-23 Richard Biener <rguenther@suse.de>
13601 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
13602 lang_hooks.decl_printable_name.
13603 * print-rtl-function.c (print_rtx_function): Likewise.
13604 * tree-pretty-print.c (dump_function_header): Likewise.
13606 2017-08-23 Richard Biener <rguenther@suse.de>
13609 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
13610 -g0 at compile-time.
13612 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13614 PR middle-end/19706
13615 * doc/sourcebuild.texi (Other hardware attributes):
13618 2017-08-23 Tamar Christina <tamar.christina@arm.com>
13620 PR middle-end/19706
13621 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
13622 Fix single-use check.
13624 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
13626 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
13628 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13630 * doc/install.texi: Modify to add more details on running selected
13633 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
13635 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
13636 is combined with -mabi=ms.
13637 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
13640 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13642 PR tree-optimization/81488
13643 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
13644 and cached_basis fields.
13645 (MAX_SPREAD): New constant.
13646 (alloc_cand_and_find_basis): Initialize new fields.
13647 (clear_visited): New function.
13648 (create_phi_basis_1): Rename from create_phi_basis, set visited
13649 and cached_basis fields.
13650 (create_phi_basis): New wrapper function.
13651 (phi_add_costs_1): Rename from phi_add_costs, add spread
13652 parameter, set visited field, short-circuit when limits reached.
13653 (phi_add_costs): New wrapper function.
13654 (record_phi_increments_1): Rename from record_phi_increments, set
13656 (record_phi_increments): New wrapper function.
13657 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
13658 (phi_incr_cost): New wrapper function.
13659 (all_phi_incrs_profitable_1): Rename from
13660 all_phi_incrs_profitable, set visited field.
13661 (all_phi_incrs_profitable): New wrapper function.
13663 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13664 Alan Hayward <alan.hayward@arm.com>
13665 David Sherwood <david.sherwood@arm.com>
13667 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
13668 that takes the outer and inner modes.
13669 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
13670 comparison as the canonical test for a paradoxical subreg.
13671 * combine.c (simplify_set): Use paradoxical_subreg_p.
13672 (make_extraction): Likewise.
13673 (force_to_mode): Likewise.
13674 (rtx_equal_for_field_assignment_p): Likewise.
13675 (gen_lowpart_for_combine): Likewise.
13676 (simplify_comparison): Likewise.
13677 * cse.c (equiv_constant): Likewise.
13678 * expmed.c (store_bit_field_1): Likewise.
13679 * final.c (alter_subreg): Likewise.
13680 * fwprop.c (propagate_rtx): Likewise.
13681 (forward_propagate_subreg): Likewise.
13682 * ira-conflicts.c (ira_build_conflicts): Likewise.
13683 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
13684 * lra-constraints.c (curr_insn_transform): Likewise.
13685 (split_reg): Likewise.
13686 * lra-eliminations.c (move_plus_up): Likewise.
13687 (lra_eliminate_regs_1): Likewise.
13688 * recog.c (general_operand): Likewise.
13689 * ree.c (combine_reaching_defs): Likewise.
13690 * reload.c (push_reload): Likewise.
13691 (find_reloads): Likewise.
13692 * reload1.c (elimination_effects): Likewise.
13693 (compute_reload_subreg_offset): Likewise.
13694 (choose_reload_regs): Likewise.
13695 * rtlanal.c (subreg_lsb_1): Likewise.
13696 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
13697 (simplify_subreg): Likewise.
13698 * var-tracking.c (track_loc_p): Likewise.
13699 * emit-rtl.c (byte_lowpart_offset): Likewise.
13700 (paradoxical_subreg_p): Delete out-of-line definition.
13702 2017-08-22 Jeff Law <law@redhat.com>
13704 PR tree-optimization/81741
13705 PR tree-optimization/71947
13706 * tree-ssa-dom.c: Include tree-inline.h.
13707 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
13708 equivalences if one is more expensive to compute than the other.
13709 * tree-ssa-scopedtables.h (class const_or_copies): Make
13710 record_const_or_copy_raw method private.
13711 (class avail_exprs_stack): New method simplify_binary_operation.
13712 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
13713 avail_exprs_stack::simplify_binary_operation as needed.
13714 (avail_exprs_stack::simplify_binary_operation): New function.
13716 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
13718 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
13719 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
13720 (DOT_SYMBOLS): Likewise.
13721 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
13722 (RELOCATABLE_NEEDS_FIXUP): Likewise.
13723 (RS6000_ABI_NAME): Likewise.
13724 (TARGET_CMODEL): Likewise.
13725 (TOC_SECTION_ASM_OP): Likewise.
13726 (SET_CMODEL): New macro.
13727 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
13729 2017-08-22 Richard Biener <rguenther@suse.de>
13731 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
13732 to allow for free-lang-data replacements similar to verify_type_variant.
13734 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
13735 Alan Hayward <alan.hayward@arm.com>
13736 David Sherwood <david.sherwood@arm.com>
13738 * config/aarch64/aarch64.md (casesi): Use DImode rather than
13739 VOIDmode for the LABEL_REF.
13741 2017-08-22 Richard Biener <rguenther@suse.de>
13743 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
13745 2017-08-22 Richard Biener <rguenther@suse.de>
13747 * common.opt (feliminate-dwarf2-dups): Ignore.
13748 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
13749 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
13750 same_die_p_wrap, compute_section_prefix,
13751 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
13752 (comdat_symbol_id, comdat_symbol_number): Likewise.
13753 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
13755 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
13756 (output_die): Mark unreachable path unreachable.
13757 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
13758 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
13759 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
13760 (dwarf2out_early_finish): Likewise.
13762 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
13764 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
13766 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
13769 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
13770 not VAR_P. Filter attribute warnings with OPT_Wattributes.
13771 (avr_attribute_table) <io, io_low, address>: Initialize
13772 .decl_required with true.
13774 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
13776 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
13777 undocumented debugging options.
13778 (-mvsx-scalar-double): Likewise.
13779 (-mallow-df-permute): Likewise.
13780 (-mvectorize-builtins): Likewise.
13781 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13782 (rs6000_builtin_vectorized_function): Likewise.
13783 (rs6000_builtin_md_vectorized_function): Likewise.
13784 (rs6000_opt_vars): Likewise.
13786 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
13789 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
13790 (*btrq_imm): Rename from *btrq.
13791 (*btcq_imm): Rename from *btcq.
13792 (btsc): New code attribute.
13793 (*<btsc><mode>): New insn pattern.
13794 (*btr<mode>): Ditto.
13795 (*<btsc><mode>_mask): New insn_and_split pattern.
13796 (*btr<mode>_mask): Ditto.
13798 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13799 Alan Hayward <alan.hayward@arm.com>
13800 David Sherwood <david.sherwood@arm.com>
13802 * function.c (pad_below): Simplify padding calculation.
13804 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13805 Alan Hayward <alan.hayward@arm.com>
13806 David Sherwood <david.sherwood@arm.com>
13808 * target.def (function_prologue): Remove frame size argument.
13809 (function_epilogue): Likewise.
13810 * doc/tm.texi: Regenerate.
13811 * final.c (final_start_function): Update call to function_prologue.
13812 (final_end_function): Update call to function_epilogue.
13813 (default_function_pro_epilogue): Remove frame size argument.
13814 * output.h (default_function_pro_epilogue): Likewise.
13815 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
13816 (arm_output_function_prologue): Likewise.
13817 * config/frv/frv.c (frv_function_prologue): Likewise.
13818 (frv_function_epilogue): Likewise.
13819 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
13820 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
13821 (ia64_output_function_epilogue): Likewise.
13822 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
13823 (m32r_output_function_epilogue): Likewise.
13824 * config/microblaze/microblaze.c (microblaze_function_prologue)
13825 (microblaze_function_epilogue): Likewise.
13826 * config/mips/mips.c (mips_output_function_prologue): Likewise.
13827 (mips_output_function_epilogue): Likewise.
13828 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
13829 (mmix_target_asm_function_epilogue): Likewise.
13830 * config/msp430/msp430.c (msp430_start_function): Likewise.
13831 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
13832 (nds32_asm_function_epilogue): Likewise.
13833 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
13834 * config/pa/pa.c (pa_output_function_prologue): Likewise.
13835 (pa_output_function_epilogue): Likewise.
13836 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
13837 (rs6000_output_function_epilogue): Likewise.
13838 * config/rl78/rl78.c (rl78_start_function): Likewise.
13839 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
13840 (rs6000_output_function_epilogue): Likewise.
13841 * config/rx/rx.c (rx_output_function_prologue): Likewise.
13842 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
13843 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
13844 (sparc_asm_function_epilogue): Likewise.
13846 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13848 * tree.h (type_has_mode_precision_p): New function.
13849 * convert.c (convert_to_integer_1): Use it.
13850 * expr.c (expand_expr_real_2): Likewise.
13851 (expand_expr_real_1): Likewise.
13852 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
13853 * match.pd: Likewise.
13854 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
13855 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
13856 * tree-tailcall.c (process_assignment): Likewise.
13857 * tree-vect-loop.c (vectorizable_reduction): Likewise.
13858 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
13859 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
13860 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
13861 (vectorizable_assignment): Likewise.
13862 (vectorizable_shift): Likewise.
13863 (vectorizable_operation): Likewise.
13864 * tree-vrp.c (register_edge_assert_for_2): Likewise.
13866 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
13868 * match.pd: Add pow (C, x) simplification.
13870 2017-08-21 Richard Biener <rguenther@suse.de>
13872 PR tree-optimization/81900
13873 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
13874 for blocks with abnormal predecessors.
13875 (compute_antic): Do not set visited flag prematurely.
13877 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
13880 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
13882 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
13884 * stor-layout.h (vector_type_mode): Move to...
13885 * tree.h (vector_type_mode): ...here.
13886 * stor-layout.c (vector_type_mode): Move to...
13887 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
13889 2017-08-21 Richard Biener <rguenther@suse.de>
13891 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
13892 register_external_die hooks.
13893 (debug_false_tree_charstarstar_uhwistar): Declare.
13894 (debug_nothing_tree_charstar_uhwi): Likewise.
13895 * debug.c (do_nothing_debug_hooks): Adjust.
13896 (debug_false_tree_charstarstar_uhwistar): New do nothing.
13897 (debug_nothing_tree_charstar_uhwi): Likewise.
13898 * dbxout.c (dbx_debug_hooks): Adjust.
13899 (xcoff_debug_hooks): Likewise.
13900 * sdbout.c (sdb_debug_hooks): Likewise.
13901 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13902 * dwarf2out.c (macinfo_label_base): New global.
13903 (dwarf2out_register_external_die): New function for the
13904 register_external_die hook.
13905 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
13906 (dwarf2_debug_hooks): Use them.
13907 (dwarf2_lineno_debug_hooks): Adjust.
13908 (struct die_struct): Add with_offset flag.
13909 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
13910 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
13911 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
13912 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
13913 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
13914 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
13915 defining section names for the early LTO debug variants.
13916 (reset_indirect_string): New helper.
13917 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
13918 (print_dw_val): Add support for offsetted symbol references.
13919 (get_ultimate_context): Split out from is_cxx.
13920 (is_cxx): Use get_ultimate_context.
13921 (is_fortran): Add decl overload.
13922 (compute_comp_unit_symbol): Split out worker from
13923 compute_section_prefix.
13924 (compute_section_prefix): Call compute_comp_unit_symbol and
13925 set comdat_type_p here.
13926 (output_die): Skip DIE symbol output for the LTO added one.
13927 Handle DIE symbol references with offset.
13928 (output_comp_unit): Guard section name mangling properly.
13929 For LTO debug sections emit a symbol at the section beginning
13930 which we use to refer to its DIEs.
13931 (add_abstract_origin_attribute): For DIEs registered via
13932 dwarf2out_register_external_die directly refer to the early
13933 DIE rather than indirectly through the shadow one we created.
13934 Remove obsolete call to dwarf2out_abstract_function for
13935 non-function/block origins.
13936 (gen_array_type_die): When generating early LTO debug do
13937 not emit DW_AT_string_length.
13938 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
13939 late when in LTO. As suggested place a gcc_unreachable for
13940 the DECL_ABSTRACT_P case.
13941 (gen_subprogram_die): Avoid another specification DIE
13942 for early built declarations/definitions for the late LTO case.
13943 (gen_variable_die): Add type references for late duplicated VLA dies
13945 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
13946 we have the abstract instance already.
13947 (process_scope_var): Adjust decl DIE contexts in LTO which
13948 first puts them in limbo.
13949 (gen_decl_die): Do not generate type DIEs late apart from
13950 types for VLAs or for decls we do not yet have a DIE. Do not
13951 call dwarf2out_abstract_function late.
13952 (dwarf2out_early_global_decl): Make sure to create DIEs
13953 for abstract instances of a decl first.
13954 (dwarf2out_late_global_decl): Adjust comment.
13955 (output_macinfo_op): With multiple macro sections use
13956 macinfo_label_base to distinguish labels.
13957 (output_macinfo): Likewise. Update macinfo_label_base.
13958 Pass in the line info label.
13959 (note_variable_value_in_expr): When generating LTO resolve
13960 all variable values here by generating DIEs as needed.
13961 (init_sections_and_labels): Add early LTO debug flag parameter
13962 and generate different sections and names if set. Add generation
13963 counter for the labels so we can have multiple of them.
13964 (reset_dies): Helper to allow DIEs to be output multiple times.
13965 (dwarf2out_finish): When outputting DIEs to the fat part of an
13966 LTO object first reset DIEs.
13967 (dwarf2out_early_finish): Output early DIEs when generating LTO.
13968 (modified_type_die): Check for decl_ultimate_origin being self
13970 (gen_type_die_with_usage): Likewise.
13971 (gen_typedef_die): Allow decl_ultimate_origin being self.
13972 (set_decl_abstract_flags): Remove.
13973 (set_block_abstract_flags): Likewise.
13974 (dwarf2out_abstract_function): Treat the early generated DIEs
13975 as the abstract copy and only add DW_AT_inline and
13976 DW_AT_artificial here and call set_decl_origin_self.
13977 If the DIE has an abstract origin don't do anything.
13978 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
13979 if we have none yet (Go fails to build one, PR78628).
13980 (variably_modified_type_p): Prevent endless recursion for Ada
13981 cyclic pointer types.
13982 * lto-streamer-in.c: Include debug.h.
13983 (dref_queue): New global.
13984 (lto_read_tree_1): Stream in DIE references.
13985 (lto_input_tree): Register DIE references.
13986 (input_function): Stream DECL_DEBUG_ARGS.
13987 * lto-streamer-out.c: Include debug.h.
13988 (lto_write_tree_1): Output DIE references.
13989 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
13990 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
13991 (output_function): Stream DECL_DEBUG_ARGS.
13992 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
13993 Stream DECL_ABSTRACT_ORIGIN.
13994 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
13995 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
13996 DECL_CONTEXT for file-scope decls.
13997 * lto-streamer.h (struct dref_entry): Declare.
13998 (dref_queue): Likewise.
13999 * cfgexpand.c (pass_expand::execute): Do not call the
14000 outlining_inline_function hook here.
14001 * lto-wrapper.c (debug_obj): New global.
14002 (tool_cleanup): Unlink it if required.
14003 (debug_objcopy): New function.
14004 (run_gcc): Handle early debug sections in the IL files by
14005 extracting them to separate files, partially linkin them and
14006 feeding the result back as result to the linker.
14007 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
14008 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
14009 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
14010 sections into a separate segment.
14011 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
14013 (darwin_asm_dwarf_section): Likewise.
14014 (darwin_asm_output_dwarf_offset): Likewise.
14015 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
14017 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14018 Alan Hayward <alan.hayward@arm.com>
14019 David Sherwood <david.sherwood@arm.com>
14021 * read-md.h (md_reader::record_potential_iterator_use): Replace
14022 pointer argument with an rtx and an index.
14023 * read-rtl.c (iterator_group::apply_iterator): Likewise.
14024 (apply_mode_iterator): Likewise.
14025 (apply_code_iterator): Likewise.
14026 (apply_int_iterator): Likewise.
14027 (apply_subst_iterator): Likewise.
14028 (record_iterator_use): Likewise.
14029 (record_attribute_use): Likewise.
14030 (md_reader::record_potential_iterator_use): Likewise. Update calls
14031 to record_iterator_use and apply_iterator.
14032 (iterator_use): Replace ptr with x and index.
14033 (attribute_use): Likewise.
14034 (apply_attribute_uses): Update calls to apply_iterator.
14035 (apply_iterators): Likewise. Update initialization of iterator_use.
14036 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
14037 and record_potential_iterator_use.
14038 (rtx_reader::read_rtx_operand): Likewise.
14040 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
14041 Alan Hayward <alan.hayward@arm.com>
14042 David Sherwood <david.sherwood@arm.com>
14044 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
14047 2017-08-21 Richard Biener <rguenther@suse.de>
14049 PR middle-end/81884
14050 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
14051 at struct end conservatively when comparing common bases.
14053 2017-08-21 Richard Biener <rguenther@suse.de>
14055 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
14056 (mem_ref_in_stmt): Remove.
14057 (determine_max_movement): Use ref index to get at the reference.
14058 (invariantness_dom_walker::before_dom_children): Deal with
14059 lim data already initialized.
14060 (gather_mem_refs_stmt): Initialize lim data and record ref index.
14062 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14064 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
14065 (TARGET_ISA_ROUND): Ditto.
14066 (TARGET_ROUND): Ditto.
14067 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
14068 * config/i386/i386.md: Ditto.
14069 * config/i386/sse.md: Ditto.
14070 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
14071 with OPTION_MASK_ISA_SSE4_1.
14073 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
14076 * doc/extend.texi (x86 Built-in Functions): Correct the name of
14077 __builtin_ia32_lzcnt_u16.
14079 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
14082 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
14083 (rs6000_set_current_function): Rewrite function to use it.
14085 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
14088 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
14089 and TYPE_WARN_IF_NOT_ALIGN.
14090 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
14091 (handle_warn_if_not_align): New.
14092 (place_union_field): Call handle_warn_if_not_align.
14093 (place_field): Call handle_warn_if_not_align.
14094 Copy TYPE_WARN_IF_NOT_ALIGN.
14095 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
14096 (layout_type): Likewise.
14097 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
14099 (tree_decl_common): Add warn_if_not_align.
14100 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
14101 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
14102 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
14103 (DECL_WARN_IF_NOT_ALIGN): Likewise.
14104 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
14105 * doc/extend.texi: Document warn_if_not_aligned attribute.
14106 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
14108 2017-08-17 Martin Liska <mliska@suse.cz>
14111 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
14112 (get_data_dependence): Use it as pointer type.
14113 (distribute_loop): Likewise.
14115 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14117 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
14118 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
14119 (p8_vmrgow_<mode>_direct): New define_insn.
14120 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
14121 handle endianness for vmrgew and vmrgow permute patterns.
14123 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14125 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
14126 * config/rs6000/rs6000-cpus.def: Remove comment.
14127 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
14128 (POWERPC_MASKS): Likewise.
14129 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
14130 use of TARGET_VSX_TIMODE.
14131 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
14132 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
14133 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
14134 (rs6000_option_override_internal): Remove dead code.
14135 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
14136 (rs6000_legitimize_reload_address): Likewise.
14137 (rs6000_legitimate_address_p): Likewise.
14138 (rs6000_opt_masks): Delete "vsx-timode".
14139 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
14140 from function comment.
14141 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
14142 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
14143 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
14145 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
14146 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
14147 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
14148 (VSX_TI): Likewise.
14150 (define_peephole2): Likewise.
14152 2017-08-17 Martin Sebor <msebor@redhat.com>
14155 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
14156 past the end of an array.
14157 (test_pp_format): Add test cases.
14159 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
14161 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
14162 missing ECF_NOTHROW flags.
14164 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
14167 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
14168 operands residing in integer registers.
14169 (*vsx_le_perm_load_<mode>): Likewise.
14170 (*vsx_le_perm_store_<mode>): Likewise.
14171 (define_peephole2): Add peepholes to optimize the above.
14173 2017-08-17 Marek Polacek <polacek@redhat.com>
14175 PR middle-end/81814
14176 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
14177 to mimic what shorten_compare did. Change the return type to bool.
14178 (fold_cond_expr_with_comparison): Update call to
14179 operand_equal_for_comparison_p.
14180 (fold_ternary_loc): Likewise.
14182 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
14184 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
14186 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
14188 2017-08-17 Richard Biener <rguenther@suse.de>
14190 * tree-ssa-structalias.c (solve_graph): When propagating
14191 to successors update the graphs succ edges and avoid duplicate work.
14193 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
14196 * config/i386/i386.c (ix86_option_override_internal): Save target
14197 specific options after ix86_stack_protector_guard_reg was changed.
14199 2017-08-17 Richard Biener <rguenther@suse.de>
14201 PR tree-optimization/81827
14202 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
14203 (new_var_info): Initialize it conservatively.
14204 (get_call_vi): Mark register vars.
14205 (new_scalar_tmp_constraint_exp): Likewise.
14206 (handle_rhs_call): Likewise.
14207 (handle_const_call): Likewise.
14208 (create_function_info_for): Likewise.
14209 (solve_constraints): Sort varinfos to separate register from
14210 non-register vars to pack points-to solution bitmaps during
14213 2017-08-17 Marek Polacek <polacek@redhat.com>
14215 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
14217 2017-08-17 Richard Biener <rguenther@suse.de>
14219 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
14220 to true when overflow is undefined and we saturated the result.
14222 2017-08-17 Alan Modra <amodra@gmail.com>
14225 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
14226 Don't use store multiple if only one reg needs saving.
14227 (interesting_frame_related_regno): New function.
14228 (rs6000_frame_related): Don't emit frame info for regs that
14230 (rs6000_emit_epilogue): Likewise.
14232 2017-08-16 Nathan Sidwell <nathan@acm.org>
14234 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
14235 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
14236 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
14237 * tree.c (free_lang_data_in_type): Use else-if chain. Always
14238 clear TYPE_LANG_1. Remove obsolete member-function stripping.
14239 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
14240 (verify_type): Adjust for TYPE_BINFO move.
14241 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
14242 process TYPE_BINFO directly.
14243 (hash_tree): Likewise.
14244 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14246 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14249 2017-08-16 David Malcolm <dmalcolm@redhat.com>
14251 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
14253 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
14256 * config/i386/i386.md (*anddi_1_btr): Change predicates of
14257 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
14258 Add ix86_binary_operator_ok to insn constraint.
14259 (*iordi_1_bts): Ditto.
14260 (*xordi_1_btc): Ditto.
14261 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
14262 Update corresponding peephole2 pattern.
14266 2017-08-16 Bin Cheng <bin.cheng@arm.com>
14268 PR tree-optimization/81832
14269 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
14270 copy loop header which has IFN_LOOP_DIST_ALIAS call.
14272 2017-08-16 Marek Polacek <polacek@redhat.com>
14275 * fold-const.c (fold_indirect_ref_1): Restore original behavior
14276 regarding size_zero_node.
14278 2017-08-16 Martin Liska <mliska@suse.cz>
14281 * config.gcc: Respect previously set extra_objs in case
14284 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
14286 PR tree-optimization/81835
14287 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
14288 the phi SSA_NAME. Check that the condition in a COND_EXPR does
14289 not depend on the phi.
14291 2017-08-16 Alan Modra <amodra@gmail.com>
14293 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
14296 2017-08-16 Alan Modra <amodra@gmail.com>
14298 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
14299 (save_reg_p): ..into this. Update all callers.
14300 (first_reg_to_save): Simplify.
14302 2017-08-16 Alan Modra <amodra@gmail.com>
14304 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
14307 2017-08-15 Joseph Myers <joseph@codesourcery.com>
14311 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
14312 constant count if that count is less than 32.
14314 2017-08-15 Nathan Sidwell <nathan@acm.org>
14316 * gcc.c (execute): Emit friendlier message if inferior is killed
14317 by an external cause.
14319 2017-08-15 Richard Biener <rguenther@suse.de>
14321 PR tree-optimization/81790
14322 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
14323 CONSTRUCTORs from simplifying and VN.
14325 2017-08-14 Martin Sebor <msebor@redhat.com>
14327 * builtin-attrs.def: Add comments.
14329 2017-08-14 Martin Sebor <msebor@redhat.com>
14332 * doc/extend.texi (attribute nonstring): Document new attribute.
14334 2017-08-14 Martin Sebor <msebor@redhat.com>
14337 * tree-diagnostic.c (default_tree_printer): Handle %G.
14338 * gimple-pretty-print.h (percent_G_format): Declare new function.
14339 * gimple-pretty-print.c (percent_G_format): Define.
14340 * tree-pretty-print.c (percent_K_format): Add argument.
14342 2017-08-14 Martin Sebor <msebor@redhat.com>
14344 PR translation/79998
14345 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
14346 Remove a stray space.
14348 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
14351 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
14352 (*iordi_1_bts): Ditto.
14353 (*xordi_1_btc): Ditto.
14355 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14358 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
14359 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14361 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
14362 quoted strings, and make more translator-friendly.
14363 (darwin_rs6000_override_options): Likewise.
14364 (rs6000_option_override_internal): Likewise.
14365 (rs6000_return_in_memory): Fix overlong line.
14366 (init_cmulative_args): Use quoted strings, and make more
14367 translator-friendly.
14368 (rs6000_pass_by_reference): Fix overlong line.
14369 (def_builtin): Use quoted strings.
14370 (altivec_expand_predicate_builtin): Use quoted strings, and make
14371 more translator-friendly.
14372 (htm_expand_builtin): Use quoted strings.
14373 (cpu_expand_builtin): Use quoted strings, and make more
14374 translator-friendly.
14375 (altivec_expand_builtin): Likewise.
14376 (paired_expand_predicate_builtin): Likewise.
14377 (rs6000_invalid_builtin): Likewise.
14378 (builtin_function_type): Use quoted strings.
14379 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
14380 more translator-friendly.
14381 (rs6000_trampoline_init): Likewise.
14382 (rs6000_handle_altivec_attribute): Likewise.
14383 (rs6000_inner_target_options): Use quoted strings.
14384 (rs6000_disable_incompatible_switches): Likewise.
14385 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
14386 strings, and make more translator-friendly.
14387 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
14389 2017-08-14 Bin Cheng <bin.cheng@arm.com>
14391 PR tree-optimization/81799
14392 * tree-loop-distribution.c (version_loop_by_alias_check): Force
14393 cond_expr to simple gimple operand.
14395 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
14397 PR middle-end/46932
14398 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
14400 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
14404 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
14405 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
14406 TARGET_GASISR_PROLOGUES.
14407 * config/avr/avr.c (avr_option_override): Same.
14408 (avr_pass_pre_proep::execute): Same.
14410 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
14413 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
14414 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
14416 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
14418 * config/i386/i386.md (*load_tp_<mode>): Redefine as
14419 define_insn_and_split. Split to a memory load from 0 in
14420 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
14421 using PTR mode iterator.
14422 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
14423 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
14424 (*add_tp_<mode>): Redefine as define_insn_and_split.
14425 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
14426 address space. Merge with *add_tp_x32 using PTR mode iterator.
14427 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
14428 Split to an add with a memory load from 0 in
14429 DEFAULT_TLS_SEG_REG address space.
14431 2017-08-12 Andrew Pinski <apinski@cavium.com>
14433 * config/aarch64/aarch64-option-extensions.def (rdma):
14434 Fix feature string to what Linux prints out in /proc/cpuinfo.
14436 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
14439 * dwarf2out.c (modified_type_die): For C typedef types that have
14440 an ultimate origin, process the ultimate origin instead of the
14442 (gen_typedef_die): Assert that input DECLs have no ultimate
14444 (gen_type_die_with_usage): For typedef variants that have an
14445 ultimate origin, just call gen_decl_die on the original DECL.
14446 (process_scope_var): Avoid creating DIEs for local typedefs and
14447 concrete static variables.
14449 2017-08-12 Alan Modra <amodra@gmail.com>
14453 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
14454 match gnu-user.h startfile.
14455 (ENDFILE_LINUX_SPEC): Similarly.
14457 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
14460 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
14462 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
14464 2017-08-11 Tamar Christina <tamar.christina@arm.com>
14465 * config/aarch64/aarch64.md (mov<mode>): Change.
14466 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14467 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
14468 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
14470 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
14472 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
14473 for storage order barriers.
14475 2017-08-11 Martin Liska <mliska@suse.cz>
14478 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
14479 variables of void type.
14481 2017-08-11 Martin Liska <mliska@suse.cz>
14483 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
14484 TARGET_SUPPORTS_ALIASES.
14485 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
14486 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
14487 (optimize_weakref): Likewise.
14488 * symtab.c (symtab_node::noninterposable_alias): Likewise.
14489 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
14490 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
14492 2017-08-11 Martin Liska <mliska@suse.cz>
14495 * config/i386/i386.c (make_resolver_func): Do complete
14496 refactoring of the function.
14498 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14501 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
14502 * config/i386/i386.c (ix86_stack_protect_guard): Use
14503 ix86_stack_protect_guard_symbol_str to generate varible declaration.
14504 * doc/invoke.texi (x86 Options): Document
14505 -mstack-protector-guard-symbol= option.
14507 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
14509 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
14510 * config/i386/i386.c (ix86_split_stack_guard): New function.
14511 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
14512 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
14513 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
14514 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
14515 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
14516 (split_stack_space_check): Call ix86_split_stack_guard.
14518 2017-08-10 Martin Sebor <msebor@redhat.com>
14520 * print-tree.c (print_node): Print location using the established
14522 Replace spaces with colons.
14523 (debug_raw, debug): Ditto.
14525 2017-08-10 Martin Sebor <msebor@redhat.com>
14528 * pretty-print.c (pp_format): Correct the handling of %s precision.
14530 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
14533 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14535 (ix86_finalize_stack_frame_flags): This. Also clear
14536 frame_pointer_needed if -fno-omit-frame-pointer is used without
14538 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14539 with ix86_finalize_stack_frame_flags.
14540 (ix86_expand_epilogue): Likewise.
14541 (ix86_expand_split_stack_prologue): Likewise.
14542 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
14544 2017-08-10 Martin Liska <mliska@suse.cz>
14547 * c-attribs.c (handle_target_attribute):
14548 Report warning for an empty string argument of target attribute.
14550 2017-08-09 Jakub Jelinek <jakub@redhat.com>
14553 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
14555 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
14556 or DECL_NONLOCAL labels.
14557 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
14558 or DECL_NONLOCAL labels here.
14560 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
14562 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
14563 to indicate when early gimple folding has been disabled.
14564 (rs6000_gimple_fold_builtin): Add debug content.
14565 (rs6000_invalid_builtin): Fix whitespace.
14566 (rs6000_expand_builtin): Fix whitespace.
14567 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
14569 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
14572 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
14573 SAVE_MULTIPLE if not all the registers that saves, should be saved.
14575 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
14577 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
14578 (qdf24xx): Likewise.
14579 * config/aarch64/aarch64.md: Include falkor.md.
14580 * config/aarch64/falkor.md: New.
14582 2017-08-09 Marek Polacek <polacek@redhat.com>
14585 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
14586 * diagnostic.c (emit_diagnostic): Add a comment.
14587 (emit_diagnostic_valist): New function.
14589 2017-08-09 Marek Polacek <polacek@redhat.com>
14592 * input.c (make_location): New overload.
14593 * input.h (make_location): Declare.
14595 2017-08-08 Alan Modra <amodra@gmail.com>
14596 H.J. Lu <hongjiu.lu@intel.com>
14599 * gcc.c (NO_PIE_SPEC): Delete.
14600 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
14602 (LINK_PIE_SPEC): ..to here.
14603 (LINK_COMMAND_SPEC): Support -no-pie.
14604 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
14605 chain of crtbegin*.o selection, update for PIE_SPEC changes and
14607 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
14608 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
14609 (ENDFILE_CRTEND_SPEC): Similarly.
14611 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
14614 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
14615 (mstack-protector-guard-offset=): Ditto.
14616 * config/i386/i386.c (ix86_option_override): Handle
14617 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
14619 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
14620 ix86_stack_protect_guard_offset variables.
14621 (TARGET_STACK_PROTECT_GUARD): Always define.
14622 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
14623 and -mstack-protector-guard-offset= options.
14625 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14627 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
14628 boundary case for the last candidate.
14630 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14632 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
14633 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
14635 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14637 PR middle-end/19706
14638 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
14639 * config/aarch64/aarch64-builtins.c
14640 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
14641 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
14642 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
14644 2017-08-08 Tamar Christina <tamar.christina@arm.com>
14645 Andrew Pinski <pinskia@gmail.com>
14647 PR middle-end/19706
14648 * internal-fn.def (XORSIGN): New.
14649 * optabs.def (xorsign_optab): New.
14650 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
14651 (convert_expand_mult_copysign): New.
14652 (pass_optimize_widening_mul::execute): Call
14653 convert_expand_mult_copysign.
14655 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14657 PR tree-optimization/81354
14658 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
14659 Insert on edges rather than explicitly creating landing pads.
14660 (analyze_candidates_and_replace): Commit edge inserts.
14662 2017-08-08 Richard Biener <rguenther@suse.de>
14664 PR middle-end/81719
14665 * tree-ssa-loop-niter.c: Include tree-dfa.h.
14666 (expand_simple_operations): Also look through ADDR_EXPRs with
14667 MEM_REF bases treating them as POINTER_PLUS_EXPR.
14669 2017-08-08 Richard Biener <rguenther@suse.de>
14671 PR tree-optimization/81723
14672 * tree-vect-slp.c (struct bst_traits): New hash traits.
14673 (bst_fail): New global.
14674 (vect_build_slp_tree_2): New worker, split out from ...
14675 (vect_build_slp_tree): ... this now wrapping it with using
14676 bst_fail set to cache SLP tree build fails. Properly handle
14678 (vect_analyze_slp_instance): Allocate and free bst_fail.
14680 2017-08-08 Martin Liska <mliska@suse.cz>
14683 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
14684 LABEL_DECLs that can be from a different function.
14686 2017-08-08 Bin Cheng <bin.cheng@arm.com>
14688 PR tree-optimization/81744
14689 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
14690 loop's number of iterations.
14692 2017-08-08 Martin Liska <mliska@suse.cz>
14694 * asan.c: Include header files.
14695 * attribs.c (build_decl_attribute_variant): New function moved
14697 (build_type_attribute_qual_variant): Likewise.
14698 (cmp_attrib_identifiers): Likewise.
14699 (simple_cst_list_equal): Likewise.
14700 (omp_declare_simd_clauses_equal): Likewise.
14701 (attribute_value_equal): Likewise.
14702 (comp_type_attributes): Likewise.
14703 (build_type_attribute_variant): Likewise.
14704 (lookup_ident_attribute): Likewise.
14705 (remove_attribute): Likewise.
14706 (merge_attributes): Likewise.
14707 (merge_type_attributes): Likewise.
14708 (merge_decl_attributes): Likewise.
14709 (merge_dllimport_decl_attributes): Likewise.
14710 (handle_dll_attribute): Likewise.
14711 (attribute_list_equal): Likewise.
14712 (attribute_list_contained): Likewise.
14713 * attribs.h (lookup_attribute): New function moved from tree.[ch].
14714 (lookup_attribute_by_prefix): Likewise.
14715 * bb-reorder.c: Include header files.
14716 * builtins.c: Likewise.
14717 * calls.c: Likewise.
14718 * cfgexpand.c: Likewise.
14719 * cgraph.c: Likewise.
14720 * cgraphunit.c: Likewise.
14721 * convert.c: Likewise.
14722 * dwarf2out.c: Likewise.
14723 * final.c: Likewise.
14724 * fold-const.c: Likewise.
14725 * function.c: Likewise.
14726 * gimple-expr.c: Likewise.
14727 * gimple-fold.c: Likewise.
14728 * gimple-pretty-print.c: Likewise.
14729 * gimple.c: Likewise.
14730 * gimplify.c: Likewise.
14731 * hsa-common.c: Likewise.
14732 * hsa-gen.c: Likewise.
14733 * internal-fn.c: Likewise.
14734 * ipa-chkp.c: Likewise.
14735 * ipa-cp.c: Likewise.
14736 * ipa-devirt.c: Likewise.
14737 * ipa-fnsummary.c: Likewise.
14738 * ipa-inline.c: Likewise.
14739 * ipa-visibility.c: Likewise.
14741 * lto-cgraph.c: Likewise.
14742 * omp-expand.c: Likewise.
14743 * omp-general.c: Likewise.
14744 * omp-low.c: Likewise.
14745 * omp-offload.c: Likewise.
14746 * omp-simd-clone.c: Likewise.
14747 * opts-global.c: Likewise.
14748 * passes.c: Likewise.
14749 * predict.c: Likewise.
14750 * sancov.c: Likewise.
14751 * sanopt.c: Likewise.
14752 * symtab.c: Likewise.
14753 * toplev.c: Likewise.
14754 * trans-mem.c: Likewise.
14755 * tree-chkp.c: Likewise.
14756 * tree-eh.c: Likewise.
14757 * tree-into-ssa.c: Likewise.
14758 * tree-object-size.c: Likewise.
14759 * tree-parloops.c: Likewise.
14760 * tree-profile.c: Likewise.
14761 * tree-ssa-ccp.c: Likewise.
14762 * tree-ssa-live.c: Likewise.
14763 * tree-ssa-loop.c: Likewise.
14764 * tree-ssa-sccvn.c: Likewise.
14765 * tree-ssa-structalias.c: Likewise.
14766 * tree-ssa.c: Likewise.
14767 * tree-streamer-in.c: Likewise.
14768 * tree-vectorizer.c: Likewise.
14769 * tree-vrp.c: Likewise.
14770 * tsan.c: Likewise.
14771 * ubsan.c: Likewise.
14772 * varasm.c: Likewise.
14773 * varpool.c: Likewise.
14774 * tree.c: Remove functions moved to attribs.[ch].
14775 * tree.h: Likewise.
14776 * config/aarch64/aarch64.c: Add attrs.h header file.
14777 * config/alpha/alpha.c: Likewise.
14778 * config/arc/arc.c: Likewise.
14779 * config/arm/arm.c: Likewise.
14780 * config/avr/avr.c: Likewise.
14781 * config/bfin/bfin.c: Likewise.
14782 * config/c6x/c6x.c: Likewise.
14783 * config/cr16/cr16.c: Likewise.
14784 * config/cris/cris.c: Likewise.
14785 * config/darwin.c: Likewise.
14786 * config/epiphany/epiphany.c: Likewise.
14787 * config/fr30/fr30.c: Likewise.
14788 * config/frv/frv.c: Likewise.
14789 * config/ft32/ft32.c: Likewise.
14790 * config/h8300/h8300.c: Likewise.
14791 * config/i386/winnt.c: Likewise.
14792 * config/ia64/ia64.c: Likewise.
14793 * config/iq2000/iq2000.c: Likewise.
14794 * config/lm32/lm32.c: Likewise.
14795 * config/m32c/m32c.c: Likewise.
14796 * config/m32r/m32r.c: Likewise.
14797 * config/m68k/m68k.c: Likewise.
14798 * config/mcore/mcore.c: Likewise.
14799 * config/microblaze/microblaze.c: Likewise.
14800 * config/mips/mips.c: Likewise.
14801 * config/mmix/mmix.c: Likewise.
14802 * config/mn10300/mn10300.c: Likewise.
14803 * config/moxie/moxie.c: Likewise.
14804 * config/msp430/msp430.c: Likewise.
14805 * config/nds32/nds32-isr.c: Likewise.
14806 * config/nds32/nds32.c: Likewise.
14807 * config/nios2/nios2.c: Likewise.
14808 * config/nvptx/nvptx.c: Likewise.
14809 * config/pa/pa.c: Likewise.
14810 * config/pdp11/pdp11.c: Likewise.
14811 * config/powerpcspe/powerpcspe.c: Likewise.
14812 * config/riscv/riscv.c: Likewise.
14813 * config/rl78/rl78.c: Likewise.
14814 * config/rx/rx.c: Likewise.
14815 * config/s390/s390.c: Likewise.
14816 * config/sh/sh.c: Likewise.
14817 * config/sol2.c: Likewise.
14818 * config/sparc/sparc.c: Likewise.
14819 * config/spu/spu.c: Likewise.
14820 * config/stormy16/stormy16.c: Likewise.
14821 * config/tilegx/tilegx.c: Likewise.
14822 * config/tilepro/tilepro.c: Likewise.
14823 * config/v850/v850.c: Likewise.
14824 * config/vax/vax.c: Likewise.
14825 * config/visium/visium.c: Likewise.
14826 * config/xtensa/xtensa.c: Likewise.
14828 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
14831 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
14832 constraints since the -mupper-regs-* switches have been
14834 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
14835 into a vector from a double word element that was extracted from
14836 another vector, and eliminate extra XXPERMDI instructions.
14837 (vsx_concat_<mode>_2): Likewise.
14838 (vsx_concat_<mode>_3): Likewise.
14839 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
14840 concat to allow optimizing inserts from previous extracts.
14842 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
14844 * config/i386/i386.c (ix86_stack_protect_guard): Generate
14845 memory reference to a SSP offset in TLS address space.
14846 (ix86_print_operand) <case '@'>: Remove.
14847 (ix86_print_operand_punct_valid_p): Remove '@' code.
14848 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
14849 UNSPEC_SP_TLS_TEST.
14850 (stack_tls_protect_set_<mode>): Remove.
14851 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
14852 (stack_tls_protect_test_<mode>): Remove.
14853 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
14855 2017-08-07 Olivier Hainque <hainque@adacore.com>
14858 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
14860 2017-08-07 Douglas Rupp <rupp@adacore.com>
14862 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
14863 variable was referenced as multidir in command.
14865 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14868 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
14871 2017-08-07 Martin Liska <mliska@suse.cz>
14873 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
14874 * config/rl78/rl78.c: Add include of attribs.h.
14875 * config/sh/sh.c: Likewise.
14876 * config/v850/v850.c: Likewise.
14878 2017-08-07 Tom de Vries <tom@codesourcery.com>
14880 PR middle-end/78266
14881 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
14883 2017-08-07 Martin Liska <mliska@suse.cz>
14885 * config/mips/mips.c: Include attribs.h.
14887 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
14890 * doc/invoke.texi: Document change in behvaior for -Ofast for
14893 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
14895 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
14897 (aarch64_pop_regs): Likewise.
14898 (aarch64_gen_load_pair): Likewise.
14899 (aarch64_save_callee_saves): Likewise.
14900 (aarch64_restore_callee_saves): Likewise.
14902 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14904 * config/i386/i386.c: Revert the last change.
14906 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14909 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
14911 (ix86_finalize_stack_frame_flags): This. Also clear
14912 frame_pointer_needed if -fno-omit-frame-pointer is used without
14914 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
14915 with ix86_finalize_stack_frame_flags.
14916 (ix86_expand_epilogue): Likewise.
14917 (ix86_expand_split_stack_prologue): Likewise.
14919 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
14922 * config/i386/i386.c (get_builtin_code_for_version): Set priority
14923 to P_AES for Westmere.
14925 2017-08-07 Jonathan Yong <10walls@gmail.com>
14927 * config/i386/mingw.opt (fset-stack-executable): Removed.
14928 * config/i386/cygming.opt (fset-stack-executable): Moved
14930 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
14932 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
14934 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
14936 2017-08-07 Marek Polacek <polacek@redhat.com>
14938 PR middle-end/81737
14939 * fold-const.c (fold_indirect_ref_1): Check type_domain.
14941 2017-08-07 Martin Liska <mliska@suse.cz>
14943 * attribs.h (canonicalize_attr_name): New function.
14944 (cmp_attribs): Move from c-format.c and adjusted.
14945 (is_attribute_p): Moved from tree.h.
14946 * tree-inline.c: Add new includes.
14947 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
14948 (private_is_attribute_p): Remove.
14949 (private_lookup_attribute): Likewise.
14950 (private_lookup_attribute_by_prefix): Simplify.
14951 (remove_attribute): Use is_attribute_p.
14952 * tree.h: Remove removed declarations.
14954 2017-08-07 Jakub Jelinek <jakub@redhat.com>
14956 PR middle-end/81698
14957 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
14958 instead of computing it in the function. Formatting fix.
14959 (expand_case): Don't rely on default_edge being the first edge,
14960 clear it if removing it, pass default_edge to
14961 emit_case_dispatch_table.
14962 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
14965 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
14967 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
14968 insn in the function, emit NOP after the insn.
14970 2017-08-06 Tom de Vries <tom@codesourcery.com>
14972 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
14975 2017-08-06 Tom de Vries <tom@codesourcery.com>
14977 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
14980 2017-08-04 Yury Gribov <tetra2005@gmail.com>
14982 PR tree-optimization/57371
14983 * match.pd: New pattern.
14985 2017-08-04 Marek Polacek <polacek@redhat.com>
14987 PR middle-end/81695
14988 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
14989 perform the computation in offset_int.
14991 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
14993 PR tree-optimization/81136
14994 * tree-vectorizer.h: Include tree-hash-traits.h.
14995 (vec_base_alignments): New typedef.
14996 (vec_info): Add a base_alignments field.
14997 (vect_record_base_alignments): Declare.
14998 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
15000 (DR_IS_CONDITIONAL_IN_STMT): New macro.
15001 (create_data_ref): Add an is_conditional_in_stmt argument.
15002 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
15003 the is_conditional_in_stmt field.
15004 (data_ref_loc): Add an is_conditional_in_stmt field.
15005 (get_references_in_stmt): Set the is_conditional_in_stmt field.
15006 (find_data_references_in_stmt): Update call to create_data_ref.
15007 (graphite_find_data_references_in_stmt): Likewise.
15008 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
15009 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
15010 (vect_record_base_alignment): New function.
15011 (vect_record_base_alignments): Likewise.
15012 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
15013 for nested statements even if we fail to compute a misalignment.
15014 Use pooled base alignments for unconditional references.
15015 (vect_find_same_alignment_drs): Compare base addresses instead
15017 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
15018 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
15020 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15022 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
15023 Add an explicit name for the enum. Use auto_vec for slp_instances
15024 and grouped_stores.
15025 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
15027 (_bb_vec_info): Add a constructor and destructor.
15028 (vinfo_for_stmt): Return NULL for uids of -1 as well.
15029 (destroy_loop_vec_info): Delete.
15030 (vect_destroy_datarefs): Likewise.
15031 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
15032 (vec_info::vec_info): New function.
15033 (vec_info::~vec_info): Likewise.
15034 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
15035 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
15036 destroy_loop_vec_info.
15037 * tree-vect-loop.c (new_loop_vec_info): Replace with...
15038 (_loop_vec_info::_loop_vec_info): ...this.
15039 (destroy_loop_vec_info): Replace with...
15040 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
15041 the stmt_vec_infos. Leave handling of vec_info information to its
15042 destructor. Remove explicit vector releases.
15043 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
15044 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
15045 * tree-vect-slp.c (new_bb_vec_info): Replace with...
15046 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
15047 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
15048 (destroy_bb_vec_info): Replace with...
15049 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
15050 information to its destructor.
15051 (vect_slp_analyze_bb_1): Use new and delete instead of
15052 new_bb_vec_info and destroy_bb_vec_info.
15053 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
15056 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
15058 * tree-data-ref.h (subscript): Add access_fn field.
15059 (data_dependence_relation): Add could_be_independent_p.
15060 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
15061 (same_access_functions): Move to tree-data-ref.c.
15062 * tree-data-ref.c (ref_contains_union_access_p): New function.
15063 (access_fn_component_p): Likewise.
15064 (access_fn_components_comparable_p): Likewise.
15065 (dr_analyze_indices): Add a reference to access_fn_component_p.
15066 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
15068 (constant_access_functions): Likewise.
15069 (add_other_self_distances): Likewise.
15070 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
15071 (initialize_data_dependence_relation): Use XCNEW and remove
15072 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
15073 of access functions that have the same type. Allow the
15074 subsequence to end with different bases in some circumstances.
15075 Record the chosen access functions in SUB_ACCESS_FN.
15076 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
15077 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
15078 (subscript_dependence_tester_1): Likewise dra and drb.
15079 (build_classic_dist_vector): Update calls accordingly.
15080 (subscript_dependence_tester): Likewise.
15081 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
15082 DDR_COULD_BE_INDEPENDENT_P.
15083 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
15084 comp_alias_ddrs instead of may_alias_ddrs.
15085 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
15087 (vect_analyze_data_ref_dependence): Use it if
15088 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
15089 distance vectors if that fails.
15090 (dependence_distance_ge_vf): New function.
15091 (vect_prune_runtime_alias_test_list): Use it. Don't clear
15092 LOOP_VINFO_MAY_ALIAS_DDRS.
15094 2017-08-04 Richard Biener <rguenther@suse.de>
15096 PR middle-end/81705
15097 * fold-const.c (fold_binary_loc): Properly restrict
15098 minus_var0 && minus_var1 case when associating undefined overflow
15101 2017-08-04 Tom de Vries <tom@codesourcery.com>
15103 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
15105 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15107 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
15108 Don't start diagnostic messages with a capital letter.
15109 * config/rs6000/rs6000.c (rs6000_option_override_internal):
15111 (rs6000_invalid_builtin): Likewise.
15112 (rs6000_trampoline_init): Likewise.
15114 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15117 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
15118 after setting changeable df flags.
15120 2017-08-03 Richard Biener <rguenther@suse.de>
15122 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
15123 up if the use is in USE - X.
15125 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
15127 * toplev.c (dumpfile.h): New include.
15128 (internal_error_reentered): New static function. Use it...
15129 (internal_error_function): ...here to handle reentered internal_error.
15131 2017-08-03 Richard Biener <rguenther@suse.de>
15133 PR middle-end/81148
15134 * fold-const.c (split_tree): Add minus_var and minus_con
15135 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
15136 here but always use minus_*.
15137 (associate_trees): Assert we never associate with MINUS_EXPR
15138 and NULL first operand. Do not recurse for PLUS_EXPR operands
15139 when associating as MINUS_EXPR either.
15140 (fold_binary_loc): Track minus_var and minus_con.
15142 2017-08-03 Tom de Vries <tom@codesourcery.com>
15145 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
15146 ACCEL_COMPILER, apply finish_options on
15147 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
15149 2017-08-03 Tom de Vries <tom@codesourcery.com>
15152 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
15153 function_entry_patch_area_size > 0.
15155 2017-08-03 Jakub Jelinek <jakub@redhat.com>
15158 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
15159 instead of 10??LU, perform unit multiplication in wide_int,
15160 don't change alloc_object_size_limit if the limit is larger
15163 PR tree-optimization/81655
15164 PR tree-optimization/81588
15165 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
15166 the case when ranges[i].low and high are 1 for unsigned type with
15169 PR middle-end/81052
15170 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
15171 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
15173 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
15175 * tree-vrp.h: Add include guard.
15177 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
15180 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
15181 (ud2): New insn pattern.
15182 * config/i386/i386.c (ix86_expand_epilogue):
15183 For naked functions, generate ud2 instead of trap insn.
15185 2017-08-02 Marek Polacek <polacek@redhat.com>
15188 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
15190 2017-08-02 Tom de Vries <tom@codesourcery.com>
15191 Cesar Philippidis <cesar@codesourcery.com>
15193 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
15194 Add missing edge probabilities.
15196 2017-08-02 Tamar Christina <tamar.christina@arm.com>
15198 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
15199 Correct endianness.
15201 2017-08-02 Jakub Jelinek <jakub@redhat.com>
15203 PR middle-end/79499
15204 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
15205 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
15206 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
15207 prologue_seq sequences - if any.
15209 2017-08-02 Richard Biener <rguenther@suse.de>
15211 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
15212 via vectors if supported, integer extracts via punning if supported
15213 or otherwise vector extracts.
15215 2017-08-02 Richard Biener <rguenther@suse.de>
15217 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
15219 (bitmap_insert_into_set): ... this.
15221 2017-08-02 Richard Biener <rguenther@suse.de>
15223 PR tree-optimization/81633
15225 2015-08-17 Alan Hayward <alan.hayward@arm.com>
15227 PR tree-optimization/71752
15228 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
15230 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
15232 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
15233 (machine_function::call_ms2sysv_pad_out): Remove field.
15234 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
15235 (ix86_compute_frame_layout): Likewise.
15237 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
15240 * config/i386/i386.c (ix86_set_func_type): Disallow naked
15241 attribute with interrupt attribute.
15243 2017-08-01 Andrew Pinski <apinski@cavium.com>
15245 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
15246 BIT_INSERT_EXPR's operand 1
15247 to see if the types precision matches.
15249 2017-08-01 Martin Liska <mliska@suse.cz>
15251 PR middle-end/70140
15252 * builtins.c (expand_builtin_memcpy_args): Remove.
15253 (expand_builtin_memcpy): Call newly added function
15254 expand_builtin_memory_copy_args.
15255 (expand_builtin_memcpy_with_bounds): Likewise.
15256 (expand_builtin_mempcpy): Remove last argument.
15257 (expand_builtin_mempcpy_with_bounds): Likewise.
15258 (expand_builtin_memory_copy_args): New function created from
15259 expand_builtin_mempcpy_args with small modifications.
15260 (expand_builtin_mempcpy_args): Remove.
15261 (expand_builtin_stpcpy): Remove unused argument.
15262 (expand_builtin): Likewise.
15263 (expand_builtin_with_bounds): Likewise.
15265 2017-08-01 Martin Liska <mliska@suse.cz>
15268 Make mempcpy more optimal (PR middle-end/70140).
15270 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15273 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
15274 __builtin_vec_cmpne verify both arguments are compatible vectors
15275 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
15276 verify arg1_type is a pointer or array type. For __builtin_vec_st,
15277 move computation of aligned to after checking the argument types.
15281 * config/rs6000/vsx.md (vextract_fp_from_shorth,
15282 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
15285 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
15286 Doug Rupp <rupp@adacore.com>
15287 Olivier Hainque <hainque@adacore.com>
15289 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
15290 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
15292 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
15293 for TARGET_OS_CPP_BUILTIN.
15294 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
15295 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
15297 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
15298 passing required abi options to the assembler for EABI configurations.
15299 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
15300 of .text.hot and .text.unlikely sections for kernel modules when
15301 using ARM style exceptions.
15302 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
15303 options. Add EXTRA_CC1_SPEC.
15304 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
15306 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
15308 (ARM_TARGET2_DWARF_FORMAT): Define.
15309 * config/arm/t-vxworks: Adjust multilib control to removal of the
15310 Diab command line options.
15312 2017-08-01 Martin Liska <mliska@suse.cz>
15314 PR gcov-profile/81561
15315 * gcov.c (unblock): Make unblocking safe as we need to preserve
15316 index correspondence of blocks and block_lists.
15318 2017-08-01 Richard Biener <rguenther@suse.de>
15320 PR tree-optimization/81181
15321 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
15322 (compute_antic): ... end of iteration here.
15324 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
15326 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
15327 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
15328 (ftree-slp-vectorize): Likewise.
15329 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
15330 can no longer be set independent of flag_tree_loop_vectorize.
15331 * omp-general.c (emp_max_vf): Likewise.
15332 * opts.c (enable_fdo_optimizations): Remove references to
15333 flag_tree_vectorize, these are now implicit.
15334 (common_handle_option): Remove handling for OPT_ftree_vectorize,
15335 and leave it for the options machinery.
15337 2017-08-01 Martin Liska <mliska@suse.cz>
15339 PR middle-end/70140
15340 * builtins.c (expand_builtin_memcpy_args): Remove.
15341 (expand_builtin_memcpy): Call newly added function
15342 expand_builtin_memory_copy_args.
15343 (expand_builtin_memcpy_with_bounds): Likewise.
15344 (expand_builtin_mempcpy): Remove last argument.
15345 (expand_builtin_mempcpy_with_bounds): Likewise.
15346 (expand_builtin_memory_copy_args): New function created from
15347 expand_builtin_mempcpy_args with small modifications.
15348 (expand_builtin_mempcpy_args): Remove.
15349 (expand_builtin_stpcpy): Remove unused argument.
15350 (expand_builtin): Likewise.
15351 (expand_builtin_with_bounds): Likewise.
15353 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15356 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
15357 print "ds:" only for immediates in generic address space.
15359 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
15362 * config/i386/i386.c (ix86_funciton_naked): New prototype.
15363 (ix86_function_ok_for_sibcall): Return false for naked functions.
15365 2017-08-01 Richard Biener <rguenther@suse.de>
15367 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
15368 (compute_antic): Seed worklist with exit block predecessors.
15369 * cfganal.c (dfs_find_deadend): For a cycle return the source
15370 of the edge closing it.
15372 2017-08-01 Tamar Christina <tamar.christina@arm.com>
15374 * config/aarch64/aarch64.c
15375 (aarch64_can_const_movi_rtx_p): Move 0 check.
15377 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15379 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
15380 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
15382 * match.pd: Ditto in address comparison pattern.
15384 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15386 PR tree-optimization/81627
15387 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
15388 closed ssa form for store-store chain.
15390 2017-08-01 Bin Cheng <bin.cheng@arm.com>
15392 PR tree-optimization/81620
15393 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
15394 for store-store chain.
15396 2017-08-01 Jakub Jelinek <jakub@redhat.com>
15398 PR tree-optimization/81588
15399 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
15400 ranges[i].in_p, invert comparison code ccode. For >/>=,
15401 swap rhs1 and rhs2 and comparison code unconditionally,
15402 for </<= don't do that. Don't swap rhs1/rhs2 again if
15403 ranges[i].in_p, instead invert comparison code ccode if
15404 opcode or oe->rank is BIT_IOR_EXPR.
15407 * optabs.def (vec_extract_optab, vec_init_optab): Change from
15408 a direct optab to conversion optab.
15409 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
15410 with GET_MODE_INNER as last argument instead of optab_handler.
15411 * expmed.c (extract_bit_field_1): Likewise. Use vector from
15412 vector extraction if possible and optab is available.
15413 * expr.c (store_constructor): Use convert_optab_handler instead
15414 of optab_handler. Use vector initialization from smaller
15415 vectors if possible and optab is available.
15416 * tree-vect-stmts.c (vectorizable_load): Likewise.
15417 * doc/md.texi (vec_extract, vec_init): Document that the optabs
15418 now have two modes.
15419 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
15420 of vec_init from half-sized vectors with the same element mode.
15421 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
15422 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
15423 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
15424 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
15425 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
15426 after mode in gen_vec_extract* calls.
15427 (vec_extract<mode>): Renamed to ...
15428 (vec_extract<mode><ssescalarmodelower>): ... this.
15429 (vec_extract<mode><ssehalfvecmodelower>): New expander.
15430 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
15431 element mode after mode in gen_vec_init* calls.
15432 (VEC_INIT_HALF_MODE): New mode iterator.
15433 (vec_init<mode>): Renamed to ...
15434 (vec_init<mode><ssescalarmodelower>): ... this.
15435 (vec_init<mode><ssehalfvecmodelower>): New expander.
15436 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
15437 (vec_extractv2sfsf): ... this.
15438 (vec_initv2sf): Renamed to ...
15439 (vec_initv2sfsf): ... this.
15440 (vec_extractv2si): Renamed to ...
15441 (vec_extractv2sisi): ... this.
15442 (vec_initv2si): Renamed to ...
15443 (vec_initv2sisi): ... this.
15444 (vec_extractv4hi): Renamed to ...
15445 (vec_extractv4hihi): ... this.
15446 (vec_initv4hi): Renamed to ...
15447 (vec_initv4hihi): ... this.
15448 (vec_extractv8qi): Renamed to ...
15449 (vec_extractv8qiqi): ... this.
15450 (vec_initv8qi): Renamed to ...
15451 (vec_initv8qiqi): ... this.
15452 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
15453 (vec_init<mode>): Renamed to ...
15454 (vec_init<mode><VEC_base_l>): ... this.
15455 (vec_extract<mode>): Renamed to ...
15456 (vec_extract<mode><VEC_base_l>): ... this.
15457 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
15458 (vec_initv2sfsf): ... this.
15459 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
15460 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15461 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
15462 element mode after mode in gen_vec_init* calls.
15463 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
15464 (vec_init<mode><Vel>): ... this.
15465 (vec_extract<mode>): Renamed to ...
15466 (vec_extract<mode><Vel>): ... this.
15467 * config/aarch64/iterators.md (Vel): New mode attribute.
15468 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
15469 Add element mode after mode in gen_vec_extract* calls.
15470 * config/s390/vector.md (non_vec_l): New mode attribute.
15471 (vec_extract<mode>): Renamed to ...
15472 (vec_extract<mode><non_vec_l>): ... this.
15473 (vec_init<mode>): Renamed to ...
15474 (vec_init<mode><non_vec_l>): ... this.
15475 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
15476 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
15478 * config/arm/iterators.md (V_elem_l): New mode attribute.
15479 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
15480 (vec_extract<mode><V_elem_l>): ... this.
15481 (vec_extractv2di): Renamed to ...
15482 (vec_extractv2didi): ... this.
15483 (vec_init<mode>): Renamed to ...
15484 (vec_init<mode><V_elem_l>): ... this.
15485 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
15486 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
15487 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
15488 Add element mode after gen_vec_extract* calls.
15489 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
15490 (vec_init<mode><unitmode>): ... this.
15491 (vec_extract<mode>): Renamed to ...
15492 (vec_extract<mode><unitmode>): ... this.
15493 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
15494 (vec_init<mode><unitmode>): ... this.
15495 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
15496 (vec_initv2sfsf): ... this.
15497 (vec_extractv2sf): Renamed to ...
15498 (vec_extractv2sfsf): ... this.
15499 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
15500 Add element mode after gen_vec_extract* calls.
15501 * config/mips/mips.md (unitmode): New mode iterator.
15502 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
15503 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
15504 * config/spu/spu.md (inner_l): New mode attribute.
15505 (vec_init<mode>): Renamed to ...
15506 (vec_init<mode><inner_l>): ... this.
15507 (vec_extract<mode>): Renamed to ...
15508 (vec_extract<mode><inner_l>): ... this.
15509 * config/sparc/sparc.md (veltmode): New mode iterator.
15510 (vec_init<VMALL:mode>): Renamed to ...
15511 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
15512 * config/ia64/vect.md (vec_initv2si): Renamed to ...
15513 (vec_initv2sisi): ... this.
15514 (vec_initv2sf): Renamed to ...
15515 (vec_initv2sfsf): ... this.
15516 (vec_extractv2sf): Renamed to ...
15517 (vec_extractv2sfsf): ... this.
15518 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
15519 (vec_init<mode>): Renamed to ...
15520 (vec_init<mode><VEC_base_l>): ... this.
15521 (vec_extract<mode>): Renamed to ...
15522 (vec_extract<mode><VEC_base_l>): ... this.
15523 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
15524 (vec_initv2sfsf): ... this.
15525 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
15526 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
15527 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
15528 gen_vec_init* calls.
15530 2017-08-01 Richard Biener <rguenther@suse.de>
15532 PR tree-optimization/81297
15533 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
15534 TREE_OVERFLOW from INTEGER_CSTs.
15536 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
15538 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
15540 2017-07-31 Carl Love <cel@us.ibm.com>
15542 * config/rs6000/rs6000-c: Add support for built-in functions
15543 vector signed char vec_xl_be (signed long long, signed char *);
15544 vector unsigned char vec_xl_be (signed long long, unsigned char *);
15545 vector signed int vec_xl_be (signed long long, signed int *);
15546 vector unsigned int vec_xl_be (signed long long, unsigned int *);
15547 vector signed long long vec_xl_be (signed long long, signed long long *);
15548 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
15549 vector signed short vec_xl_be (signed long long, signed short *);
15550 vector unsigned short vec_xl_be (signed long long, unsigned short *);
15551 vector double vec_xl_be (signed long long, double *);
15552 vector float vec_xl_be (signed long long, float *);
15553 * config/rs6000/altivec.h (vec_xl_be): Add #define.
15554 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
15555 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
15557 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
15558 (altivec_expand_builtin): Add switch statement to call
15559 altivec_expand_xl_be for each builtin.
15560 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
15561 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
15562 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
15563 __builtin_vsx_le_be_v16qi.
15564 * doc/extend.texi: Update the built-in documentation file for the
15565 new built-in functions.
15567 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15570 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
15572 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
15574 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15576 * config.gcc: Add z14.
15577 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
15578 CPU model numbers for z13s and z14.
15579 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
15581 * config/s390/s390-opts.h (enum processor_type): Rename
15582 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15583 * config/s390/s390.c (processor_table): Add field for CPU name to
15584 be passed to Binutils.
15585 (s390_asm_output_machine_for_arch): Use the new field in
15586 processor_table for Binutils.
15587 (s390_expand_builtin): Replace arch12 with z14.
15588 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
15589 (s390_get_sched_attrmask): Likewise.
15590 (s390_get_unit_mask): Likewise.
15591 * config/s390/s390.opt: Add z14 to processor_type enum.
15593 2017-07-31 Martin Jambor <mjambor@suse.cz>
15596 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
15597 regardless of optimization level.
15599 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
15600 Martin Liska <mliska@suse.cz>
15602 * predict.def: Remove old comment and adjust probability.
15603 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
15604 PREDICT statements.
15606 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
15609 * config/i386/i386.c (ix86_function_naked): New function.
15610 (ix86_can_use_return_insn_p): Return false for naked functions.
15611 (ix86_expand_prologue): Skip prologue for naked functions.
15612 (ix86_expand_epilogue): Skip epilogue for naked functions
15613 and emit trap instruction.
15614 (ix86_warn_func_return): New function.
15615 (ix86_attribute_table): Add "naked" attribute specification.
15616 (TARGET_WARN_FUNC_RETURN): Define.
15617 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
15619 2017-07-31 Martin Liska <mliska@suse.cz>
15621 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
15622 (dump_gimple_bb_header): Always dump BB info.
15623 (pp_cfg_jump): Do not append info about BB when dumping a jump.
15625 2017-07-31 Martin Liska <mliska@suse.cz>
15628 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
15629 also with current_function_decl non-null equality.
15631 2017-07-31 Jakub Jelinek <jakub@redhat.com>
15634 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
15635 change type to the element type, instead add eltype variable and
15636 use it where we are interested in the element type.
15638 PR tree-optimization/81603
15639 * ipa-polymorphic-call.c
15640 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
15641 offset arithmetic in offset_int, bail out if the resulting bit offset
15642 doesn't fit into shwi.
15644 2017-07-31 Martin Liska <mliska@suse.cz>
15646 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
15647 (gimplify_save_expr): Fix comment.
15649 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
15652 * config/i386/i386.c (ix86_function_arg): Update arguments for
15654 (ix86_compute_frame_layout): Set the initial stack offset to
15655 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
15656 INCOMING_FRAME_SP_OFFSET.
15657 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
15658 stack before exception handler returns.
15659 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
15660 the 'ERROR_CODE' for exception handler.
15662 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
15664 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
15665 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
15666 (ASM_OUTPUT_REG_POP): Ditto.
15667 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
15668 instead of asm_fprintf to output pure string.
15670 2017-07-29 Jakub Jelinek <jakub@redhat.com>
15672 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
15673 to imported_module_or_decl hook.
15674 (debug_nothing_tree_tree_tree_bool): Remove.
15675 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
15676 * debug.c (do_nothing_debug_hooks): Use
15677 debug_nothing_tree_tree_tree_bool_bool instead of
15678 debug_nothing_tree_tree_tree_bool.
15679 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
15680 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
15681 * sdbout.c (sdb_debug_hooks): Likewise.
15682 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
15683 (gen_namespace_die): Add DW_AT_export_symbols attribute if
15685 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
15686 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
15687 attribute, don't add anything.
15689 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15691 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
15692 (fold_build2_stat_loc): Likewise.
15693 (fold_build3_stat_loc): Likewise.
15694 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
15695 (fold_build1_loc): Remove macro.
15696 (fold_build2_loc): Likewise.
15697 (fold_build3_loc): Likewise.
15699 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15701 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
15702 (gimple_build_debug_bind_source_stat): Likewise.
15703 * gimple.h (gimple_build_debug_bind): Remove macro.
15704 (gimple_build_debug_bind_source): Likewise.
15706 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15708 * bitmap.c (bitmap_alloc): Adjust.
15709 (bitmap_gc_alloc): Likewise.
15710 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
15712 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15714 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
15715 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
15716 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
15717 (bitmap_gc_alloc_stat): Likewise.
15718 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
15720 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15722 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
15723 * rtl.h (shallow_copy_rtx): Remove macro.
15725 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15727 * emit-rtl.c (gen_raw_REG): Adjust.
15728 * gengenrtl.c (gendef): Likewise.
15729 * rtl.c (rtx_alloc_stat): Remove _stat from name.
15730 * rtl.h (rtx_alloc): Remove macro.
15732 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15734 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
15735 (build_tree_list_stat): Likewise.
15736 * tree.h (build_tree_list): Remove macro.
15737 (build_tree_list_vec): Likewise.
15739 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15741 * tree.c (make_vector_stat): Remove _stat from name.
15742 (build_vector_stat): Likewise.
15743 * tree.h (make_vector_stat): Remove macro.
15744 (build_vector_stat): Likewise.
15746 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15748 * tree.h (build_var_debug_value): Remove prototype.
15750 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15752 * tree.c (tree_cons_stat): Remove _stat from name.
15753 * tree.h (tree_cons): Remove macro.
15755 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15757 * tree.c (build_vl_exp_stat): Remove _stat from name.
15758 * tree.h (build_vl_exp): Remove macro.
15760 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15762 * tree.c (build_decl_stat): Remove _stat from name.
15763 * tree.h (build_decl): Remove macro.
15765 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15767 * gimple.c (gimple_build_with_ops_stat): Adjust.
15768 (gimple_alloc_stat): Remove _stat from name.
15769 * gimple.h (gimple_alloc): Remove macro.
15771 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15773 * tree.c (make_tree_vec_stat): Remove _stat from name.
15774 (grow_tree_vec_stat): Likewise.
15775 * tree.h (make_tree_vec_stat): Adjust prototype.
15776 (grow_tree_vec_stat): Likewise.
15777 (make_tree_vec): Remove macro.
15778 (grow_tree_vec): Likewise.
15780 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15782 * fold-const.c (fold_build1_stat_loc): Adjust.
15783 (fold_build2_stat_loc): Likewise.
15784 (fold_build3_stat_loc): Likewise.
15785 * tree.c (build0_stat): Remove _stat from name.
15786 (build1_stat): Likewise.
15787 (build2_stat): Likewise.
15788 (build3_stat): Likewise.
15789 (build4_stat): Likewise.
15790 (build5_stat): Likewise.
15791 * tree.h (build1_loc): Remove macro, and rename _stat function
15793 (build2_loc): Likewise.
15794 (build3_loc): Likewise.
15795 (build4_loc): Likewise.
15796 (build5_loc): Likewise.
15798 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15800 * tree.c (make_int_cst_stat): Remove _stat from name.
15801 * tree.h (make_int_cst_stat): Adjust prototype.
15802 (make_int_cst): Remove macro.
15804 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15806 * tree.c (make_tre_binfo_stat): Remove _stat from name.
15807 * tree.h (make_tree_binfo_stat): Adjust prototype.
15808 (make_tree_binfo): Remove.
15810 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15812 * tree.c (copy_node_stat): Rename to copy_node.
15813 (build_distinct_type_copy): Adjust.
15814 * tree.h (copy_node_stat): Adjust prototype.
15815 (copy_node): Remove macro.
15817 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
15819 * tree.c (make_node_stat): rename to make_node.
15820 (build_tree_list_stat): Adjust.
15821 (build0_stat): Likewise.
15822 (build2_stat): Likewise.
15823 (build3_stat): Likewise.
15824 (build4_stat): Likewise.
15825 (build5_stat): Likewise.
15826 (build_decl_stat): Likewise.
15827 * tree.h (make_node_stat): Adjust prototype.
15828 (make_node): remove macro.
15830 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
15832 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
15833 (PPC_FEATURE2_SCV): Likewise.
15834 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
15836 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15838 * config/aarch64/aarch64.c
15839 (aarch64_internal_mov_immediate): Add new special pattern.
15840 * config/aarch64/aarch64.md (*movdi_aarch64):
15841 Add reg/32bit const mov case.
15843 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15844 Richard Sandiford <richard.sandiford@linaro.org>
15846 * config/aarch64/aarch64.md (mov<mode>): Generalize.
15847 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
15848 Add integer and movi cases.
15849 (movi-split-hf-df-sf split, fp16): New.
15850 (enabled): Added TARGET_FP_F16INST.
15851 * config/aarch64/iterators.md (GPF_HF): New.
15852 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
15854 2017-07-28 Tamar Christina <tamar.christina@arm.com>
15856 * config/aarch64/aarch64.c
15857 (aarch64_simd_container_mode): Add prototype.
15858 (aarch64_expand_mov_immediate): Add HI support.
15859 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
15860 (aarch64_can_const_movi_rtx_p): New.
15861 (aarch64_preferred_reload_class):
15862 Remove restrictions of using FP registers for certain SIMD operations.
15863 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
15864 (aarch64_valid_floating_const): Add integer move validation.
15865 (aarch64_simd_imm_scalar_p): Remove.
15866 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
15867 (aarch64_legitimate_constant_p): Expand list of supported cases.
15868 * config/aarch64/aarch64-protos.h
15869 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
15870 (aarch64_reinterpret_float_as_int): New.
15871 (aarch64_simd_imm_scalar_p): Remove.
15872 * config/aarch64/constraints.md (Uvi): New.
15873 (Dd): Split into Ds and new Dd.
15874 * config/aarch64/aarch64.md (*movsi_aarch64):
15876 (*movdi_aarch64): Add SIMD mov case.
15878 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15880 * tree-predcom.c: (struct chain): Handle store-store chain in which
15881 stores for elimination only store loop invariant values.
15882 (execute_pred_commoning_chain): Ditto.
15883 (prepare_initializers_chain_store_elim): Ditto.
15884 (prepare_finalizers): Ditto.
15885 (is_inv_store_elimination_chain): New function.
15886 (initialize_root_vars_store_elim_1): New function.
15888 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15890 * tree-predcom.c: Revise general description of the pass.
15891 (enum chain_type): New enum type for store elimination.
15892 (struct chain): New field supporting store elimination.
15893 (struct component): Ditto.
15894 (dump_chain): Dump store-stores chain.
15895 (release_chain): Release resources.
15896 (split_data_refs_to_components): Compute and create component
15897 contains only stores for elimination.
15898 (get_chain_last_ref_at): New function.
15899 (make_invariant_chain): Initialization.
15900 (make_rooted_chain): Specify chain type in parameter and record it.
15901 (add_looparound_copies): Skip for store-stores chain.
15902 (determine_roots_comp): Compute type of chain and pass it to
15904 (initialize_root_vars_store_elim_2): New function.
15905 (finalize_eliminated_stores): New function.
15906 (remove_stmt): Handle store for elimination.
15907 (execute_pred_commoning_chain): Execute predictive commoning on
15908 store-store chains.
15909 (determine_unroll_factor): Skip unroll for store-stores chain.
15910 (prepare_initializers_chain_store_elim): New function.
15911 (prepare_initializers_chain): Hanlde store-store chain.
15912 (prepare_finalizers_chain, prepare_finalizers): New function.
15913 (tree_predictive_commoning_loop): Return integer value indicating
15914 if loop is unrolled or lcssa form is corrupted.
15915 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
15917 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15919 * tree-predcom.c (initialize_root): Delete.
15920 (execute_pred_commoning_chain): Initialize root vars and replace
15921 reference of non-combined chain directly, rather than call above
15924 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15926 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
15927 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
15929 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15931 * tree-predcom.c (struct chain): New field init_seq.
15932 (release_chain): Release init_seq.
15933 (prepare_initializers_chain): Record intialization stmts in above
15935 (insert_init_seqs): New function.
15936 (tree_predictive_commoning_loop): Call insert_init_seqs.
15938 2017-07-28 Bin Cheng <bin.cheng@arm.com>
15940 * tree-predcom.c (determine_roots_comp): Skip trivial components.
15942 2017-07-28 Richard Biener <rguenther@suse.de>
15944 * match.pd: Remove superfluous :c.
15945 * genmatch.c (simplify::id): Add member.
15946 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
15948 (current_id): New global.
15949 (dt_node::parent): Move from ...
15950 (dt_operand::parent): ... here. Add for_id member.
15951 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
15952 (decision_tree::find_node): Relax order requirement when
15953 merging DT_TRUE nodes to ones inbetween the current simplify
15954 and the one we try to merge with. Add diagnostic whenever
15955 we need to enforce pattern order by not merging.
15956 (decision_tree::insert): Set current_id.
15957 (decision_tree::print_node): Dump parent node and for_id.
15958 (parser::last_id): Add member.
15959 (parser::push_simplify): Assign unique id.
15960 (parser::parser): Initialize last_id.
15962 2017-07-28 Martin Liska <mliska@suse.cz>
15965 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
15966 gimple_build_debug_bind.
15968 2017-07-28 Richard Biener <rguenther@suse.de>
15970 PR tree-optimization/81502
15971 * match.pd: Add pattern combining BIT_INSERT_EXPR with
15973 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
15975 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
15976 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
15977 for BIT_FIELD_REF args.
15978 * fold-const.c (make_bit_field_ref): Likewise.
15979 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
15981 2017-07-28 Jakub Jelinek <jakub@redhat.com>
15984 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
15985 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
15986 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
15987 Or it into SANITIZER_UNDEFINED.
15988 * ubsan.c: Include gimple-fold.h and varasm.h.
15989 (ubsan_expand_ptr_ifn): New function.
15990 (instrument_pointer_overflow): New function.
15991 (maybe_instrument_pointer_overflow): New function.
15992 (instrument_object_size): Formatting fix.
15993 (pass_ubsan::execute): Call instrument_pointer_overflow
15994 and maybe_instrument_pointer_overflow.
15995 * internal-fn.c (expand_UBSAN_PTR): New function.
15996 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
15997 * sanitizer.def (__ubsan_handle_pointer_overflow,
15998 __ubsan_handle_pointer_overflow_abort): New builtins.
15999 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
16000 * internal-fn.def (UBSAN_PTR): New internal function.
16001 * opts.c (sanitizer_opts): Add pointer-overflow.
16002 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
16003 * fold-const.c (build_range_check): Compute pointer range check in
16004 integral type if pointer arithmetics would be needed. Formatting
16007 2017-07-28 Martin Liska <mliska@suse.cz>
16010 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
16011 parameters that are of a variable-length.
16013 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
16015 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
16017 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
16018 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
16019 (CRT_CALL_STATIC_FUNCTION): Likewise.
16020 (ASM_DEFAULT_SPEC): New define.
16021 (ASM_SPEC32): Likewise.
16022 (ASM_SPEC64): Likewise.
16023 (ASM_SPEC_COMMON): Likewise.
16024 (ASM_SPEC): Likewise.
16025 (INVALID_64BIT): Likewise.
16026 (LINK_OS_DEFAULT_SPEC): Likewise.
16027 (LINK_OS_SPEC32): Likewise.
16028 (LINK_OS_SPEC64): Likewise.
16029 (POWERPC_LINUX): Likewise.
16030 (PTRDIFF_TYPE): Likewise.
16031 (RESTORE_FP_PREFIX): Likewise.
16032 (RESTORE_FP_SUFFIX): Likewise.
16033 (SAVE_FP_PREFIX): Likewise.
16034 (SAVE_FP_SUFFIX): Likewise.
16035 (SIZE_TYPE): Likewise.
16036 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
16037 (TARGET_64BIT): Likewise.
16038 (TARGET_64BIT): Likewise.
16039 (TARGET_AIX): Likewise.
16040 (WCHAR_TYPE_SIZE): Likewise.
16041 (WCHAR_TYPE): Undefine.
16042 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
16043 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
16044 (CPP_OS_RTEMS_SPEC): Delete.
16045 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
16046 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
16048 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
16050 2017-07-28 Jakub Jelinek <jakub@redhat.com>
16052 PR tree-optimization/81578
16053 * tree-parloops.c (build_new_reduction): Bail out if
16054 reduction_code isn't one of the standard OpenMP reductions.
16055 Move the details printing after that decision.
16057 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16059 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
16060 related to reload_in_progress.
16061 (splat_input_operand): Likewise.
16062 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
16064 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
16066 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
16067 (TARGET_INSTANTIATE_DECLS): Likewise.
16068 (legitimate_indexed_address_p): Delete reload_in_progress code.
16069 (rs6000_debug_legitimate_address_p): Likewise.
16070 (rs6000_eliminate_indexed_memrefs): Likewise.
16071 (rs6000_emit_le_vsx_store): Likewise.
16072 (rs6000_emit_move_si_sf_subreg): Likewise.
16073 (rs6000_emit_move): Likewise.
16074 (register_to_reg_type): Likewise.
16075 (rs6000_pre_atomic_barrier): Likewise.
16076 (rs6000_machopic_legitimize_pic_address): Likewise.
16077 (rs6000_allocate_stack_temp): Likewise.
16078 (rs6000_address_for_fpconvert): Likewise.
16079 (rs6000_address_for_altivec): Likewise.
16080 (rs6000_secondary_memory_needed_rtx): Delete function.
16081 (rs6000_check_sdmode): Likewise.
16082 (rs6000_alloc_sdmode_stack_slot): Likewise.
16083 (rs6000_instantiate_decls): Likewise.
16084 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
16085 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
16086 Delete reload_in_progress.
16087 (*vec_reload_and_plus_<mptrsize>): Likewise.
16088 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
16089 (vsx_div_v2di): Likewise.
16090 (vsx_udiv_v2di): Likewise.
16092 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16094 * config/rs6000/rs6000.opt (mlra): Replace with stub.
16095 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
16096 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
16097 (rs6000_debug_reg_global): Delete print of LRA status.
16098 (rs6000_option_override_internal): Delete dead LRA related code.
16099 (rs6000_lra_p): Delete function.
16100 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
16102 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
16104 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
16105 * config/riscv/rtems.h: New file.
16107 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16108 Sudakshina Das <sudi.das@arm.com>
16110 * config/aarch64/aarch64.md
16111 (define_split for and<mode>3nr_compare): Move
16112 non aarch64_logical_operand to a register.
16113 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
16114 register immediate operand to a register.
16115 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
16117 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
16119 PR middle-end/81564
16120 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
16122 2017-07-27 Richard Biener <rguenther@suse.de>
16124 PR tree-optimization/81573
16125 PR tree-optimization/81494
16126 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
16127 multi defuse cycle case.
16129 2017-07-27 Richard Biener <rguenther@suse.de>
16131 PR tree-optimization/81571
16132 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
16135 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
16137 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
16138 earlier and only if MASK_FPU is set. Adjust formatting.
16140 2017-07-27 Martin Liska <mliska@suse.cz>
16142 * opt-functions.awk: Add validation of value of Init.
16143 * optc-gen.awk: Pass new argument.
16145 2017-07-27 Martin Liska <mliska@suse.cz>
16147 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
16148 Fix wrong condition.
16150 2017-07-27 Martin Liska <mliska@suse.cz>
16152 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
16153 BBs and edges seen by autoFDO.
16155 2017-07-27 Richard Biener <rguenther@suse.de>
16157 PR tree-optimization/81502
16158 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
16159 with incompatible but same sized type.
16160 (execute_update_addresses_taken): Likewise.
16162 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
16164 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
16165 flag_tree_loop_vectorize rather than flag_tree_vectorize.
16167 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16170 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
16171 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
16172 Change s_operand to memory_operand.
16174 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
16176 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
16177 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
16178 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
16179 Emit instructions rather than returning an expression. Handle TFmode
16180 and KFmode by casting to TImode.
16181 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
16182 (rs6000_emit_le_vsx_store): Likewise.
16183 * config/rs6000/vsx.md (VSX_TI): New iterator.
16184 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
16185 (*vsx_le_undo_permute_<mode>): Likewise.
16186 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
16187 emit the split sequence.
16188 (*vsx_le_perm_store_<mode>): Likewise.
16190 2017-07-27 Jakub Jelinek <jakub@redhat.com>
16192 PR tree-optimization/81555
16193 PR tree-optimization/81556
16194 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
16195 if true, force CHANGED for the recursive invocation.
16196 (reassociate_bb): Remember original length of ops array, pass
16197 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
16199 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
16200 attributes for noipa attribute. For naked attribute use
16201 lookup_attribute first before lookup_attribute_spec.
16202 * final.c (rest_of_handle_final): Disable IPA RA for functions with
16204 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
16205 for functions with noipa attribute.
16206 (cgraph_externally_visible_p): Return true for functions with noipa
16208 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
16209 for functions with noipa attribute.
16210 * doc/extend.texi: Document noipa function attribute.
16211 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
16212 also for functions with noipa attribute.
16213 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
16215 2017-07-26 Andrew Pinski <apinski@cavium.com>
16217 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
16218 vec_unalign_load_cost and vec_unalign_store_cost.
16220 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
16222 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
16223 -mvsx-small-integer option.
16224 (ISA_3_0_MASKS_IEEE): Likewise.
16225 (OTHER_VSX_VECTOR_MASKS): Likewise.
16226 (POWERPC_MASKS): Likewise.
16227 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
16228 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
16229 code, only testing for DImode being allowed in non-VSX floating
16231 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
16232 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
16234 (rs6000_option_override_internal): Delete -mvsx-small-integer.
16235 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
16236 TARGET_P8_VECTOR test.
16237 (rs6000_secondary_reload_simple_move): Likewise.
16238 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
16239 since TARGET_P9_VECTOR was already tested.
16240 (rs6000_opt_masks): Remove -mvsx-small-integer.
16241 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
16242 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16244 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
16245 test for TARGET_VEXTRACTUB was used, and that uses
16247 (p9 extract splitter): Likewise.
16248 (vsx_extract_<mode>_di_p9): Likewise.
16249 (vsx_extract_<mode>_store_p9): Likewise.
16250 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
16251 for TARGET_P9_VECTOR was used. Delete code that is now dead with
16252 the elimination of TARGET_VSX_SMALL_INTEGER.
16253 (vsx_extract_<mode>_p8): Likewise.
16254 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
16255 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
16256 (vsx_set_<mode>_p9): Likewise.
16257 (vsx_set_v4sf_p9): Likewise.
16258 (vsx_set_v4sf_p9_zero): Likewise.
16259 (vsx_insert_extract_v4sf_p9): Likewise.
16260 (vsx_insert_extract_v4sf_p9_2): Likewise.
16261 * config/rs6000/rs6000.md (sign extend splitter): Change
16262 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
16263 (floatsi<mode>2_lfiwax_mem): Likewise.
16264 (floatunssi<mode>2_lfiwzx_mem): Likewise.
16265 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
16266 since a test for TARGET_P9_VECTOR was used.
16267 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16268 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
16269 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16270 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16271 TARGET_P8_VECTOR test.
16272 (fix_trunc<mode>si2_stfiwx): Likewise.
16273 (fix_trunc<mode>si2_internal): Likewise.
16274 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
16275 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16277 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16278 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
16279 TARGET_P8_VECTOR test.
16280 (fixuns_trunc<mode>si2_stfiwx): Likewise.
16281 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
16282 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
16284 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
16285 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
16286 since a test for TARGET_P9_VECTOR was used.
16287 (splitter for loading small constants): Likewise.
16289 2017-07-26 Andrew Pinski <apinski@cavium.com>
16291 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
16294 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
16297 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
16298 (fp_valid_at): Likewise.
16300 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16302 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
16303 (qdf24xx_addrcost_table): Likewise.
16304 (cortexa57_tunings): Update to use generic_branch_cost.
16305 (cortexa72_tunings): Likewise.
16306 (cortexa73_tunings): Likewise.
16307 (qdf24xx_tunings): Likewise.
16309 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
16311 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
16312 (thunderx2t99_branch_cost): Likewise.
16313 (cortexa35_tunings): Update to use generic_branch_cost.
16314 (cortexa53_tunings): Likewise.
16315 (cortexa57_tunings): Likewise.
16316 (cortexa72_tunings): Likewise.
16317 (cortexa73_tunings): Likewise.
16318 (thunderx2t99_tunings): Likewise.
16320 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16322 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
16323 (sparc_option_override): Honour MASK_FSMULD.
16324 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
16325 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
16326 * config/sparc/sparc.opt (mfsmuld): New option.
16327 * doc/invoke.texi (mfsmuld): Document option.
16329 2017-07-26 Marek Polacek <polacek@redhat.com>
16331 PR middle-end/70992
16332 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
16334 2017-07-26 Richard Biener <rguenther@suse.de>
16336 * gimple-match-head.c (do_valueize): Return OP if valueize
16338 (get_def): New helper to get at the def stmt of a SSA name
16339 if valueize allows.
16340 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
16341 do_valueize to get at the def stmt.
16342 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
16344 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
16346 PR middle-end/46932
16347 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
16349 2017-07-26 Martin Liska <mliska@suse.cz>
16352 * function.c (expand_function_start): Make expansion of
16353 nonlocal_goto_save_area after parm_birth_insn.
16355 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
16357 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
16358 from all CPU target flags enable members.
16360 2017-07-26 Richard Biener <rguenther@suse.de>
16362 * genmatch.c (dt_simplify::gen): Make iterator vars const.
16363 (decision_tree::gen): Make 'type' const.
16364 (write_predicate): Likewise.
16366 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16368 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
16369 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
16370 (rs6000_option_override_internal): Likewise.
16371 (rs6000_expand_vector_set): Likewise.
16372 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
16373 (TARGET_UPPER_REGS_SF): Likewise.
16374 (TARGET_UPPER_REGS_DI): Likewise.
16375 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
16376 (TARGET_DIRECT_MOVE_64BIT): Likewise.
16377 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
16378 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
16379 (Splitters for DI constants in Altivec registers): Likewise.
16380 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
16381 (vsx_set_v4sf_p9): Likewise.
16382 (vsx_set_v4sf_p9_zero): Likewise.
16383 (vsx_insert_extract_v4sf_p9): Likewise.
16384 (vsx_insert_extract_v4sf_p9_2): Likewise.
16386 2017-07-25 Carl Love <cel@us.ibm.com>
16388 * doc/extend.texi: Update the built-in documentation file for the
16389 existing built-in functions
16390 vector signed char vec_cnttz (vector signed char);
16391 vector unsigned char vec_cnttz (vector unsigned char);
16392 vector signed short vec_cnttz (vector signed short);
16393 vector unsigned short vec_cnttz (vector unsigned short);
16394 vector signed int vec_cnttz (vector signed int);
16395 vector unsigned int vec_cnttz (vector unsigned int);
16396 vector signed long long vec_cnttz (vector signed long long);
16397 vector unsigned long long vec_cnttz (vector unsigned long long);
16399 2017-07-25 Andrew Pinski <apinski@cavium.com>
16401 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
16402 accesses where the use is for the first operand of a BIT_INSERT.
16404 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16407 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16408 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16410 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
16412 * config/i386/gstabs.h: Delete.
16413 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
16415 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
16417 * config/i386/i386.c (ix86_decompose_address): Do not check for
16418 register RTX when looking at index_reg or base_reg.
16419 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
16421 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
16423 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
16424 to update EH info here.
16426 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16428 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
16430 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16432 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
16434 2017-07-25 Torsten Duwe <duwe@suse.de>
16436 * common.opt: Introduce -fpatchable-function-entry
16437 command line option, and its variables function_entry_patch_area_size
16438 and function_entry_patch_area_start.
16439 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
16440 including a two-value parser.
16441 * target.def (print_patchable_function_entry): New target hook.
16442 * targhooks.h (default_print_patchable_function_entry): New function.
16443 * targhooks.c (default_print_patchable_function_entry): Likewise.
16444 * toplev.c (process_options): Switch off IPA-RA if
16445 patchable function entries are being generated.
16446 * varasm.c (assemble_start_function): Look at the
16447 patchable-function-entry command line switch and current
16448 function attributes and maybe generate NOP instructions by
16449 calling the print_patchable_function_entry hook.
16450 * doc/extend.texi: Document patchable_function_entry attribute.
16451 * doc/invoke.texi: Document -fpatchable_function_entry
16452 command line option.
16453 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
16455 * doc/tm.texi: Re-generate.
16457 2017-07-25 Jakub Jelinek <jakub@redhat.com>
16460 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
16461 TARGET_AVX512DQ rather than TARGET_AVX512BW.
16463 2017-07-25 Tamar Christina <tamar.christina@arm.com>
16465 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
16467 2017-07-25 Richard Biener <rguenther@suse.de>
16469 PR tree-optimization/81455
16470 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
16471 not walk in cycles when looking for guards.
16473 2017-07-25 Richard Biener <rguenther@suse.de>
16475 PR tree-optimization/81529
16476 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
16477 when optimizing backedge uses.
16479 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
16481 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
16483 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
16484 to dl_section_ref. On AIX, append an expression to subtract
16485 the size of the section length to dl_section_ref.
16487 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
16489 * configure.ac: If any of the config.* scripts fail, exit 1.
16490 * configure: Regenerate.
16492 2017-07-25 Richard Biener <rguenther@suse.de>
16494 PR middle-end/81546
16495 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
16496 of immediate uses, be more verbose on errors.
16498 2017-07-25 Richard Biener <rguenther@suse.de>
16500 PR tree-optimization/81510
16501 * tree-vect-loop.c (vect_is_simple_reduction): When the
16502 reduction stmt is not inside the loop bail out.
16504 2017-07-25 Richard Biener <rguenther@suse.de>
16506 PR tree-optimization/81303
16507 * tree-vect-loop-manip.c (vect_loop_versioning): Build
16508 profitability check against LOOP_VINFO_NITERSM1.
16510 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
16512 * domwalk.c (cmp_bb_postorder): Simplify.
16513 (sort_bbs_postorder): New function. Use it...
16514 (dom_walker::walk): ...here to optimize common cases.
16516 2017-07-25 Martin Liska <mliska@suse.cz>
16519 * ipa-visibility.c (function_and_variable_visibility): Make the
16520 redirection just on target that supports aliasing.
16521 Fix GNU coding style.
16523 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
16526 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
16528 * config/arm/rtems.h: Likewise.
16529 * config/bfin/rtems.h: Likewise.
16530 * config/i386/rtemself.h: Likewise.
16531 * config/lm32/rtems.h: Likewise.
16532 * config/m32c/rtems.h: Likewise.
16533 * config/m68k/rtemself.h: Likewise.
16534 * config/microblaze/rtems.h: Likewise.
16535 * config/mips/rtems.h: Likewise.
16536 * config/moxie/rtems.h: Likewise.
16537 * config/nios2/rtems.h: Likewise.
16538 * config/powerpcspe/rtems.h: Likewise.
16539 * config/rs6000/rtems.h: Likewise.
16540 * config/rtems.h: Likewise.
16541 * config/sh/rtems.h: Likewise.
16542 * config/sh/rtemself.h: Likewise.
16543 * config/sparc/rtemself.h: Likewise.
16545 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16548 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
16549 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
16550 * tree-ssa-structalias.c (alias_get_name): Same.
16552 2017-07-25 Bin Cheng <bin.cheng@arm.com>
16555 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
16556 instructions if no du chain is found.
16558 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
16560 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
16562 2017-07-25 Richard Biener <rguenther@suse.de>
16564 PR middle-end/81505
16565 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
16568 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
16570 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
16571 upper-regs options.
16572 (ISA_2_7_MASKS_SERVER): Likewise.
16573 (ISA_3_0_MASKS_IEEE): Likewise.
16574 (OTHER_P8_VECTOR_MASKS): Likewise.
16575 (OTHER_VSX_VECTOR_MASKS): Likewise.
16576 (POWERPC_MASKS): Likewise.
16577 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
16578 duplicate list of options.
16579 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
16580 explicit -mupper-regs options.
16581 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
16582 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
16583 alias for -mupper-regs-df.
16584 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
16585 (rs6000_init_hard_regno_mode_ok): Likewise.
16586 (rs6000_option_override_internal): Likewise.
16587 (rs6000_opt_masks): Likewise.
16588 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
16589 options in terms of whether -mvsx or -mpower8-vector was used.
16590 (TARGET_UPPER_REGS_DI): Likewise.
16591 (TARGET_UPPER_REGS_SF): Likewise.
16592 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
16593 -mupper-regs-* options.
16595 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16597 * passes.c (emergency_dump_function): Print some empty lines and a
16598 header before the RTL dump.
16600 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
16602 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
16604 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
16607 * config/aarch64/aarch64.c (aarch64_classify_symbol):
16608 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
16610 2017-07-24 Carl Love <cel@us.ibm.com>
16612 * config/rs6000/rs6000-c.c: Add support for built-in functions
16613 vector float vec_extract_fp32_from_shorth (vector unsigned short);
16614 vector float vec_extract_fp32_from_shortl (vector unsigned short);
16615 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
16616 vec_extract_fp_from_shortl): Add defines for the two builtins.
16617 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
16618 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
16620 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
16621 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
16622 * doc/extend.texi: Update the built-in documentation file for the
16623 new built-in function.
16625 2017-07-24 Jakub Jelinek <jakub@redhat.com>
16628 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
16630 * doc/generic.texi: Likewise.
16631 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
16632 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
16634 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
16636 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
16637 (aarch64_mls_elt_merge<mode>): Likewise.
16639 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
16641 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16642 having __cxa_atexit.
16644 2017-07-23 Michael Collison <michael.collison@arm.com>
16646 * config/arm/arm.c (arm_option_override): Deprecate
16647 use of -mstructure-size-boundary.
16648 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
16649 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
16651 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16654 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
16655 Reduce cost estimate for direct moves.
16657 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
16660 * config/i386/i386.c (ix86_option_override_internal): Disable
16661 BMI, BMI2 and TBM instructions for -m16.
16663 2017-07-21 Carl Love <cel@us.ibm.com>
16665 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16666 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
16667 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
16668 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
16669 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
16670 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
16671 VMULOSW): New enum "unspec" values.
16672 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
16673 altivec_vmulosw): New patterns.
16674 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
16675 VMULOSW): Add definitions.
16677 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
16679 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
16680 (qdf24xx): Likewise.
16681 * config/aarch64/aarch64-options-extensions.def (rdma); New.
16682 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
16683 (AARCH64_FL_V8_1): Renumber.
16684 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
16685 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
16686 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
16687 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
16688 rdma to feature modifiers list.
16690 2017-07-21 Yury Gribov <tetra2005@gmail.com>
16692 PR middle-end/56727
16693 * ipa-visibility (function_and_variable_visibility): Convert
16694 recursive PLT call to direct call if appropriate.
16696 2017-07-21 Andrew Pinski <apinski@cavium.com>
16698 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
16699 operand 1 to see if the types precision matches.
16700 * fold-const.c (operand_equal_p): Likewise.
16702 2017-07-21 Richard Biener <rguenther@suse.de>
16704 PR tree-optimization/81303
16705 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
16706 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
16707 (vect_peeling_hash_get_lowest_cost): Adjust.
16708 (vect_enhance_data_refs_alignment): Likewise. Use
16709 vect_get_peeling_costs_all_drs to compute the penalty for no
16710 peeling to match up costs.
16712 2017-07-21 Richard Biener <rguenther@suse.de>
16714 PR tree-optimization/81500
16715 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
16716 we didn't identify a reduction path.
16718 2017-07-21 Tom de Vries <tom@codesourcery.com>
16719 Cesar Philippidis <cesar@codesourcery.com>
16721 PR gcov-profile/81442
16722 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
16725 2017-07-21 Tom de Vries <tom@codesourcery.com>
16728 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
16730 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
16731 nvptx_override_options_after_change.
16733 2017-07-21 Ulrich Drepper <drepper@redhat.com>
16735 * dwarf2out.c (output_file_names): Avoid double testing for
16736 dwarf_version >= 5.
16738 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
16740 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
16742 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16744 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
16746 (try_crossjump_to_edge): Do not punt on partitioned functions.
16748 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
16750 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
16751 Put all BBs reachable only via paths crossing cold region to cold
16753 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
16755 2016-07-21 Richard Biener <rguenther@suse.de>
16757 PR tree-optimization/81303
16758 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
16759 into account prologue and epilogue iterations when raising
16760 min_profitable_iters to sth at least covering one vector iteration.
16762 2017-07-21 Tamar Christina <tamar.christina@arm.com>
16764 * config/arm/arm.c (arm_test_cpu_arch_dat):
16767 2017-07-20 Nathan Sidwell <nathan@acm.org>
16769 Remove TYPE_METHODS.
16770 * tree.h (TYPE_METHODS): Delete.
16771 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
16772 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
16773 (dbxout_type_methods): Scan TYPE_FIELDS.
16774 (dbxout_type): Don't check TYPE_METHODS here.
16775 * function.c (use_register_for_decl): Always ignore register for
16776 class types when not optimizing.
16777 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
16778 * tree.c (free_lang_data_in_type): Stitch out member functions and
16779 templates from TYPE_FIELDS.
16780 (build_distinct_type_copy, verify_type_variant,
16781 verify_type): Member fns are on TYPE_FIELDS.
16782 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
16783 * tree-pretty-print.c (dump_generic_node): Likewise.
16785 2017-07-20 Jakub Jelinek <jakub@redhat.com>
16788 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
16789 V2TImode and V4TImode.
16790 (ix86_expand_vector_extract): Likewise.
16791 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
16792 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
16793 (ssescalarmode): Handle V4TImode and V2TImode.
16794 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
16795 (*vec_extractv2ti, *vec_extractv4ti): New insns.
16796 (VEXTRACTI128_MODE): New mode iterator.
16797 (splitter for *vec_extractv?ti first element): New.
16798 (VEC_INIT_MODE): New mode iterator.
16799 (vec_init<mode>): Consolidate 3 expanders into one using
16800 VEC_INIT_MODE mode iterator.
16802 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16804 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
16805 non_spilled_static_chain_regno_p.
16807 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
16809 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
16811 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
16813 * bb-reorder.c (connect_traces): Allow copying of blocks within
16816 2017-07-20 Richard Biener <rguenther@suse.de>
16818 * gimple.h (gimple_phi_result): Add gphi * overload.
16819 (gimple_phi_result_ptr): Likewise.
16820 (gimple_phi_arg): Likewise. Adjust index assert to only
16821 allow actual argument accesses rather than all slots available
16823 (gimple_phi_arg_def): Add gphi * overload.
16824 * tree-phinodes.c (make_phi_node): Initialize only actual
16826 (resize_phi_node): Clear memory not covered by old node,
16827 do not initialize excess argument slots.
16828 (reserve_phi_args_for_new_edge): Initialize new argument slot
16831 2017-07-20 Bin Cheng <bin.cheng@arm.com>
16833 PR tree-optimization/81388
16835 2016-07-21 Bin Cheng <bin.cheng@arm.com>
16837 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
16838 by removing computation of may_be_zero.
16840 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
16841 Tom de Vries <tom@codesourcery.com>
16843 PR middle-end/81030
16844 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
16845 when gimple level profile disagrees with what RTL expander did.
16847 2017-07-20 Richard Biener <rguenther@suse.de>
16849 PR tree-optimization/61171
16850 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
16851 (vect_analyze_stmt): Add slp instance parameter.
16852 (vectorizable_reduction): Likewise.
16853 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
16854 (vect_is_simple_reduction): Deal with chains not detected
16855 as SLP reduction chain, specifically not properly associated
16856 chains containing a mix of plus/minus.
16857 (get_reduction_op): Remove.
16858 (get_initial_defs_for_reduction): Simplify, pass in whether
16859 this is a reduction chain, pass in the SLP node for the PHIs.
16860 (vect_create_epilog_for_reduction): Get the SLP instance as
16862 (vectorizable_reduction): Get the SLP instance as arg.
16863 During analysis remember the SLP node with the PHIs in the
16864 instance. Simplify getting at the vectorized reduction PHIs.
16865 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
16866 through SLP instance.
16867 (vect_slp_analyze_operations): Likewise.
16868 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
16869 (vect_transform_stmt): Likewise.
16871 2017-07-20 Tom de Vries <tom@codesourcery.com>
16873 PR tree-optimization/81489
16874 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
16875 read of phi arg location to before loop that modifies phi.
16877 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16879 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
16882 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16884 PR middle-end/81331
16885 * except.c (execute): Fix ordering issue.
16887 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16889 PR rtl-optimization/81423
16890 * combine.c (make_compound_operation_int): Don't try to optimize
16891 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
16893 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
16895 PR rtl-optimization/81423
16896 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
16897 with a constant that is -1 in the truncated to mode.
16899 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16901 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
16902 (determine_unlikely_bbs): ... here.
16903 * predict.h (propagate_unlikely_bbs_forward): Declare.
16904 * cfgexpand.c (pass_expand::execute): Use it.
16905 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
16907 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
16908 propagate_unlikely_bbs_forward.
16910 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
16912 PR middle-end/81331
16913 * except.c (maybe_add_nop_after_section_switch): New function.
16916 2017-07-19 Tom de Vries <tom@codesourcery.com>
16918 * gimple.h (gimple_phi_set_arg): Make assert more strict.
16920 2017-07-19 Tom de Vries <tom@codesourcery.com>
16922 * gimple.h (gimple_phi_arg): Make assert more strict.
16924 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
16926 * config.gcc (powerpc*-*-*): Add mmintrin.h.
16927 * config/rs6000/mmintrin.h: New file.
16928 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
16930 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16932 PR tree-optimization/81346
16933 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
16935 2017-07-19 Tom de Vries <tom@codesourcery.com>
16937 * config/nvptx/nvptx.md (VECIM): Add V2DI.
16939 2017-07-19 Tom de Vries <tom@codesourcery.com>
16941 * config/nvptx/nvptx-modes.def: Add V2DImode.
16942 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
16943 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
16944 (nvptx_output_mov_insn): Handle lack of mov.b128.
16945 (nvptx_print_operand): Handle 'H' and 'L' codes.
16946 (nvptx_vector_mode_supported): Allow V2DImode.
16947 (nvptx_preferred_simd_mode): New function.
16948 (nvptx_data_alignment): New function.
16949 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
16950 nvptx_preferred_simd_mode.
16951 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
16953 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
16955 2017-07-19 Tom de Vries <tom@codesourcery.com>
16957 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
16958 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
16959 (nvptx_vector_mode_supported): New function. Allow V2SImode.
16960 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
16961 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
16962 (mov<VECIM>_insn): New define_insn.
16963 (define_expand "mov<VECIM>): New define_expand.
16965 2017-07-19 Tom de Vries <tom@codesourcery.com>
16967 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
16969 2017-07-19 Jakub Jelinek <jakub@redhat.com>
16971 PR tree-optimization/81346
16972 * fold-const.h (fold_div_compare, range_check_type): Declare.
16973 * fold-const.c (range_check_type): New function.
16974 (build_range_check): Use range_check_type.
16975 (fold_div_compare): No longer static, rewritten into
16976 a match.pd helper function.
16977 (fold_comparison): Don't call fold_div_compare here.
16978 * match.pd (X / C1 op C2): New optimization using fold_div_compare
16979 as helper function.
16981 2017-07-19 Nathan Sidwell <nathan@acm.org>
16983 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
16984 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
16985 * tree.c (find_decls_types_r, verify_type): Use
16986 TYPE_{MIN,MAX}_VALUE_RAW.
16987 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
16988 (hash_tree): Likewise.
16989 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
16991 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
16994 2017-07-18 Tom de Vries <tom@codesourcery.com>
16996 PR middle-end/81464
16997 * omp-expand.c (expand_omp_for_static_chunk): Handle
16998 equal-argument loop exit phi.
17000 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
17003 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
17004 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
17005 operand 2 predicate.
17006 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
17007 operand 2 predicate.
17008 (ror,rol -> rorx splitters): Use const_int_operand as
17009 operand 2 predicate.
17011 2017-06-18 Richard Biener <rguenther@suse.de>
17013 PR tree-optimization/81410
17014 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
17015 the gap in the ! slp_perm SLP case after each group.
17017 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17019 PR middle-end/81463
17020 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
17023 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
17025 PR middle-end/81462
17026 * predict.c (set_even_probabilities): Cleanup; do not affect
17027 probabilities that are already known.
17028 (combine_predictions_for_bb): Call even when count is set.
17030 2017-07-18 Nathan Sidwell <nathan@acm.org>
17032 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
17035 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17038 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
17039 optimization for loop niter analysis.
17041 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
17044 * config/avr/avr.c (avr_optimize_casesi): Don't use
17045 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
17047 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
17049 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
17050 body_cost_vec from _vect_peel_extended_info.
17051 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
17052 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
17055 2017-07-18 Bin Cheng <bin.cheng@arm.com>
17057 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
17059 2017-07-18 Richard Biener <rguenther@suse.de>
17061 PR tree-optimization/80620
17062 PR tree-optimization/81403
17063 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
17064 info when re-using a VN table entry.
17066 2017-07-18 Richard Biener <rguenther@suse.de>
17068 PR tree-optimization/81418
17069 * tree-vect-loop.c (vectorizable_reduction): Properly compute
17070 vectype_in. Verify that with lane-reducing reduction operations
17071 we have a single def-use cycle.
17073 2017-07-17 Carl Love <cel@us.ibm.com>
17075 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
17077 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17078 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
17079 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
17080 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
17081 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
17082 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
17083 VMULOSW): New enum "unspec" values.
17084 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
17085 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
17086 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
17087 altivec_vmulosw): New patterns.
17088 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
17089 VMULOSW): Add definitions.
17090 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
17092 * config/alpha/alpha.c: Include predict.h.
17094 2017-07-17 Yury Gribov <tetra2005@gmail.com>
17096 * tree-vrp.c (compare_assert_loc): Fix comparison function
17097 to return predictable results.
17099 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17101 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
17103 (subdi3): Likewise.
17104 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
17105 * doc/invoke.texi (mexpand-adddi): Update text.
17107 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17109 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
17110 that also clobbers the CC register. The old expand code is moved
17112 (*arc_clzsi2): ... here.
17113 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
17114 the CC register. The old expand code is moved to ...
17115 (arc_ctzsi2): ... here.
17117 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17119 * config/arc/arc.opt (mindexed-loads): Use initial value
17120 TARGET_INDEXED_LOADS_DEFAULT.
17121 (mauto-modify-reg): Use initial value
17122 TARGET_AUTO_MODIFY_REG_DEFAULT.
17123 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17124 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17125 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
17126 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
17128 2017-07-17 Martin Liska <mliska@suse.cz>
17131 * opts.c (finish_options): Do not allow -fgnu-tm
17132 w/ -fsanitize={kernel-,}address. Say sorry.
17134 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17137 * tree-loop-distribution.c (classify_partition): Only assert on
17138 numer of iterations.
17139 (merge_dep_scc_partitions): Delete prameter. Update function call.
17140 (distribute_loop): Remove code handling loop with unknown niters.
17141 (pass_loop_distribution::execute): Skip loop with unknown niters.
17143 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17146 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
17147 function sort_partitions_by_post_order.
17149 2017-07-17 Bin Cheng <bin.cheng@arm.com>
17151 PR tree-optimization/81374
17152 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
17153 the max index of basic blocks, rather than number of basic blocks.
17155 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17157 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
17159 (arc_legitimate_pic_operand_p): Likewise.
17160 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
17162 (arc_needs_pcl_p): Likewise.
17163 (arc_legitimate_pc_offset_p): Likewise.
17164 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
17165 function is also used in constrains.md.
17166 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
17167 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
17168 PLUS. Only return true/false in known cases, otherwise assert.
17169 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
17170 is already called in arc_legitimate_constant_p.
17171 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
17173 (LEGITIMATE_PIC_OPERAND_P): Use
17174 arc_raw_symbolic_reference_mentioned_p function.
17175 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
17180 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
17181 Andrew Burgess <andrew.burgess@embecosm.com>
17183 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
17184 (arc_return_address_register): New function.
17185 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
17186 (arc_handle_fndecl_attribute): Add naked attribute.
17187 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
17188 (TARGET_WARN_FUNC_RETURN): Likewise.
17189 (arc_allocate_stack_slots_for_args): New function.
17190 (arc_warn_func_return): Likewise.
17191 (machine_function): Change type fn_type.
17192 (arc_compute_function_type): Consider new naked function type,
17193 change function return type.
17194 (arc_must_save_register): Adapt to handle new
17195 arc_compute_function_type's return type.
17196 (arc_expand_prologue): Likewise.
17197 (arc_expand_epilogue): Likewise.
17198 (arc_return_address_regs): Delete.
17199 (arc_return_address_register): New function.
17200 (arc_epilogue_uses): Use above function.
17201 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
17202 (arc_function_type): Change encoding, add naked type.
17203 (ARC_INTERRUPT_P): Change to handle the new encoding.
17204 (ARC_FAST_INTERRUPT_P): Likewise.
17205 (ARC_NORMAL_P): Define.
17206 (ARC_NAKED_P): Likewise.
17207 (arc_compute_function_type): Delete prototype.
17208 * config/arc/arc.md (in_ret_delay_slot): Use
17209 arc_return_address_register function.
17210 (simple_return): Likewise.
17211 (p_return_i): Likewise.
17213 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17215 PR tree-optimization/81428
17216 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
17217 can't be built for those types.
17219 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17221 Remove stuff dead since r239246.
17223 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
17224 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
17225 (avr_inform_devices): Remove dead stuff.
17227 2017-07-17 Tamar Christina <tamar.christina@arm.com>
17229 * config/arm/arm_neon.h: Fix softp typo.
17231 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17233 PR tree-optimization/81365
17234 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
17235 aggregate moves onto bb predecessor edges, make sure there are no
17236 loads that could alias the lhs in between the start of bb and the
17239 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
17242 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
17243 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
17244 [LSHIFTRT, outer_code = TRUNCATE]: Same.
17246 2017-07-17 Jakub Jelinek <jakub@redhat.com>
17248 PR tree-optimization/81396
17249 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
17250 (init_symbolic_number): Initialize it to 1.
17251 (perform_symbolic_merge): Add n_ops from both operands into the new
17253 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
17254 without base_addr as useless if they need more than one operation.
17255 (bswap_replace): Handle !bswap case for NULL base_addr.
17257 2017-07-17 Tom de Vries <tom@codesourcery.com>
17260 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
17263 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
17265 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
17266 conditional builtin define __FIX_LEON3FT_B2BST.
17268 2017-07-17 Daniel Cederman <cederman@gaisler.com>
17270 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
17271 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
17274 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17276 PR rtl-optimization/81424
17277 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
17278 to remove potential trapping from operands if -fnon-call-exceptions.
17280 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17282 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
17283 profile_proability for scalling.
17284 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
17286 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17288 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
17290 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17292 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
17293 fixpoint arithmetics.
17295 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17297 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
17298 fixpoint arithmetics.
17300 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17302 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
17303 fixpoint arithmetics.
17305 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
17307 * profile-count.h (profile_probability::from_reg_br_prob_note,
17308 profile_probability::to_reg_br_prob_note): New functions.
17309 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
17310 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
17311 * predict.c (probability_reliable_p): Update.
17312 (edge_probability_reliable_p): Update.
17313 (br_prob_note_reliable_p): Update.
17314 (invert_br_probabilities): Update.
17315 (add_reg_br_prob_note): New function.
17316 (combine_predictions_for_insn): Update.
17317 * asan.c (asan_clear_shadow): Update.
17318 * cfgbuild.c (compute_outgoing_frequencies): Update.
17319 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17320 (update_br_prob_note): Update.
17321 (rtl_verify_edges): Update.
17322 (purge_dead_edges): Update.
17323 (fixup_reorder_chain): Update.
17324 * emit-rtl.c (try_split): Update.
17325 * ifcvt.c (cond_exec_process_insns): Update.
17326 (cond_exec_process_if_block): Update.
17327 (dead_or_predicable): Update.
17328 * internal-fn.c (expand_addsub_overflow): Update.
17329 (expand_neg_overflow): Update.
17330 (expand_mul_overflow): Update.
17331 * loop-doloop.c (doloop_modify): Update.
17332 * loop-unroll.c (compare_and_jump_seq): Update.
17333 * optabs.c (emit_cmp_and_jump_insn_1): Update.
17334 * predict.h: Update.
17335 * reorg.c (mostly_true_jump): Update.
17337 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
17338 * config/alpha/alpha.c (emit_unlikely_jump): Update.
17339 * config/arc/arc.c: (emit_unlikely_jump): Update.
17340 * config/arm/arm.c: (emit_unlikely_jump): Update.
17341 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
17342 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
17343 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
17344 (ix86_print_operand): Update.
17345 (ix86_split_fp_branch): Update.
17346 (predict_jump): Update.
17347 * config/ia64/ia64.c (ia64_print_operand): Update.
17348 * config/mmix/mmix.c (mmix_print_operand): Update.
17349 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
17350 (rs6000_expand_split_stack_prologue): Update.
17351 * config/rs6000/rs6000.c: Update.
17352 * config/s390/s390.c (s390_expand_vec_strlen): Update.
17353 (s390_expand_vec_movstr): Update.
17354 (s390_expand_cs_tdsi): Update.
17355 (s390_expand_split_stack_prologue): Update.
17356 * config/sh/sh.c (sh_print_operand): Update.
17357 (expand_cbranchsi4): Update.
17358 (expand_cbranchdi4): Update.
17359 * config/sparc/sparc.c (output_v9branch): Update.
17360 * config/spu/spu.c (get_branch_target): Update.
17361 (ea_load_store_inline): Update.
17362 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
17363 * config/tilepro/tilepro.c: Update.
17365 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
17367 * gimplify.c (mostly_copy_tree_r): Revert latest change.
17368 (gimplify_save_expr): Likewise.
17370 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17372 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
17374 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17376 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
17378 * timevar.def (TV_IPA_FNSUMMARY): Define.
17380 2017-07-16 Daniel Cederman <cederman@gaisler.com>
17382 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
17383 to back store errata sensitive sequence from being generated.
17384 (sqrtdf2_fix): Likewise.
17386 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17388 * tree-ssa-threadupdate.c (compute_path_counts,
17389 update_joiner_offpath_counts): Use profile_probability.
17391 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
17394 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17396 * config/arm/arm-c.c (arm_cpu_builtins): Define
17397 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17399 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
17401 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17402 array entries to represent __ieee128 versions of the
17403 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
17404 scalar_extract_sig, and scalar_insert_exp built-in functions.
17405 (altivec_resolve_overloaded_builtin): Add special case handling
17406 for the __builtin_scalar_insert_exp function, as represented by
17407 the P9V_BUILTIN_VEC_VSIEDP constant.
17408 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
17409 exponent support for __ieee128 argument.
17410 (VSESQP): Add scalar extract signature support for __ieee128
17412 (VSTDCNQP): Add scalar test negative support for __ieee128
17414 (VSIEQP): Add scalar insert exponent support for __int128 argument
17415 with __ieee128 result.
17416 (VSIEQPF): Add scalar insert exponent support for __ieee128
17417 argument with __ieee128 result.
17418 (VSTDCQP): Add scalar test data class support for __ieee128
17420 (VSTDCNQP): Add overload support for scalar test negative with
17421 __ieee128 argument.
17422 (VSTDCQP): Add overload support for scalar test data class
17423 __ieee128 argument.
17424 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
17425 UNSPEC_VSX_SXSIGDP.
17426 (UNSPEC_VSX_SIEXPQP): New constant.
17427 (xsxexpqp): New insn for VSX scalar extract exponent quad
17429 (xsxsigqp): New insn for VSX scalar extract significand quad
17431 (xsiexpqpf): New insn for VSX scalar insert exponent quad
17432 precision with floating point argument.
17433 (xststdcqp): New expand for VSX scalar test data class quad
17435 (xststdcnegqp): New expand for VSX scalar test negative quad
17437 (xststdcqp): New insn to match expansions for VSX scalar test data
17438 class quad precision and VSX scalar test negative quad precision.
17439 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
17440 special case operand checking to enforce that second operand of
17441 VSX scalar test data class with quad precision argument is a 7-bit
17443 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
17444 prototypes and descriptions of __ieee128 versions of
17445 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
17446 scalar_test_data_class, and scalar_test_neg built-in functions.
17448 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17450 PR tree-optimization/81162
17451 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
17452 replace a negate with an add.
17454 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
17456 * doc/invoke.texi (arm/-mcpu): Document +crypto.
17458 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17460 * config/arm/arm-c.c (arm_cpu_builtins): Define
17461 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
17463 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17465 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
17466 (armv8-r): Set ARM Cortex-R52 as default CPU.
17467 * config/arm/arm-tables.opt: Regenerate.
17468 * config/arm/arm-tune.md: Regenerate.
17469 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17471 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
17472 extension for -mcpu=cortex-r52.
17474 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
17476 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
17477 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
17478 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
17479 (fp-armv8): Define it as FP_ARMv8 only.
17480 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
17481 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
17483 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
17484 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
17485 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
17486 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
17487 than TARGET_FPU_ARMV8.
17488 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
17489 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
17490 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
17492 * config/arm/neon.md (neon_vrint): Likewise.
17493 (neon_vcvt): Likewise.
17494 (neon_<fmaxmin_op><mode>): Likewise.
17495 (<fmaxmin><mode>3): Likewise.
17496 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
17497 * config/arm/predicates.md (arm_cond_move_operator): Check against
17498 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
17500 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
17502 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
17503 to top of function.
17505 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17507 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
17508 loop in comment with memset.
17510 2017-07-14 Martin Liska <mliska@suse.cz>
17512 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
17513 * dwarf2out.c (is_java): Remove the function.
17514 (output_pubname): Remove usage of the function.
17515 (lower_bound_default): Remove usage of DW_LANG_Java.
17516 (gen_compile_unit_die): Likewise.
17517 * gcc.c: Remove compiler defaults for .java and .zip files.
17518 * gimple-expr.c (remove_suffix): Change as there's no longer
17519 extension than 4-letter one.
17520 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
17521 (gimplify_save_expr): Likewise.
17522 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
17523 as it's possible even for other languages than Java.
17524 * langhooks.h (struct lang_hooks): Remove Java from a comment.
17525 * lto-opts.c (lto_write_options): Remove reference to Java.
17526 * opts.c (strip_off_ending): Update file extension handling.
17527 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
17528 * tree-eh.c (lower_resx): Likewise.
17529 * tree.c (free_lang_data_in_type): Remove dead code.
17530 (find_decls_types_r): Likewise.
17531 (build_common_builtin_nodes): Remove Java from a comment.
17532 (verify_type): Remove dead code.
17533 * varasm.c (assemble_external): Remove Java from a comment.
17535 2017-07-14 Martin Liska <mliska@suse.cz>
17537 * opts.c (finish_options): Add quotes.
17538 (common_handle_option): Likewise.
17540 2017-07-14 Martin Liska <mliska@suse.cz>
17542 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
17543 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
17544 Remove N_SO_PASCAL.
17545 * dwarf2out.c (lower_bound_default): Do not handle
17547 (gen_compile_unit_die): Likewise.
17548 * gcc.c: Remove default extension binding for GNU Pascal.
17549 * stmt.c: Remove Pascal language from a comment.
17550 * xcoffout.c: Likewise.
17552 2017-07-13 David Malcolm <dmalcolm@redhat.com>
17555 * diagnostic-show-locus.c (fixit_cmp): New function.
17556 (layout::layout): Sort m_fixit_hints.
17557 (column_range::column_range): Assert that the values are valid.
17558 (struct char_span): New struct.
17559 (correction::overwrite): New method.
17560 (struct source_line): New struct.
17561 (line_corrections::add_hint): Add assertions. Reimplement memcpy
17562 calls in terms of classes source_line and char_span, and
17563 correction::overwrite.
17564 (selftest::test_overlapped_fixit_printing_2): New function.
17565 (selftest::diagnostic_show_locus_c_tests): Call it.
17567 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
17569 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
17570 early if there is no lhs.
17572 2017-07-13 Martin Liska <mliska@suse.cz>
17574 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
17575 (gen_reference_type_die): Likewise.
17576 * stor-layout.c: Remove Pascal-related comment.
17578 2017-07-13 Martin Liska <mliska@suse.cz>
17580 * opts.c (finish_options): Add quotes to error messages.
17581 (parse_sanitizer_options): Likewise.
17583 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
17585 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
17587 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
17589 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
17591 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
17593 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
17595 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
17597 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
17600 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
17601 provides the hardware capability bits, define the macro
17602 __BUILTIN_CPU_SUPPORTS__.
17603 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
17604 if GLIBC does not provide the hardware capability bits. Add a
17605 gcc_unreachable call if the built-in cpu function is neither
17606 __builtin_cpu_is nor __builtin_cpu_supports.
17607 (rs6000_get_function_versions_dispatcher): Change the warning
17608 that an old GLIBC is used which does not export the capability
17609 bits to be an error.
17610 * doc/extend.texi (target_clones attribute): Document the
17611 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
17612 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
17613 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
17614 the macros defined by GCC if the newer GLIBC is available.
17616 2017-07-12 Jeff Law <law@redhat.com>
17618 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
17619 remaining includes slightly.
17620 * config/riscv/riscv-builtins.c: Include profile-count.h.
17622 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17625 * config/avr/avr.c (avr_set_current_function): In diagnostic
17626 messages: Quote keywords and (parts of) identifiers.
17627 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
17630 2017-07-12 Carl Love <cel@us.ibm.com>
17632 * config/rs6000/rs6000-c.c: Add support for built-in functions
17633 vector bool char vec_revb (vector bool char);
17634 vector bool short vec_revb (vector short char);
17635 vector bool int vec_revb (vector bool int);
17636 vector bool long long vec_revb (vector bool long long);
17637 * doc/extend.texi: Update the built-in documentation file for the
17638 new built-in functions.
17640 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17642 * config/s390/s390.md: Remove movcc splitter.
17644 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17646 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
17647 load/store on condition.
17649 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
17652 * config/avr/avr.c (avr_encode_section_info)
17653 [progmem && !TREE_READONLY]: Error if progmem object needs
17656 2017-07-11 Michael Collison <michael.collison@arm.com>
17658 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
17661 2017-07-11 Carl Love <cel@us.ibm.com>
17663 * config/rs6000/rs6000-c.c: Add support for builtins
17664 vector unsigned int vec_parity_lsbb (vector signed int);
17665 vector unsigned int vec_parity_lsbb (vector unsigned int);
17666 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
17667 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
17668 vector unsigned long long vec_parity_lsbb (vector signed long long);
17669 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
17670 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
17671 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17672 * doc/extend.texi: Update the built-in documentation file for the
17673 new built-in functions.
17675 2017-07-11 David Malcolm <dmalcolm@redhat.com>
17677 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
17678 (layout::m_primary_loc): New field.
17679 (layout::layout): Initialize new field. Move location filtering
17680 logic from here to...
17681 (layout::maybe_add_location_range): ...this new method. Add
17682 support for filtering to just the lines already specified by other
17684 (layout::will_show_line_p): New method.
17685 (gcc_rich_location::add_location_if_nearby): New method.
17686 (selftest::test_add_location_if_nearby): New test function.
17687 (selftest::diagnostic_show_locus_c_tests): Call it.
17688 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
17691 2017-07-11 Tom de Vries <tom@codesourcery.com>
17693 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
17694 (bb_first_real_insn): New function.
17695 (nvptx_single): Add extra initialization of broadcasted condition
17698 2017-07-11 Nathan Sidwell <nathan@acm.org>
17700 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
17702 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
17704 * doc/extend.texi (AVR Function Attributes): Remove weblink to
17705 Binutils doc as TEXI will mess them up.
17706 * doc/invoke.texi (AVR Options): Same here.
17708 2017-07-11 Daniel Cederman <cederman@gaisler.com>
17710 * config/sparc/sparc.opt (mfix-ut700): New option.
17711 (mfix-gr712rc): Likewise.
17712 (sparc_fix_b2bst): New variable.
17713 * doc/invoke.texi (SPARC options): Document them.
17714 (ARM options): Fix warnings.
17715 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
17716 instructions to prevent sequences that can trigger the store-store
17717 errata for certain LEON3FT processors.
17718 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
17719 (sparc_option_override): Set sparc_fix_b2bst appropriately.
17720 * config/sparc/sparc.md (fix_b2bst): New attribute.
17721 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
17723 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
17726 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
17728 (*rsqrtsf2_sse): Ditto.
17730 (div<mode>3): Macroize insn from divdf3 and divsf3
17731 using MODEF mode iterator.
17733 2017-07-10 Martin Sebor <msebor@redhat.com>
17735 PR tree-optimization/80397
17736 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
17737 instead of testing for equality to INTEGER_TYPE.
17739 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
17741 * config.gcc: Remove uclibc from arc target spec.
17743 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
17745 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
17747 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
17750 * lto-wrapper.c (remove_option): New function.
17751 (merge_and_complain): Merge PIC/PIE options more realistically.
17753 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17755 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
17759 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
17760 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
17761 * config.in: Regenerate.
17762 * configure: Regenerate.
17763 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
17764 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
17765 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
17766 (TARGET_GASISR_PROLOGUES): ...target mask.
17767 * common/config/avr/avr-common.c
17768 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
17769 Set -mgas-isr-prologues.
17770 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
17771 INSERT_PASS_BEFORE for it.
17772 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
17773 * config/avr/avr.c (avr_option_override)
17774 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
17775 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
17776 (avr_attribute_table) <no_gccisr>: Add new function attribute.
17777 (avr_set_current_function) <is_no_gccisr>: Init machine field.
17778 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
17780 (make_avr_pass_pre_proep): New function.
17781 (emit_push_sfr) <treg>: Add argument to function and use it
17782 instead of TMP_REG.
17783 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
17784 and set machine->gasisr.yes.
17785 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
17786 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
17787 __gcc_isr.n_pushed to .L__stack_usage.
17788 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
17789 (avr_asm_final_postscan_insn): ...this new static function.
17790 * config/avr/avr.h (machine_function)
17791 <is_no_gccisr, use_L__stack_usage>: New fields.
17792 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
17793 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
17794 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
17795 (gasisr, *gasisr): New expander and insn.
17796 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
17797 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
17798 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
17800 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
17802 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
17805 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
17807 Move jump-tables out of .text again.
17810 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
17811 (ASM_OUTPUT_ADDR_VEC): New function.
17812 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
17813 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
17814 INSN_ADDRESSes as asm comment.
17815 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
17816 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
17817 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
17818 * config/avr/avr.md (*tablejump): Adjust comment.
17819 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
17820 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
17822 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
17823 (avr_output_addr_vec): New proto.
17824 (avr_log_t) <insn_addresses>: New field.
17826 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
17829 * config/i386/i386.c (ix86_function_arg_advance): Set
17830 outgoing_args_on_stack to true if there are outgoing arguments
17832 (ix86_function_arg): Likewise.
17833 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
17834 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
17835 * config/i386/i386.h (machine_function): Add
17836 outgoing_args_on_stack.
17838 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
17840 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
17841 supporting pthreds.
17842 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
17844 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
17846 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
17847 (REAL_H): Remove $(MACHMODE_H).
17848 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
17850 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
17851 $(MACHMODE_H) and double-int.h.
17852 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
17854 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
17857 2017-07-07 Andrew Pinski <apinski@cavium.com>
17859 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
17860 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
17862 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
17864 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
17865 Add warning if GCC was not configured to link against a GLIBC that
17866 exports the hardware capability bits.
17867 (make_resolver_func): Make resolver function private and not a
17868 COMDAT function. Create the name with clone_function_name instead
17869 of make_unique_name.
17872 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
17873 correct operand in doing the split.
17875 2017-07-07 Carl Love <cel@us.ibm.com>
17877 * config/rs6000/rs6000-c: Add support for built-in function
17878 vector unsigned short vec_pack_to_short_fp32 (vector float,
17880 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
17881 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
17882 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
17883 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
17884 (convert_4f32_8i16): Add define_expand.
17885 * doc/extend.texi: Update the built-in documentation file for the
17886 new built-in function.
17888 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17890 * config/sparc/m8.md: New file.
17891 * config/sparc/sparc.md: Include m8.md.
17893 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17895 * config/sparc/sparc.opt: New option -mvis4b.
17896 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
17897 (sparc_option_override): Handle VIS4B.
17898 (enum sparc_builtins): Define
17899 SPARC_BUILTIN_DICTUNPACK{8,16,32},
17900 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
17901 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
17902 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
17903 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
17904 (check_constant_argument): New function.
17905 (sparc_vis_init_builtins): Define builtins
17906 __builtin_vis_dictunpack{8,16,32},
17907 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
17908 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
17909 __builtin_vis_fpcmpde{8,16,32}shl and
17910 __builtin_vis_fpcmpur{8,16,32}shl.
17911 (sparc_expand_builtin): Check that the constant operands to
17912 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
17913 constant and in range.
17914 * config/sparc/sparc-c.c (sparc_target_macros): Handle
17916 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
17917 (SPARC_IMM5_P): Likewise.
17918 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
17919 (enabled): Handle vis4b.
17920 (UNSPEC_DICTUNPACK): New unspec.
17921 (UNSPEC_FPCMPSHL): Likewise.
17922 (UNSPEC_FPUCMPSHL): Likewise.
17923 (UNSPEC_FPCMPDESHL): Likewise.
17924 (UNSPEC_FPCMPURSHL): Likewise.
17925 (cpu_feature): New CPU feature `vis4b'.
17926 (dictunpack{8,16,32}): New insns.
17927 (FPCSMODE): New mode iterator.
17928 (fpcscond): New code iterator.
17929 (fpcsucond): Likewise.
17930 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
17931 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
17932 (fpcmpde{8,16,32}{si,di}shl): Likewise.
17933 (fpcmpur{8,16,32}{si,di}shl): Likewise.
17934 * config/sparc/constraints.md: Define constraints `q' for unsigned
17935 2-bit integer constants and `t' for unsigned 5-bit integer
17937 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
17939 (imm5_operand_dictunpack16): Likewise.
17940 (imm5_operand_dictunpack32): Likewise.
17941 (imm2_operand): Likewise.
17942 * doc/invoke.texi (SPARC Options): Document -mvis4b.
17943 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
17944 ditunpack* and fpcmp*shl builtins.
17946 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17948 * config.gcc: Handle m8 in --with-{cpu,tune} options.
17949 * config.in: Add HAVE_AS_SPARC6 define.
17950 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
17952 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
17954 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
17955 (CPP_CPU_SPEC): Handle m8.
17956 (ASM_CPU_SPEC): Likewise.
17957 * config/sparc/sparc-opts.h (enum processor_type): Add
17959 * config/sparc/sparc.c (m8_costs): New struct.
17960 (sparc_option_override): Handle TARGET_CPU_m8.
17961 (sparc32_initialize_trampoline): Likewise.
17962 (sparc64_initialize_trampoline): Likewise.
17963 (sparc_issue_rate): Likewise.
17964 (sparc_register_move_cost): Likewise.
17965 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
17966 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
17967 (ASM_CPU64_DEFAULT_SPEC): Likewise.
17968 (CPP_CPU_SPEC): Handle M8.
17969 (ASM_CPU_SPEC): Likewise.
17970 (AS_M8_FLAG): Define.
17971 * config/sparc/sparc.md: Add m8 to the cpu attribute.
17972 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
17973 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
17975 * configure: Regenerate.
17976 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
17979 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
17981 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
17983 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
17984 ("*movdi_insn_sp32"): Do not set v3pipe.
17985 ("*movsi_insn"): Likewise.
17986 ("*movdi_insn_sp64"): Likewise.
17987 ("*movsf_insn"): Likewise.
17988 ("*movdf_insn_sp32"): Likewise.
17989 ("*movdf_insn_sp64"): Likewise.
17990 ("*zero_extendsidi2_insn_sp64"): Likewise.
17991 ("*sign_extendsidi2_insn"): Likewise.
17992 ("*mov<VM32:mode>_insn"): Likewise.
17993 ("*mov<VM64:mode>_insn_sp64"): Likewise.
17994 ("*mov<VM64:mode>_insn_sp32"): Likewise.
17995 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
17996 ("<vlop:code><VL:mode>3"): Likewise.
17997 ("*not_<vlop:code><VL:mode>3"): Likewise.
17998 ("*nand<VL:mode>_vis"): Likewise.
17999 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
18000 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
18001 ("one_cmpl<VL:mode>2"): Likewise.
18002 ("faligndata<VM64:mode>_vis"): Likewise.
18003 ("alignaddrsi_vis"): Likewise.
18004 ("alignaddrdi_vis"): Likweise.
18005 ("alignaddrlsi_vis"): Likewise.
18006 ("alignaddrldi_vis"): Likewise.
18007 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18008 ("bmaskdi_vis"): Likewise.
18009 ("bmasksi_vis"): Likewise.
18010 ("bshuffle<VM64:mode>_vis"): Likewise.
18011 ("cmask8<P:mode>_vis"): Likewise.
18012 ("cmask16<P:mode>_vis"): Likewise.
18013 ("cmask32<P:mode>_vis"): Likewise.
18014 ("pdistn<P:mode>_vis"): Likewise.
18015 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18017 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18019 * config/sparc/sparc.md ("subtype"): New insn attribute.
18020 ("*wrgsr_sp64"): Set insn subtype.
18021 ("*rdgsr_sp64"): Likewise.
18022 ("alignaddrsi_vis"): Likewise.
18023 ("alignaddrdi_vis"): Likewise.
18024 ("alignaddrlsi_vis"): Likewise.
18025 ("alignaddrldi_vis"): Likewise.
18026 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
18027 ("fexpand_vis"): Likewise.
18028 ("fpmerge_vis"): Likewise.
18029 ("faligndata<VM64:mode>_vis"): Likewise.
18030 ("bshuffle<VM64:mode>_vis"): Likewise.
18031 ("cmask8<P:mode>_vis"): Likewise.
18032 ("cmask16<P:mode>_vis"): Likewise.
18033 ("cmask32<P:mode>_vis"): Likewise.
18034 ("fchksm16_vis"): Likewise.
18035 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
18036 ("fmean16_vis"): Likewise.
18037 ("fp<plusminus_insn>64_vis"): Likewise.
18038 ("<plusminus_insn>v8qi3"): Likewise.
18039 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
18040 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
18041 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
18042 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
18043 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
18044 ("*movqi_insn"): Likewise.
18045 ("*movhi_insn"): Likewise.
18046 ("*movsi_insn"): Likewise.
18047 ("movsi_pic_gotdata_op"): Likewise.
18048 ("*movdi_insn_sp32"): Likewise.
18049 ("*movdi_insn_sp64"): Likewise.
18050 ("movdi_pic_gotdata_op"): Likewise.
18051 ("*movsf_insn"): Likewise.
18052 ("*movdf_insn_sp32"): Likewise.
18053 ("*movdf_insn_sp64"): Likewise.
18054 ("*zero_extendhisi2_insn"): Likewise.
18055 ("*zero_extendqihi2_insn"): Likewise.
18056 ("*zero_extendqisi2_insn"): Likewise.
18057 ("*zero_extendqidi2_insn"): Likewise.
18058 ("*zero_extendhidi2_insn"): Likewise.
18059 ("*zero_extendsidi2_insn_sp64"): Likewise.
18060 ("ldfsr"): Likewise.
18061 ("prefetch_64"): Likewise.
18062 ("prefetch_32"): Likewise.
18063 ("tie_ld32"): Likewise.
18064 ("tie_ld64"): Likewise.
18065 ("*tldo_ldub_sp32"): Likewise.
18066 ("*tldo_ldub1_sp32"): Likewise.
18067 ("*tldo_ldub2_sp32"): Likewise.
18068 ("*tldo_ldub_sp64"): Likewise.
18069 ("*tldo_ldub1_sp64"): Likewise.
18070 ("*tldo_ldub2_sp64"): Likewise.
18071 ("*tldo_ldub3_sp64"): Likewise.
18072 ("*tldo_lduh_sp32"): Likewise.
18073 ("*tldo_lduh1_sp32"): Likewise.
18074 ("*tldo_lduh_sp64"): Likewise.
18075 ("*tldo_lduh1_sp64"): Likewise.
18076 ("*tldo_lduh2_sp64"): Likewise.
18077 ("*tldo_lduw_sp32"): Likewise.
18078 ("*tldo_lduw_sp64"): Likewise.
18079 ("*tldo_lduw1_sp64"): Likewise.
18080 ("*tldo_ldx_sp64"): Likewise.
18081 ("*mov<VM32:mode>_insn"): Likewise.
18082 ("*mov<VM64:mode>_insn_sp64"): Likewise.
18083 ("*mov<VM64:mode>_insn_sp32"): Likewise.
18085 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18087 * config/sparc/sparc.md ("type"): New insn type viscmp.
18088 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
18090 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
18091 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
18092 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
18093 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
18095 ("n7_vis_logical_11cycle"): Likewise.
18096 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
18097 * config/sparc/niagara2.md ("niag3_vis": Likewise.
18098 * config/sparc/niagara.md ("niag_vis"): Likewise.
18099 * config/sparc/ultra3.md ("us3_fga"): Likewise.
18100 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
18102 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
18104 * config/sparc/sparc.md: New instruction type `bmask'.
18105 (bmaskdi_vis): Use the `bmask' type.
18106 (bmasksi_vis): Likewise.
18107 * config/sparc/ultra3.md (us3_array): Likewise.
18108 * config/sparc/niagara7.md (n7_array): Likewise.
18109 * config/sparc/niagara4.md (n4_array): Likewise.
18110 * config/sparc/niagara2.md (niag2_vis): Likewise.
18111 (niag3_vis): Likewise.
18112 * config/sparc/niagara.md (niag_vis): Likewise.
18114 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18116 * ipa-comdats.c: Remove optimize check from gate.
18117 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
18118 for functions not optimized.
18119 (ipa_fn_summary_read): Skip optimize check.
18120 (ipa_fn_summary_write): Likewise.
18121 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
18123 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
18125 (can_inline_edge_p): Check flag_pcc_struct_return for match.
18126 (check_callers): Give up on caller which is not optimized.
18127 (inline_small_functions): Likewise.
18128 (ipa_inline): Do not give up when not optimizing.
18129 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
18130 away unoptimizes cdtors.
18131 (whole_program_function_and_variable_visibility): Do
18132 ipa_discover_readonly_nonaddressable_vars in LTO mode.
18133 * ipa.c (process_references): Do not check optimize.
18134 (symbol_table::remove_unreachable_nodes): Update optimize check.
18135 (set_writeonly_bit): Update optimize check.
18136 (pass_ipa_cdtor_merge::gate): Do not check optimize.
18137 (pass_ipa_single_use::gate): Remove.
18139 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18141 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
18142 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
18143 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
18144 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
18145 permute_load, permute_store, adjust_extract, adjust_splat,
18146 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
18147 replace_swap_with_copy, dump_swap_insn_table,
18148 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
18149 recombine_lvx_pattern, recombine_stvx_pattern,
18150 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
18151 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
18152 to file rs6000-p8swap.c.
18153 * config/rs6000/rs6000-p8swap.c: New file.
18154 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
18155 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
18156 and rs6000*-*-* targets.
18158 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18160 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
18162 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18164 * lto-wrapper.c (merge_and_complain): Do not merge
18165 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
18166 fsigned_zeros, ftrapping_math, fwrapv.
18167 (append_compiler_options): Do not track these options.
18168 (append_linker_options): Likewie
18170 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18172 * cgraphunit.c (cgraph_node::finalize_function): When
18173 !flag_toplevel_reorde set no_reorder flag.
18174 (varpool_node::finalize_decl): Likewise.
18175 (symbol_table::compile): Drop no toplevel reorder path.
18177 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18179 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
18180 edges; zero probability is not better than uninitialized.
18182 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18184 * asan.h (asan_sanitize_allocas_p): Declare.
18185 * asan.c (asan_sanitize_allocas_p): New function.
18186 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
18187 (handle_builtin_alloca): Likewise.
18188 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
18189 if !asan_sanitize_allocas_p.
18190 * params.def (asan-instrument-allocas): Add new option.
18191 * params.h (ASAN_PROTECT_ALLOCAS): Define.
18192 * opts.c (common_handle_option): Disable allocas sanitization for
18195 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
18197 * asan.c: Include gimple-fold.h.
18198 (get_last_alloca_addr): New function.
18199 (handle_builtin_stackrestore): Likewise.
18200 (handle_builtin_alloca): Likewise.
18201 (asan_emit_allocas_unpoison): Likewise.
18202 (get_mem_refs_of_builtin_call): Add new parameter, remove const
18203 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
18204 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
18205 (instrument_builtin_call): Pass gimple iterator to
18206 get_mem_refs_of_builtin_call.
18207 (last_alloca_addr): New global.
18208 * asan.h (asan_emit_allocas_unpoison): Declare.
18209 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
18210 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
18211 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
18212 if function calls alloca.
18213 * gimple-fold.c (replace_call_with_value): Remove static keyword.
18214 * gimple-fold.h (replace_call_with_value): Declare.
18215 * internal-fn.c: Include asan.h.
18216 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
18217 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
18219 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18221 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
18222 (C_SELFTEST_FLAGS): New.
18223 (CPP_SELFTEST_FLAGS): New.
18224 (SELFTEST_DEPS): New, from deps of s-selftest.
18225 (C_SELFTEST_DEPS): New, from deps of s-selftest.
18226 (CPP_SELFTEST_DEPS): New.
18227 (selftest): Add dependency on s-selftest-c++.
18228 (s-selftest): Rename to...
18229 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
18230 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
18231 than SELFTEST_FLAGS.
18232 (selftest-gdb): Rename to...
18233 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
18235 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
18236 (selftest-valgrind): Rename to...
18237 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
18239 (selftest-valgrind): Reintroduce as an alias for
18240 selftest-c-valgrind.
18241 (s-selftest-c++): New.
18242 (selftest-c++-gdb): New.
18243 (selftest-c++-valgrind): New.
18245 2017-07-06 Olivier Hainque <hainque@adacore.com>
18247 * gcc.c (process_command): When deciding if undefined variables
18248 should be ignored when processing specs, accept "gcc -v" as well.
18250 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
18252 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
18253 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
18255 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
18257 * config/arm/arm-cpus.in (armv8-r): Add new entry.
18258 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
18259 * config/arm/arm-tables.opt: Regenerate.
18260 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
18262 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
18264 2017-07-06 Carl Love <cel@us.ibm.com>
18266 * ChangeLog: Clean up from mid air collision
18268 2017-07-06 Carl Love <cel@us.ibm.com>
18270 * config/rs6000/rs6000-c.c: Add support for built-in functions
18271 vector signed int vec_subc (vector signed int, vector signed int);
18272 vector signed __int128 vec_subc (vector signed __int128,
18273 vector signed __int128);
18274 vector unsigned __int128 vec_subc (vector unsigned __int128,
18275 vector unsigned __int128);
18276 vector signed int vec_sube (vector signed int, vector signed int,
18277 vector signed int);
18278 vector unsigned int vec_sube (vector unsigned int,
18279 vector unsigned int,
18280 vector unsigned int);
18281 vector signed __int128 vec_sube (vector signed __int128,
18282 vector signed __int128,
18283 vector signed__int128);
18284 vector unsigned __int128 vec_sube (vector unsigned __int128,
18285 vector unsigned __int128,
18286 vector unsigned __int128);
18287 vector signed int vec_subec (vector signed int, vector signed int,
18288 vector signed int);
18289 vector unsigned int vec_subec (vector unsigned int,
18290 vector unsigned int,
18291 vector unsigned int);
18292 vector signed __int128 vec_subec (vector signed __int128,
18293 vector signed __int128,
18294 vector signed__int128);
18295 vector unsigned __int128 vec_subec (vector unsigned __int128,
18296 vector unsigned __int128,
18297 vector unsigned __int128);
18298 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
18299 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
18300 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
18301 BU_ALTIVEC_OVERLOAD_X definitions.
18302 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
18303 * doc/extend.texi: Update the built-in documentation file for the new
18304 built-in functions.
18306 2017-07-06 David Malcolm <dmalcolm@redhat.com>
18309 * diagnostic-show-locus.c (layout::layout): Use start and finish
18310 spelling location for the start and finish of each range.
18311 * genmatch.c (linemap_client_expand_location_to_spelling_point):
18312 Add unused aspect param.
18313 * input.c (expand_location_1): Add "aspect" param, and use it
18314 to access the correct part of the location.
18315 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
18317 (expand_location_to_spelling_point): Likewise.
18318 (linemap_client_expand_location_to_spelling_point): Add "aspect"
18319 param, and pass it to expand_location_1.
18321 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
18323 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
18324 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
18325 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
18326 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
18327 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
18328 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
18329 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
18330 _mm_maskz_getmant_ss): New intrinsics.
18331 (__builtin_ia32_getexpss128_mask): Changed to ...
18332 __builtin_ia32_getexpss128_round ... this.
18333 (__builtin_ia32_getexpsd128_mask): Changed to ...
18334 __builtin_ia32_getexpsd128_round ... this.
18335 * config/i386/i386-builtin-types.def
18336 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
18337 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
18338 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
18339 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
18340 __builtin_ia32_getmantss_mask_round): New builtins.
18341 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
18342 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
18343 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
18344 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
18345 * config/i386/sse.md
18346 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
18347 avx512f_sgetexp<mode><mask_scalar_name>
18348 <round_saeonly_scalar_name> ... this.
18349 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
18350 %0, %1, %2<round_saeonly_op3>}): Changed to ...
18351 vgetexp<ssescalarmodesuffix>
18352 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18353 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
18354 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
18355 avx512f_vgetmant<mode><mask_scalar_name>
18356 <round_saeonly_scalar_name> ... this.
18357 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
18358 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
18359 vgetmant<ssescalarmodesuffix>
18360 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
18361 %0<mask_scalar_operand4>, %1, %2
18362 <round_saeonly_scalar_mask_op4>, %3} ... this.
18363 * config/i386/subst.md (mask_scalar_operand4,
18364 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
18365 round_saeonly_scalar_nimm_predicate): New subst attributes.
18367 2017-07-06 Julia Koval <julia.koval@intel.com>
18369 * config/i386/i386.c (ix86_erase_embedded_rounding):
18370 Remove code for old rounding pattern.
18372 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
18374 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
18376 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
18378 * doc/sourcebuild.texi (Test Directives, Variants of
18379 dg-require-support): Add documentation for dg-require-stack-check.
18381 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
18383 * config/i386/subst.md (mask_scalar, round_scalar,
18384 round_saeonly_scalar): New meta-templates.
18385 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
18386 round_scalar_mask_operand3, round_scalar_mask_op3,
18387 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
18388 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
18389 round_saeonly_scalar_constraint,
18390 round_saeonly_scalar_prefix): New subst attribute.
18391 * config/i386/sse.md
18392 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
18393 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
18394 <round_scalar_name> ... this.
18395 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
18396 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
18397 <round_scalar_name> ... this.
18398 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
18399 <sse>_vm<code><mode>3<mask_scalar_name>
18400 <round_saeonly_scalar_name> ... this.
18401 (v<plusminus_mnemonic><ssescalarmodesuffix>
18402 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18403 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18404 v<plusminus_mnemonic><ssescalarmodesuffix>
18405 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18406 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18407 (v<multdiv_mnemonic><ssescalarmodesuffix>
18408 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
18409 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
18410 v<multdiv_mnemonic><ssescalarmodesuffix>
18411 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18412 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
18413 (v<maxmin_float><ssescalarmodesuffix>
18414 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
18415 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
18416 v<maxmin_float><ssescalarmodesuffix>
18417 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
18418 %0<mask_scalar_operand3>, %1, %<iptr>2
18419 <round_saeonly_scalar_mask_op3>} ... this.
18421 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
18423 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
18424 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
18426 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
18427 Alan Hayward <alan.hayward@arm.com>
18428 David Sherwood <david.sherwood@arm.com>
18430 * combine.c (simplify_if_then_else): Remove "enum" before
18432 * compare-elim.c (can_eliminate_compare): Likewise.
18433 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
18435 (aarch64_lookup_simd_builtin_type): Likewise.
18436 (aarch64_simd_builtin_type): Likewise.
18437 (aarch64_init_simd_builtin_types): Likewise.
18438 (aarch64_simd_expand_args): Likewise.
18439 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
18441 (aarch64_reverse_mask): Likewise.
18442 (aarch64_simd_emit_reg_reg_move): Likewise.
18443 (aarch64_gen_adjusted_ldpstp): Likewise.
18444 (aarch64_ccmp_mode_to_code): Likewise.
18445 (aarch64_operands_ok_for_ldpstp): Likewise.
18446 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18447 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18449 (aarch64_min_divisions_for_recip_mul): Likewise.
18450 (aarch64_reassociation_width): Likewise.
18451 (aarch64_get_condition_code_1): Likewise.
18452 (aarch64_simd_emit_reg_reg_move): Likewise.
18453 (aarch64_simd_attr_length_rglist): Likewise.
18454 (aarch64_reverse_mask): Likewise.
18455 (aarch64_operands_ok_for_ldpstp): Likewise.
18456 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
18457 (aarch64_gen_adjusted_ldpstp): Likewise.
18458 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
18460 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
18461 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
18462 (arm_lookup_simd_builtin_type): Likewise.
18463 (arm_simd_builtin_type): Likewise.
18464 (arm_init_simd_builtin_types): Likewise.
18465 (arm_expand_builtin_args): Likewise.
18466 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
18467 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
18468 (ft32_setup_incoming_varargs): Likewise.
18469 (ft32_function_arg): Likewise.
18470 (ft32_function_arg_advance): Likewise.
18471 (ft32_pass_by_reference): Likewise.
18472 (ft32_arg_partial_bytes): Likewise.
18473 (ft32_valid_pointer_mode): Likewise.
18474 (ft32_addr_space_pointer_mode): Likewise.
18475 (ft32_addr_space_legitimate_address_p): Likewise.
18476 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
18478 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
18479 (ix86_emit_outlined_ms2sysv_restore): Likewise.
18480 (iamcu_alignment): Likewise.
18481 (canonicalize_vector_int_perm): Likewise.
18482 (ix86_noce_conversion_profitable_p): Likewise.
18483 (ix86_mpx_bound_mode): Likewise.
18484 (ix86_operands_ok_for_move_multiple): Likewise.
18485 * config/microblaze/microblaze-protos.h
18486 (microblaze_expand_conditional_branch_reg): Likewise.
18487 * config/microblaze/microblaze.c
18488 (microblaze_expand_conditional_branch_reg): Likewise.
18489 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
18491 (rs6000_reassociation_width): Likewise.
18492 (rs6000_invalid_binary_op): Likewise.
18493 (fusion_p9_p): Likewise.
18494 (emit_fusion_p9_load): Likewise.
18495 (emit_fusion_p9_store): Likewise.
18496 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
18498 (riscv_hard_regno_mode_ok_p): Likewise.
18499 (riscv_address_insns): Likewise.
18500 (riscv_split_symbol): Likewise.
18501 (riscv_legitimize_move): Likewise.
18502 (riscv_function_value): Likewise.
18503 (riscv_hard_regno_nregs): Likewise.
18504 (riscv_expand_builtin): Likewise.
18505 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
18506 (riscv_build_integer): Likewise.
18507 (riscv_split_integer): Likewise.
18508 (riscv_legitimate_constant_p): Likewise.
18509 (riscv_cannot_force_const_mem): Likewise.
18510 (riscv_regno_mode_ok_for_base_p): Likewise.
18511 (riscv_valid_base_register_p): Likewise.
18512 (riscv_valid_offset_p): Likewise.
18513 (riscv_valid_lo_sum_p): Likewise.
18514 (riscv_classify_address): Likewise.
18515 (riscv_legitimate_address_p): Likewise.
18516 (riscv_address_insns): Likewise.
18517 (riscv_load_store_insns): Likewise.
18518 (riscv_force_binary): Likewise.
18519 (riscv_split_symbol): Likewise.
18520 (riscv_force_address): Likewise.
18521 (riscv_legitimize_address): Likewise.
18522 (riscv_move_integer): Likewise.
18523 (riscv_legitimize_const_move): Likewise.
18524 (riscv_legitimize_move): Likewise.
18525 (riscv_address_cost): Likewise.
18526 (riscv_subword): Likewise.
18527 (riscv_output_move): Likewise.
18528 (riscv_canonicalize_int_order_test): Likewise.
18529 (riscv_emit_int_order_test): Likewise.
18530 (riscv_function_arg_boundary): Likewise.
18531 (riscv_pass_mode_in_fpr_p): Likewise.
18532 (riscv_pass_fpr_single): Likewise.
18533 (riscv_pass_fpr_pair): Likewise.
18534 (riscv_get_arg_info): Likewise.
18535 (riscv_function_arg): Likewise.
18536 (riscv_function_arg_advance): Likewise.
18537 (riscv_arg_partial_bytes): Likewise.
18538 (riscv_function_value): Likewise.
18539 (riscv_pass_by_reference): Likewise.
18540 (riscv_setup_incoming_varargs): Likewise.
18541 (riscv_print_operand): Likewise.
18542 (riscv_elf_select_rtx_section): Likewise.
18543 (riscv_save_restore_reg): Likewise.
18544 (riscv_for_each_saved_reg): Likewise.
18545 (riscv_register_move_cost): Likewise.
18546 (riscv_hard_regno_mode_ok_p): Likewise.
18547 (riscv_hard_regno_nregs): Likewise.
18548 (riscv_class_max_nregs): Likewise.
18549 (riscv_memory_move_cost): Likewise.
18550 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
18551 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
18552 (rl78_addr_space_address_mode): Likewise.
18553 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
18555 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
18556 (rs6000_reassociation_width): Likewise.
18557 (rs6000_invalid_binary_op): Likewise.
18558 (fusion_p9_p): Likewise.
18559 (emit_fusion_p9_load): Likewise.
18560 (emit_fusion_p9_store): Likewise.
18561 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
18562 (ok_for_simple_move_operands): Likewise.
18563 (ok_for_simple_move_strict_operands): Likewise.
18564 (ok_for_simple_arith_logic_operands): Likewise.
18565 (visium_legitimize_reload_address): Likewise.
18566 (visium_select_cc_mode): Likewise.
18567 (output_cbranch): Likewise.
18568 (visium_split_double_move): Likewise.
18569 (visium_expand_copysign): Likewise.
18570 (visium_expand_int_cstore): Likewise.
18571 (visium_expand_fp_cstore): Likewise.
18572 * config/visium/visium.c (visium_pass_by_reference): Likewise.
18573 (visium_function_arg): Likewise.
18574 (visium_function_arg_advance): Likewise.
18575 (visium_libcall_value): Likewise.
18576 (visium_setup_incoming_varargs): Likewise.
18577 (visium_legitimate_constant_p): Likewise.
18578 (visium_legitimate_address_p): Likewise.
18579 (visium_legitimize_address): Likewise.
18580 (visium_secondary_reload): Likewise.
18581 (visium_register_move_cost): Likewise.
18582 (visium_memory_move_cost): Likewise.
18583 (prepare_move_operands): Likewise.
18584 (ok_for_simple_move_operands): Likewise.
18585 (ok_for_simple_move_strict_operands): Likewise.
18586 (ok_for_simple_arith_logic_operands): Likewise.
18587 (visium_function_value_1): Likewise.
18588 (rtx_ok_for_offset_p): Likewise.
18589 (visium_legitimize_reload_address): Likewise.
18590 (visium_split_double_move): Likewise.
18591 (visium_expand_copysign): Likewise.
18592 (visium_expand_int_cstore): Likewise.
18593 (visium_expand_fp_cstore): Likewise.
18594 (visium_split_cstore): Likewise.
18595 (visium_select_cc_mode): Likewise.
18596 (visium_split_cbranch): Likewise.
18597 (output_cbranch): Likewise.
18598 (visium_print_operand_address): Likewise.
18599 * expmed.c (flip_storage_order): Likewise.
18600 * expmed.h (emit_cstore): Likewise.
18601 (flip_storage_order): Likewise.
18602 * genrecog.c (validate_pattern): Likewise.
18603 * hsa-gen.c (gen_hsa_addr): Likewise.
18604 * internal-fn.c (expand_arith_overflow): Likewise.
18605 * ira-color.c (allocno_copy_cost_saving): Likewise.
18606 * lra-assigns.c (find_hard_regno_for_1): Likewise.
18607 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
18608 (process_invariant_for_inheritance): Likewise.
18609 * lra-eliminations.c (move_plus_up): Likewise.
18610 * omp-low.c (lower_oacc_reductions): Likewise.
18611 * simplify-rtx.c (simplify_subreg): Likewise.
18612 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
18613 (TARGET_CHKP_BOUND_MODE): Likewise..
18614 * targhooks.c (default_chkp_bound_mode): Likewise.
18615 (default_setup_incoming_vararg_bounds): Likewise.
18616 * targhooks.h (default_chkp_bound_mode): Likewise.
18617 (default_setup_incoming_vararg_bounds): Likewise.
18618 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
18619 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
18620 (have_whole_vector_shift): Likewise.
18621 * tree-vect-stmts.c (vectorizable_load): Likewise.
18622 * doc/tm.texi: Regenerate.
18624 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18626 Graceful degrade if Binutils PR21472 is not available.
18629 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
18630 .rodata in flash test fails.
18631 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
18632 * confgure: Regenerate.
18633 * config.in: Regenerate.
18634 * config/avr/avr.c (avr_asm_named_section)
18635 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
18636 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
18637 (avr_asm_init_sections): Same.
18639 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18641 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
18642 (fma<VH:mode>4_intrinsic): Likewise.
18643 (*fmsub<VCVTF:mode>4): Likewise.
18644 (*fmsub<VH:mode>4_intrinsic): Likewise.
18646 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
18649 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
18650 Don't depend on "optimize > 0".
18651 (out_movhi_r_mr, out_movqi_mr_r): Same.
18652 (out_movhi_mr_r, out_movqi_r_mr): Same.
18653 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
18654 io_address_operand on "optimize > 0".
18656 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18658 * tree-loop-distribution.c: Add general explanantion on the pass.
18659 (generate_loops_for_partition): Mark distributed loop.
18660 (pg_add_dependence_edges): New parameter. Handle alias data
18661 dependence specially and record it in the parameter if asked.
18662 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
18663 (init_partition_graph_vertices, add_partition_graph_edge): New.
18664 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
18665 (free_partition_graph_vdata, build_partition_graph): New.
18666 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
18667 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
18668 (data_ref_segment_size, latch_dominated_by_data_ref): New.
18669 (compute_alias_check_pairs, version_loop_by_alias_check): New.
18670 (version_for_distribution_p, finalize_partitions): New.
18671 (distribute_loop): Handle alias data dependence specially. Factor
18672 out loop fusion code as functions and call these functions.
18674 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18676 * tree-loop-distribution.c (classify_partition): New parameter and
18677 better handle reduction statement.
18678 (rdg_build_partitions): Revise comment.
18679 (distribute_loop): Compute statements in all partitions and pass it
18680 to classify_partition.
18682 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18684 * tree-loop-distribution.c (enum partition_type): New.
18685 (struct partition): New field type.
18686 (partition_merge_into): Add parameter. Update partition type.
18687 (data_dep_in_cycle_p, update_type_for_merge): New functions.
18688 (build_rdg_partition_for_vertex): Compute partition type.
18689 (rdg_build_partitions): Dump partition type.
18690 (distribute_loop): Update calls to partition_merge_into.
18692 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18694 * tree-loop-distribution.c (struct ddr_hasher): New.
18695 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
18697 (classify_partition): Call get_data_dependence.
18698 (pg_add_dependence_edges): Ditto.
18699 (distribute_loop): Release data dependence hash table.
18701 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18703 * tree-loop-distribution.c (ref_base_address): Delete.
18704 (similar_memory_accesses): Rename ...
18705 (share_memory_accesses): ... to this. Check if partitions access
18706 the same memory reference.
18707 (distribute_loop): Call share_memory_accesses.
18709 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18711 * tree-loop-distribution.c (struct partition): New field recording
18712 its data reference.
18713 (partition_alloc, partition_free): Init and release data refs.
18714 (partition_merge_into): Merge data refs.
18715 (build_rdg_partition_for_vertex): Collect data refs for partition.
18716 (pg_add_dependence_edges): Change parameters from vector to bitmap.
18718 (distribute_loop): Remve data refs from vertice data of partition
18721 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18723 * tree-loop-distribution.c (params.h): Include header file.
18724 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
18725 (datarefs_vec): New global var.
18726 (create_rdg_vertices): Use datarefs_vec directly.
18727 (free_rdg): Don't free data references.
18728 (build_rdg): Update use. Don't free data references.
18729 (distribute_loop): Compute global variable for data references.
18730 Bail out if there are too many data references.
18732 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18734 * tree-loop-distribution.c (loop_nest): New global var.
18735 (build_rdg): Use loop directly, rather than loop nest.
18736 (pg_add_dependence_edges): Remove loop nest parameter. Use global
18738 (distribute_loop): Compute global variable loop nest. Update use.
18740 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18742 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
18743 (partition_merge_into): New parameter. Dump reason for fusion.
18744 (distribute_loop): Update use of partition_merge_into.
18746 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18748 * tree-loop-distribution.c (bb_top_order_index): New.
18749 (bb_top_order_index_size, bb_top_order_cmp): New.
18750 (stmts_from_loop): Use topological order.
18751 (pass_loop_distribution::execute): Compute and release topological
18752 order for basic blocks.
18754 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18756 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
18759 2017-07-05 Bin Cheng <bin.cheng@arm.com>
18761 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
18762 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
18763 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
18764 * internal-fn.def (LOOP_DIST_ALIAS): New.
18765 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
18766 (fold_loop_internal_call): ... this.
18767 (vect_loop_dist_alias_call): New function.
18768 (set_uid_loop_bbs): Call fold_loop_internal_call.
18769 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
18772 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18775 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
18776 Require dead FLAGS_REG at the beginning of a peephole.
18778 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
18781 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
18782 arguments in the call to __builtin_ia32_sbb_u32.
18783 (_subborrow_u64): Swap _X and _Y arguments in the call to
18784 __builtin_ia32_sbb_u64.
18786 2017-07-04 Jakub Jelinek <jakub@redhat.com>
18789 * tree-vrp.c (compare_assert_loc): Turn into a function template
18790 with stable template parameter. Only test if a->e is NULL,
18791 !a->e == !b->e has been verified already. Use e == NULL or
18792 e != NULL instead of e or ! e tests. If stable is true, don't use
18793 iterative_hash_expr, on the other side allow a or b or both NULL
18794 and sort the NULLs last.
18795 (process_assert_insertions): Sort using compare_assert_loc<false>
18796 instead of compare_assert_loc, later sort using
18797 compare_assert_loc<true> before calling process_assert_insertions_for
18798 in a loop. Use break instead of continue once seen NULL pointer.
18800 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
18802 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
18803 Cortex-R7 and Cortex-R8 processors.
18805 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18807 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
18808 uninitialized while src is not.
18810 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
18812 * common/config/arm/arm-common.c: Adjust include path for
18814 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
18815 (arm-cpu.h): Create in build directory. Adjust dependency rules.
18816 (arm-cpu-data.h): Likewise.
18817 (arm-cpu-cdata.h): Likewise.
18818 * config/arm/arm-cpu.h: Delete.
18819 * config/arm/arm-cpu-cdata.h: Delete.
18820 * config/arm/arm-cpu-data.h: Delete.
18822 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
18824 * config/arm/arm-cpus.in (cortex-a55): New.
18825 (cortex-a75): Likewise.
18826 (cortex-a75.cortex-a55): Likewise.
18827 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
18829 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
18830 * config/arm/arm-cpu-cdata.h: Regenerate.
18831 * config/arm/arm-cpu-data.h: Regenerate.
18832 * config/arm/arm-cpu.h: Regenerate.
18833 * config/arm/arm-tables.opt: Regenerate.
18834 * config/arm/arm-tune.md: Regenerate.
18836 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18838 * haifa-sched.c (sched_create_recovery_edges): Update profile.
18840 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
18842 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
18845 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
18847 PR tree-optimization/81292
18848 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
18849 full_string_p, also call adjust_related_strinfos if the adjustment
18850 is simple, otherwise invalidate related strinfos.
18852 2017-07-04 Martin Liska <mliska@suse.cz>
18855 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
18856 newly created variable as DECL_IGNORED_P.
18858 2017-07-04 Martin Liska <mliska@suse.cz>
18861 * ipa-inline.c (inline_small_functions):
18862 Use xstrdup_for_dump.
18864 2017-07-04 Tom de Vries <tom@codesourcery.com>
18866 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
18868 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18871 * config/darwin.c (darwin_function_switched_text_sections):
18874 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
18876 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
18878 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
18880 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
18882 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18884 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
18885 min_profitable_iters, and th as inclusive lower bounds.
18886 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
18887 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
18888 for min_profitable_iters and min_profitable_estimate.
18889 (vect_transform_loop): Treat th as an inclusive lower bound.
18890 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
18892 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
18895 * config/darwin.c (darwin_function_switched_text_sections):
18896 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
18897 in two pieces, and suppress the use of buf.
18899 2017-07-03 Nathan Sidwell <nathan@acm.org>
18901 * hash-table.h (hash_table_mod1): Fix indentation.
18903 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
18905 PR middle-end/81290
18906 * predict.c (force_edge_cold): Be more careful about propagation
18908 * profile-count.h (profile_probability::guessed,
18909 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
18911 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
18913 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
18915 * doc/invoke.texi (rcpc architecture extension): Document it.
18917 2017-07-03 Richard Biener <rguenther@suse.de>
18919 PR tree-optimization/60510
18920 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
18921 the scalar reduction PHI and use it.
18922 (vectorizable_reduction): Properly guard the single_defuse_cycle
18923 path for non-SLP reduction chains where we cannot use it.
18924 Rework reduc_def/index and vector type deduction. Rework
18925 vector operand gathering during reduction op code-gen.
18926 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
18927 chains dissolve the chain and leave it to non-SLP reduction
18930 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18932 * tree-data-ref.h (dr_alignment): Declare.
18933 * tree-data-ref.c (dr_alignment): New function.
18934 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
18935 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
18937 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
18939 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18941 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
18942 and base_misalignment fields.
18943 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
18944 * tree-data-ref.c: Include builtins.h.
18945 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
18946 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
18947 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
18948 * tree-vect-data-refs.c: Include tree-cfg.h.
18949 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
18950 fields instead of calculating an alignment here.
18951 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
18952 innermost_loop_behavior fields.
18954 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18956 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
18958 (DR_STEP_ALIGNMENT): New macro.
18959 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
18960 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
18961 (create_data_ref): Print it.
18962 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
18963 to tell whether the step preserves vector (mis)alignment.
18964 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18965 Move the check for an integer step and generalise to all INTEGER_CST.
18966 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
18967 Print the outer step alignment.
18969 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18971 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
18972 with offset_alignment.
18973 (DR_ALIGNED_TO): Delete.
18974 (DR_OFFSET_ALIGNMENT): New macro.
18975 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
18976 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
18977 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
18978 (create_data_ref): Likewise.
18979 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
18980 (vect_analyze_data_refs): Likewise.
18981 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
18982 creating dummy innermost behavior.
18984 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18986 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
18987 with a "innermost_loop_behavior *" and refeence tree.
18988 * tree-data-ref.c (dr_analyze_innermost): Likewise.
18989 (create_data_ref): Update call accordingly.
18990 * tree-predcom.c (find_looparound_phi): Likewise.
18992 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
18994 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
18995 fields with dr_wrt_vec_loop.
18996 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
18997 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
18998 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
18999 (vect_dr_behavior): New function.
19000 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19001 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
19002 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
19003 track whether the step preserves the misalignment.
19004 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
19005 Use vect_dr_behavior.
19006 (vect_setup_realignment): Update call accordingly.
19007 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
19008 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
19009 call to vect_create_addr_base_for_vector_ref.
19010 (vect_create_cond_for_align_checks): Likewise.
19011 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
19012 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
19013 (vect_recog_mask_conversion_pattern): Likewise.
19014 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
19015 (new_stmt_vec_info): Remove redundant zeroing.
19017 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
19019 * common/config/arm/arm-common.c (arm_be8_option): New function.
19020 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
19021 (ISA_ARMv6): Add isa_bit_be8.
19022 * config/arm/arm.h (arm_be8_option): Add prototype.
19023 (BE8_SPEC_FUNCTION): New define.
19024 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
19025 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
19026 (mlittle-endian): Similarly.
19027 (mbe8, mbe32): New options.
19028 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
19029 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
19031 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19033 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
19035 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19037 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
19038 (cleanup_tree_cfg_bb): Use it.
19039 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
19041 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
19043 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19046 * loop-doloop.c (add_test): Update profile.
19048 2017-07-03 Martin Liska <mliska@suse.cz>
19051 * sanopt.c (rewrite_usage_of_param): New function.
19052 (sanitize_rewrite_addressable_params): Likewise.
19053 (pass_sanopt::execute): Call rewrite_usage_of_param.
19055 2017-07-03 Richard Biener <rguenther@suse.de>
19057 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
19058 back to using VIEW_CONVERT_EXPR.
19060 2017-07-03 Martin Liska <mliska@suse.cz>
19063 * doc/extend.texi: Document when a resolver function is
19064 generated for target_clones.
19066 2017-07-03 Martin Liska <mliska@suse.cz>
19068 * asan.c (asan_emit_stack_protection): Unpoison just red zones
19069 and shadow memory of auto variables which are subject of
19070 use-after-scope sanitization.
19071 (asan_expand_mark_ifn): Add do set only when is_poison.
19073 2016-07-03 Richard Biener <rguenther@suse.de>
19075 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
19077 (vect_force_simple_reduction): Record reduction def -> phi mapping.
19078 (vectorizable_reduction): Perform reduction PHI creation when
19079 visiting a reduction PHI and adjust and simplify code generation
19080 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
19081 (vect_transform_loop): Visit reduction PHIs.
19082 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
19083 defs into the SLP tree.
19084 (vect_build_slp_tree): Reduction defs terminate the recursion.
19085 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
19087 (vect_get_vec_defs_for_stmt_copy): Export.
19088 (vect_get_vec_defs): Likewise.
19089 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
19091 (vect_get_vec_defs_for_stmt_copy): Declare.
19092 (vect_get_vec_defs): Likewise.
19094 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
19096 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
19097 parameter with a "loop" parameter and use it instead of the
19098 loop containing DR_STMT. Don't check simple_iv when doing
19099 BB analysis. Describe the two analysis modes in the comment.
19101 2017-07-03 Tom de Vries <tom@codesourcery.com>
19103 PR tree-optimization/69468
19104 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
19105 (find_same_succ_bb): Handle ignore_edge_flags.
19107 2017-07-03 Tom de Vries <tom@codesourcery.com>
19109 PR tree-optimization/81192
19110 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
19112 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
19114 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
19116 2017-07-03 Tom de Vries <tom@codesourcery.com>
19118 PR tree-optimization/81192
19119 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
19120 BB_SAME_SUCC (bb) == NULL.
19122 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19124 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
19127 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19129 * dumpfile.c: Include profile-count.h
19130 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
19132 (insert_cond_bb): Update profile.
19133 * tree-cfg.h (insert_cond_bb): Update prototype.
19134 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
19135 * tree-dump.c: Do not include tree-cfg.
19137 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19139 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
19141 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19143 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
19146 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
19148 * tree-complex.c (expand_complex_div_wide): update profile.
19150 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19151 Alan Hayward <alan.hayward@arm.com>
19152 David Sherwood <david.sherwood@arm.com>
19154 * Makefile.in (MACHMODE_H): Remove insn-modes.h
19155 (CORETYPES_H): New define.
19156 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
19157 (insn-modes-inline.h, s-modes-inline-h): New rules.
19158 (generated_files): Add insn-modes-inline.h.
19159 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
19160 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
19161 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
19162 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
19163 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
19164 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
19165 (build/gencodes.o, build/genconditions.o): Likewise.
19166 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
19167 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
19168 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
19169 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
19170 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
19171 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
19172 * coretypes.h: Include everything up to real.h for generators.
19173 Include insn-modes.h first. Include wide-int-print.h after
19174 wide-int.h. Include insn-modes-inline.h and then machmode.h.
19175 * machmode.h: Don't include insn-modes.h here.
19176 * function-tests.c: Remove includes of signop.h, machmode.h,
19177 double-int.h and wide-int.h.
19179 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
19181 * optc-save-gen.awk: Likewise.
19182 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
19183 * godump.c: Remove include of wide-int-print.h.
19184 * pretty-print.h: Likewise.
19185 * wide-int-print.cc: Likewise.
19186 * wide-int.cc: Likewise.
19187 * hash-map-tests.c: Remove include of signop.h.
19188 * hash-set-tests.c: Likewise.
19189 * rtl-tests.c: Likewise.
19190 * mkconfig.sh: Remove include of machmode.h.
19191 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
19193 (emit_insn_modes_inline_h): ...this new function. Emit the code
19194 into an insn-modes-inline.h header file, adding appropriate
19195 include guards and end comments.
19196 (emit_insn_modes_c_header): Remove include of machmode.h.
19197 (emit_min_insn_modes_c_header): Include coretypes.h rather than
19199 (main): Handle -i flag and call emit_insn_modes_inline_h when
19202 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19204 * tree-ssa-strlen.c (strinfo): Rename the length field to
19205 nonzero_chars. Add a full_string_p field.
19206 (compare_nonzero_chars, zero_length_string_p): New functions.
19207 (get_addr_stridx): Add an offset_out parameter.
19208 Use compare_nonzero_chars.
19209 (get_stridx): Update accordingly. Use compare_nonzero_chars.
19210 (new_strinfo): Update after above changes to strinfo.
19211 (set_endptr_and_length): Set full_string_p.
19212 (get_string_length): Update after above changes to strinfo.
19213 (unshare_strinfo): Update call to new_strinfo.
19214 (maybe_invalidate): Likewise.
19215 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
19216 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
19217 as a uhwi instead of an shwi. Update after above changes to
19218 strinfo and new_strinfo.
19219 (zero_length_string): Assert that chainsi contains full strings.
19220 Use zero_length_string_p. Update call to new_strinfo.
19221 (adjust_related_strinfos): Update after above changes to strinfo.
19222 Copy full_string_p from origsi.
19223 (adjust_last_stmt): Use zero_length_string_p.
19224 (handle_builtin_strlen): Update after above changes to strinfo and
19225 new_strinfo. Install the lhs as the string length if the previous
19226 entry didn't describe a full string.
19227 (handle_builtin_strchr): Update after above changes to strinfo
19229 (handle_builtin_strcpy): Likewise.
19230 (handle_builtin_strcat): Likewise.
19231 (handle_builtin_malloc): Likewise.
19232 (handle_pointer_plus): Likewise.
19233 (handle_builtin_memcpy): Likewise. Track nonzero characters
19234 that aren't necessarily followed by a nul terminator.
19235 (handle_char_store): Likewise.
19237 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19239 PR tree-optimization/80769
19240 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
19241 for malloc and calloc. Document the new invariant that all related
19242 strinfos have delayed lengths or none do.
19243 (verify_related_strinfos): Move earlier in file.
19244 (set_endptr_and_length): New function, split out from...
19245 (get_string_length): ...here. Also set the lengths of related
19247 (zero_length_string): Assert that chainsi has known (rather than
19249 (adjust_related_strinfos): Likewise.
19251 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
19253 PR tree-optimization/81136
19254 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
19255 assert that two references with the same misalignment have the same
19256 compile-time misalignment if those compile-time misalignments
19259 2017-07-01 Andi Kleen <ak@linux.intel.com>
19261 * print-tree.c (print_node): Print all attributes.
19263 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19265 * cfg.c (scale_bbs_frequencies): New function.
19266 * cfg.h (scale_bbs_frequencies): Declare it.
19267 * cfgloopanal.c (single_likely_exit): Cleanup.
19268 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
19270 (scale_loop_profile): Likewise.
19271 (loop_version): Likewise.
19272 (create_empty_loop_on_edge): Update.
19273 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
19274 scale_loop_frequencies, scale_loop_profile, loopify,
19275 loop_version): Update prototypes.
19276 * modulo-sched.c (sms_schedule): Update.
19277 * predict.c (unlikely_executed_edge_p): Also check probability.
19278 (probably_never_executed_edge_p): Fix typo.
19279 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19280 * tree-parloops.c (gen_parallel_loop): Update.
19281 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
19282 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19283 * tree-ssa-loop-split.c (split_loop): Update.
19284 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19285 * tree-vect-loop-manip.c (vect_do_peeling): Update.
19286 (vect_loop_versioning): Update.
19287 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19289 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19291 * trans-mem.c (split_bb_make_tm_edge): Update profile.
19293 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19295 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
19296 to keep profile consistent.
19298 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19300 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
19301 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
19302 * profile-count.h (max_safe_multiplier): Make unsigned.
19303 (profile_count::guessed_zero): New.
19305 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19307 * bb-reorder.c (fix_up_crossing_landing_pad,
19308 fix_crossing_conditional_branches): Use make_single_succ_edge
19309 to keep profile consistent.
19311 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
19313 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
19316 2017-07-01 Jakub Jelinek <jakub@redhat.com>
19319 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
19320 the right scopes, make sure cond_jump isn't preserved between multiple
19321 iterations. Search for fallthru edge whenever there are 3+ edges and
19322 use find_fallthru_edge for it.
19324 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19326 Patch by Alexander Monakov <amonakov@ispras.ru>
19327 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
19328 probabilities consistently.
19330 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19332 * pa.c (pa_expand_compare_and_swap_loop): Update call of
19333 emit_cmp_and_jump_insns.
19335 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19338 * tree-inline.c (expand_call_inline): Combine profile statuses.
19340 2017-06-30 Andrew Pinski <apinski@cavium.com>
19342 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
19343 fold_stmt returned true.
19345 2017-06-30 Nathan Sidwell <nathan@acm.org>
19347 * ggc.h (empty_string): Delete.
19348 * cfgexpand.c (expand_asm_stmt): Use plain "".
19349 * optabs.c (expand_asm_memory_barrier): Likewise.
19350 * stringpool.c (empty_string): Delete.
19351 (digit_vector, digit_string): Delete.
19352 (ggc_alloc_string): Use plain "", don't optimize single digit
19353 strings. Use ggc_alloc_atomic.
19355 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
19357 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
19358 comparison set and one other set, use the cost of the non-comparison
19361 2017-06-30 Nathan Sidwell <nathan@acm.org>
19363 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
19366 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
19368 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
19369 loops. Remove now unneeded calls to gimple_switch_set_label() that
19370 just set removed labels to NULL_TREE.
19372 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
19374 * tree-ssanames.c (set_range_info_raw): Abstract from ...
19375 (set_range_info): ...here. Only call set_range_info_raw if domain
19377 (set_nonzero_bits): Call set_range_info_raw.
19378 * tree-ssanames.h (set_range_info_raw): New.
19380 2017-06-30 Jakub Jelinek <jakub@redhat.com>
19383 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
19384 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
19385 of nonimmediate_operand and <store_mask_constraint> instead of m
19386 for the input operand. For V8FI iterator, always split if input
19387 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
19388 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
19389 <store_mask_predicate> instead of register_operand and
19390 <store_mask_constraint> instead of v for the input operand. Make
19391 sure both operands aren't MEMs for if not <mask_applied>.
19393 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
19395 * lto-wrapper.c (copy_file) Close both file descriptors before
19398 2017-06-30 Martin Liska <mliska@suse.cz>
19401 * multiple_target.c (create_dispatcher_calls): Make ifunc
19402 also for function that don't have calls or are not referenced.
19404 2017-06-30 Richard Biener <rguenther@suse.de>
19406 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
19407 analyze the first scalar stmt. Move vector type computation
19408 for the BB case here from ...
19409 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
19410 live operation processing in the SLP case properly.
19412 2017-06-30 Richard Biener <rguenther@suse.de>
19414 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
19416 2017-06-30 Martin Liska <mliska@suse.cz>
19419 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
19420 before BUILT_IN_UNWIND_RESUME when ASAN is used.
19422 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
19424 * doc/invoke.texi (AArch64): Add missing options and remove redundant
19427 2017-06-30 Richard Biener <rguenther@suse.de>
19429 PR tree-optimization/81249
19430 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
19431 condition reduction result to original scalar type.
19433 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19435 * profile-count.h (enum profile_quality): Fix typos and whitespace
19438 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19440 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
19441 type for branch probabilities.
19443 2017-06-29 Julian Brown <julian@codesourcery.com>
19444 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19446 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
19447 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
19448 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
19449 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
19451 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19453 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
19454 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
19455 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
19456 CC usage from generic code to here.
19457 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
19458 CC usage into the target macros.
19460 2017-06-29 Maya Rashish <coypu@sdf.org>
19462 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
19465 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19467 * arm/arm-builtins.c: Include profile-count.h
19468 * except.c (sjlj_emit_function_enter): Use
19469 profile_probability::unlikely.
19471 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19473 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
19474 and tocrel_offset be pointer args rather than implicitly using
19476 (legitimate_constant_pool_address_p, rs6000_emit_move,
19477 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
19478 tocrel_offset and use in toc_relative_expr_p call.
19479 (print_operand, print_operand_address): Use static tocrel_base_oac
19480 and tocrel_offset_oac.
19481 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
19484 2017-06-29 Maya Rashish <coypu@sdf.org>
19486 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
19488 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
19490 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
19491 objects, take into account only the alignment of 'op0' and 'mode1' if
19494 2017-06-29 Steve Ellcey <sellcey@cavium.com>
19496 * ccmp.c (ccmp_tree_comparison_p): New function.
19497 (ccmp_candidate_p): Update to use above function.
19498 (get_compare_parts): New function.
19499 (expand_ccmp_next): Update to use new functions.
19500 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
19502 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
19503 take mode as argument.
19504 * ccmp.h (expand_ccmp_expr): Add mode as argument.
19505 * expr.c (expand_expr_real_1): Pass mode as argument.
19507 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
19509 * combine.c (combine_instructions): Print insns to dump_file, together
19512 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
19514 * asan.c (asan_emit_stack_protection): Update.
19515 (create_cond_insert_point): Update.
19516 * auto-profile.c (afdo_propagate_circuit): Update.
19517 * basic-block.h (struct edge_def): Turn probability to
19518 profile_probability.
19519 (EDGE_FREQUENCY): Update.
19520 * bb-reorder.c (find_traces_1_round): Update.
19521 (better_edge_p): Update.
19522 (sanitize_hot_paths): Update.
19523 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
19524 (make_single_succ_edge): Update.
19525 (check_bb_profile): Update.
19526 (dump_edge_info): Update.
19527 (update_bb_profile_for_threading): Update.
19528 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
19529 probabilitycount to 0.
19530 * cfgbuild.c (compute_outgoing_frequencies): Update.
19531 * cfgcleanup.c (try_forward_edges): Update.
19532 (outgoing_edges_match): Update.
19533 (try_crossjump_to_edge): Update.
19534 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
19535 (expand_gimple_tailcall): Update.
19536 (construct_init_block): Use make_single_succ_edge.
19537 (construct_exit_block): Use make_single_succ_edge.
19538 * cfghooks.c (verify_flow_info): Update.
19539 (redirect_edge_succ_nodup): Update.
19540 (split_edge): Update.
19541 (account_profile_record): Update.
19542 * cfgloopanal.c (single_likely_exit): Update.
19543 * cfgloopmanip.c (scale_loop_profile): Update.
19544 (set_zero_probability): Remove.
19545 (duplicate_loop_to_header_edge): Update.
19546 * cfgloopmanip.h (loop_version): Update prototype.
19547 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
19548 (force_nonfallthru_and_redirect): Update.
19549 (update_br_prob_note): Update.
19550 (rtl_verify_edges): Update.
19551 (purge_dead_edges): Update.
19552 (rtl_lv_add_condition_to_bb): Update.
19553 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
19554 * cgraphunit.c (init_lowered_empty_function): Update.
19555 (cgraph_node::expand_thunk): Update.
19556 * cilk-common.c: Include profile-count.h
19557 * dojump.c (inv): Remove.
19558 (jumpifnot): Update.
19559 (jumpifnot_1): Update.
19560 (do_jump_1): Update.
19562 (do_jump_by_parts_greater_rtx): Update.
19563 (do_compare_rtx_and_jump): Update.
19564 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
19565 do_jump_1. do_compare_rtx_and_jump): Update prototype.
19566 * dwarf2cfi.c: Include profile-count.h
19567 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
19568 (sjlj_emit_dispatch_table): Likewise.
19569 * explow.c: Include profile-count.h
19570 * expmed.c (emit_store_flag_force): Update.
19571 (do_cmp_and_jump): Update.
19572 * expr.c (compare_by_pieces_d::generate): Update.
19573 (compare_by_pieces_d::finish_mode): Update.
19574 (emit_block_move_via_loop): Update.
19575 (store_expr_with_bounds): Update.
19576 (store_constructor): Update.
19577 (expand_expr_real_2): Update.
19578 (expand_expr_real_1): Update.
19579 * expr.h (try_casesi, try_tablejump): Update prototypes.
19580 * gimple-pretty-print.c (dump_probability): Update.
19581 (dump_profile): New.
19582 (dump_gimple_label): Update.
19583 (dump_gimple_bb_header): Update.
19584 * graph.c (draw_cfg_node_succ_edges): Update.
19585 * hsa-gen.c (convert_switch_statements): Update.
19586 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
19587 (find_if_case_1): Update.
19588 (find_if_case_2): Update.
19589 * internal-fn.c (expand_arith_overflow_result_store): Update.
19590 (expand_addsub_overflow): Update.
19591 (expand_neg_overflow): Update.
19592 (expand_mul_overflow): Update.
19593 (expand_vector_ubsan_overflow): Update.
19594 * ipa-cp.c (good_cloning_opportunity_p): Update.
19595 * ipa-split.c (split_function): Use make_single_succ_edge.
19596 * ipa-utils.c (ipa_merge_profiles): Update.
19597 * loop-doloop.c (add_test): Update.
19598 (doloop_modify): Update.
19599 * loop-unroll.c (compare_and_jump_seq): Update.
19600 (unroll_loop_runtime_iterations): Update.
19601 * lra-constraints.c (lra_inheritance): Update.
19602 * lto-streamer-in.c (input_cfg): Update.
19603 * lto-streamer-out.c (output_cfg): Update.
19604 * mcf.c (adjust_cfg_counts): Update.
19605 * modulo-sched.c (sms_schedule): Update.
19606 * omp-expand.c (expand_omp_for_init_counts): Update.
19607 (extract_omp_for_update_vars): Update.
19608 (expand_omp_ordered_sink): Update.
19609 (expand_omp_for_ordered_loops): Update.
19610 (expand_omp_for_generic): Update.
19611 (expand_omp_for_static_nochunk): Update.
19612 (expand_omp_for_static_chunk): Update.
19613 (expand_cilk_for): Update.
19614 (expand_omp_simd): Update.
19615 (expand_omp_taskloop_for_outer): Update.
19616 (expand_omp_taskloop_for_inner): Update.
19617 * omp-simd-clone.c (simd_clone_adjust): Update.
19618 * optabs.c (expand_doubleword_shift): Update.
19619 (expand_abs): Update.
19620 (emit_cmp_and_jump_insn_1): Update.
19621 (expand_compare_and_swap_loop): Update.
19622 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
19623 * predict.c (predictable_edge_p): Update.
19624 (edge_probability_reliable_p): Update.
19625 (set_even_probabilities): Update.
19626 (combine_predictions_for_insn): Update.
19627 (combine_predictions_for_bb): Update.
19628 (propagate_freq): Update.
19629 (estimate_bb_frequencies): Update.
19630 (force_edge_cold): Update.
19631 * profile-count.c (profile_count::dump): Add missing space into dump.
19632 (profile_count::debug): Add newline.
19633 (profile_count::differs_from_p): Explicitly convert to unsigned.
19634 (profile_count::stream_in): Update.
19635 (profile_probability::dump): New member function.
19636 (profile_probability::debug): New member function.
19637 (profile_probability::differs_from_p): New member function.
19638 (profile_probability::differs_lot_from_p): New member function.
19639 (profile_probability::stream_in): New member function.
19640 (profile_probability::stream_out): New member function.
19641 * profile-count.h (profile_count_quality): Rename to ...
19642 (profile_quality): ... this one.
19643 (profile_probability): New.
19644 (profile_count): Update.
19645 * profile.c (compute_branch_probabilities): Update.
19646 * recog.c (peep2_attempt): Update.
19647 * sched-ebb.c (schedule_ebbs): Update.
19648 * sched-rgn.c (find_single_block_region): Update.
19649 (compute_dom_prob_ps): Update.
19650 (schedule_region): Update.
19651 * sel-sched-ir.c (compute_succs_info): Update.
19652 * stmt.c (struct case_node): Update.
19653 (do_jump_if_equal): Update.
19654 (get_outgoing_edge_probs): Update.
19655 (conditional_probability): Update.
19656 (emit_case_dispatch_table): Update.
19657 (expand_case): Update.
19658 (expand_sjlj_dispatch_table): Update.
19659 (emit_case_nodes): Update.
19660 * targhooks.c: Update.
19661 * tracer.c (better_p): Update.
19662 (find_best_successor): Update.
19663 * trans-mem.c (expand_transaction): Update.
19664 * tree-call-cdce.c: Update.
19665 * tree-cfg.c (gimple_split_edge): Upate.
19666 (move_sese_region_to_fn): Upate.
19667 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
19668 * tree-eh.c (lower_resx): Upate.
19669 (cleanup_empty_eh_move_lp): Upate.
19670 * tree-if-conv.c (version_loop_for_if_conversion): Update.
19671 * tree-inline.c (copy_edges_for_bb): Update.
19672 (copy_cfg_body): Update.
19673 * tree-parloops.c (gen_parallel_loop): Update.
19674 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
19675 (gimple_gen_time_profiler): Update.
19676 * tree-ssa-dce.c (remove_dead_stmt): Update.
19677 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
19678 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
19679 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
19680 (unloop_loops): Update.
19681 (try_peel_loop): Update.
19682 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
19683 * tree-ssa-loop-split.c (connect_loops): Update.
19684 (split_loop): Update.
19685 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
19686 (hoist_guard): Update.
19687 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
19688 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
19689 (value_replacement): Update.
19690 * tree-ssa-reassoc.c (branch_fixup): Update.
19691 * tree-ssa-tail-merge.c (replace_block_by): Update.
19692 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
19693 (create_edge_and_update_destination_phis): Update.
19694 (compute_path_counts): Update.
19695 (recompute_probabilities): Update.
19696 (update_joiner_offpath_counts): Update.
19697 (freqs_to_counts_path): Update.
19698 (duplicate_thread_path): Update.
19699 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
19700 (struct switch_conv_info): Update.
19701 (gen_inbound_check): Update.
19702 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
19703 (vect_do_peeling): Update.
19704 (vect_loop_versioning): Update.
19705 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
19706 (optimize_mask_stores): Update.
19707 * ubsan.c (ubsan_expand_null_ifn): Update.
19708 * value-prof.c (gimple_divmod_fixed_value): Update.
19709 (gimple_divmod_fixed_value_transform): Update.
19710 (gimple_mod_pow2): Update.
19711 (gimple_mod_pow2_value_transform): Update.
19712 (gimple_mod_subtract): Update.
19713 (gimple_mod_subtract_transform): Update.
19714 (gimple_ic): Update.
19715 (gimple_stringop_fixed_value): Update.
19716 (gimple_stringops_transform): Update.
19717 * value-prof.h: Update.
19719 2017-06-29 Carl Love <cel@us.ibm.com>
19721 * config/rs6000/rs6000-c.c: Add support for built-in functions
19722 vector signed int vec_signed (vector float);
19723 vector signed long long vec_signed (vector double);
19724 vector signed int vec_signed2 (vector double, vector double);
19725 vector signed int vec_signede (vector double);
19726 vector signed int vec_signedo (vector double);
19727 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
19728 instruction generator.
19729 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19730 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
19731 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
19733 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
19734 vunsignede_v2df): Add define_expands.
19735 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
19736 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
19737 VEC_UNSIGNEDO): Add definitions.
19738 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
19739 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
19740 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
19741 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
19742 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
19743 * config/rs6000/altivec.h (vec_signed, vec_signed2,
19744 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
19745 vec_unsignede, vec_unsignedo): Add builtin defines.
19746 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
19748 * doc/extend.texi: Update the built-in documentation file for the
19749 new built-in functions.
19751 2017-06-29 Richard Biener <rguenther@suse.de>
19753 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
19754 reduction chains to LOOP_VINFO_REDUCTIONS.
19755 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
19756 SLP reductions after processing reduction chains.
19758 2017-06-29 Nathan Sidwell <nathan@acm.org>
19760 * builtins.c (fold_builtin_FUNCTION): Use
19761 lang_hooks.decl_printable_name.
19763 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
19765 PR middle-end/81194
19766 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
19767 with only one label.
19768 * stmt.c (expand_case): Assert NCASES is greater than one.
19770 2017-06-29 Richard Biener <rguenther@suse.de>
19772 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
19774 (group_case_labels): Likewise.
19775 (find_taken_edge): Push sanity checking on val to workers...
19776 (find_taken_edge_cond_expr): ... here
19777 (find_taken_edge_switch_expr): ... and here, handle cases
19778 with just a default label.
19779 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
19780 (group_case_labels): Likewise.
19781 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
19782 group_case_labels does anything cleanup the CFG again.
19784 2017-06-29 Bin Cheng <bin.cheng@arm.com>
19786 PR tree-optimization/81196
19787 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
19788 exit condition comparing two IVs.
19790 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
19792 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
19793 profile to the dummy entry at the end of the list of architectures.
19794 * config/arm/arm-cpu-cdata.h: Regenerated.
19796 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19797 Michael Collison <michael.collison@arm.com>
19800 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
19802 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
19803 (*aarch64_reg_<mode>3_minus_mask): New pattern.
19804 (*aarch64_<optab>_reg_di3_mask2): New pattern.
19805 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
19806 of shift when the shift amount is masked with constant equal to
19807 the size of the mode.
19808 * config/aarch64/predicates.md (subreg_lowpart_operator): New
19811 2017-06-29 Martin Liska <mliska@suse.cz>
19813 * config/i386/i386.opt: Change range from [1,5] to [0,5].
19815 2017-06-29 Yury Gribov <tetra2005@gmail.com>
19818 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
19821 (edge_growth_cache_entry::edge_growth_cache_entry): New
19823 (reset_edge_growth_cache): Update to use constructor.
19825 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19827 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
19828 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19829 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
19831 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
19833 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
19834 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
19836 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
19838 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
19839 (*-linux-uclibc*): Add t-uclibc tmake_file.
19840 * config/t-musl: New.
19841 * config/t-uclibc: New.
19843 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
19845 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
19847 (gen_comm_data): Emit architectural setting of arch_prof.
19848 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
19850 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
19851 (armv8-m.base, armv8-m.main): Likewise.
19852 * arm-protos.h (arm_build_target): Add profile field.
19853 (arch_option): Likewise.
19854 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
19856 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
19857 arm_active_target.profile.
19859 2017-06-28 Richard Biener <rguenther@suse.de>
19861 PR middle-end/81227
19862 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
19863 TYPE_OVERFLOW_WRAPS.
19864 * match.pd (negate_expr_p): Likewise.
19865 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
19866 fold_build2, not fold_binary.
19868 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19870 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
19871 Convert memory address to Pmode.
19872 (aarch64_print_operand): Assert MEM operands are always Pmode.
19874 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
19877 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
19878 Remove redundant if.
19879 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
19880 * config/arm/aarch-common-protos.h
19881 (aarch_forward_to_shift_is_not_shifted_re): Remove.
19882 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
19884 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
19887 * multiple_target.c (create_dispatcher_calls): Set the default
19888 clone to be static, not public.
19890 2017-06-28 Richard Biener <rguenther@suse.de>
19892 * tree-vect-loop.c (vectorizable_reduction): Move special
19893 cond reduction IV var creation ...
19894 (vect_create_epilog_for_reduction): ... here. Remove induction_index
19895 parameter. Use STMT_VINFO_VECTYPE.
19896 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
19899 2017-06-28 Martin Liska <mliska@suse.cz>
19902 * ipa-visibility.c (non_local_p): Handle visibility.
19904 2017-06-28 Martin Liska <mliska@suse.cz>
19907 * common.opt: Add IntegerRange to various options.
19908 * opt-functions.awk (integer_range_info): New function.
19909 * optc-gen.awk: Add integer_range_info to cl_options struct.
19910 * opts-common.c (decode_cmdline_option): Handle
19911 CL_ERR_INT_RANGE_ARG.
19912 (cmdline_handle_error): Likewise.
19913 * opts.c (print_filtered_help): Show valid interval in
19914 when --help is provided.
19915 * opts.h (struct cl_option): Add range_min and range_max fields.
19916 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
19918 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
19920 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
19921 (x * C EQ/NE y * C): New transformation.
19923 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
19925 * genmultilib (combination_space): Accept '+' in option names.
19927 2017-06-28 Martin Liska <mliska@suse.cz>
19930 * asan.c (instrument_derefs): Bail out inner references
19931 that are hard register variables.
19933 2017-06-28 Jakub Jelinek <jakub@redhat.com>
19936 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
19937 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
19939 2017-06-28 Richard Biener <rguenther@suse.de>
19941 * tree-vectorizer.h (vect_get_vec_defs): Remove.
19942 (vect_get_slp_defs): Adjust.
19943 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
19945 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
19947 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
19948 get_initial_defs_for_reduction instead of vect_get_vec_defs.
19949 (vectorizable_reduction): Adjust.
19950 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
19952 (vect_get_slp_defs): Likewise.
19953 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
19954 (vectorizable_bswap): Adjust.
19955 (vectorizable_call): Likewise.
19956 (vectorizable_conversion): Likewise.
19957 (vectorizable_assignment): Likewise.
19958 (vectorizable_shift): Likewise.
19959 (vectorizable_operation): Likewise.
19960 (vectorizable_store): Likewise.
19961 (vectorizable_condition): Likewise.
19962 (vectorizable_comparison): Likewise.
19964 2017-06-28 Michael Collison <michael.collison@arm.com>
19967 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
19969 (arm_gen_movmemqi): Removed unused variable 'i'.
19970 Convert 'for' loop into 'while' loop.
19971 (arm_expand_prologue): Remove last unnecessary set of insn.
19972 (thumb_pop): Remove unused variable 'pushed_words'.
19973 (thumb_exit): Remove last unnecessary set of regs_to_pop.
19975 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19977 * config/s390/predicates.md: Use s390_rel_address_ok_p.
19978 * config/s390/s390-protos.h: Add prototype of
19979 s390_rel_address_ok_p.
19980 * config/s390/s390.c (s390_got_symbol): New function.
19981 (s390_rel_address_ok_p): New function.
19982 (legitimize_pic_address): Use s390_rel_address_ok_p.
19983 (s390_load_got): Use s390_got_symbol.
19984 (s390_option_override): Issue error if
19985 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
19986 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
19988 * config/s390/s390.opt: New option mpic-data-is-text-relative.
19990 2017-06-27 Andrew Pinski <apinski@cavium.com>
19992 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
19993 (X * copysign (1.0, X)): New pattern.
19994 (X * copysign (1.0, -X)): New pattern.
19995 (copysign (-1.0, CST)): New pattern.
19997 2017-06-27 Joseph Myers <joseph@codesourcery.com>
19999 * genmultilib (combination_space): Remove variable.
20000 Validate reuse rules against regular expression for any sequence
20001 of multilib options in any order.
20003 2017-06-27 Michael Collison <michael.collison@arm.com>
20005 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
20006 call aarch64_split_simd_combine.
20007 * (aarch64_combine_internal<mode>): Delete pattern.
20008 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
20009 Allow register and subreg operands.
20011 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20013 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
20014 specific need, just fallback on defaults.
20015 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
20017 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20018 Olivier Hainque <hainque@adacore.com>
20020 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
20022 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
20023 targets. Pick a default if no particular attempt applied.
20024 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
20027 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20029 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
20030 (x86_64-wrs-vxworks7): Likewise.
20032 2017-06-27 Marek Polacek <polacek@redhat.com>
20035 * ubsan.c (instrument_null): Check get_base_address's result for null.
20037 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20039 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
20041 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
20043 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
20044 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
20045 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
20046 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
20047 New function types.
20048 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
20049 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
20050 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
20051 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
20052 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
20053 BUILT_IN_FEUPDATEENV): New builtins.
20054 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
20055 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
20056 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
20057 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
20059 (builtin_structptr_types): Adjust size.
20060 * tree.c (builtin_structptr_types): Add four entries.
20062 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20063 Olivier Hainque <hainque@adacore.com>
20065 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
20066 (TLS_SYM): New local macro, forcing reference to __tls__ on
20067 link command lines for VxWorks 7 RTPs, triggering initialization
20069 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
20070 OS features TLS support, true for RTPs on VxWorks 7.
20071 * config/vxworks.c (vxworks_override_options): Setup emutls
20074 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20076 * predict.c (test_prediction_value_range): Use -1U instead of -1
20077 to avoid narrowing conversion warning.
20078 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
20079 to avoid narrowing conversion warning.
20080 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
20082 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
20084 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
20086 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
20087 64bit configurations.
20088 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
20089 (SIZE_TYPE): Likewise.
20090 * config/vxworks.c (vxworks_emutls_var_fields): Use
20091 long_unsigned_type_node instead of unsigned_type_node as the offset
20092 field type, which is "pointer" mode in emutls.c.
20094 2017-06-27 Jakub Jelinek <jakub@redhat.com>
20097 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
20099 PR middle-end/81207
20100 * gimple-fold.c (replace_call_with_call_and_fold): Handle
20101 gimple_vuse copying separately from gimple_vdef copying.
20103 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20105 * value-prof.c (free_hist): Remove call to memset and the enclosing if
20108 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20109 Olivier Hainque <hainque@adacore.com>
20111 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
20112 for all vxworks7 targets.
20113 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
20114 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
20115 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
20116 variations for VX6/VX7 and 32/64bits later on in ...
20117 (VXWORKS_LIB_SPEC): Leverage new macros.
20118 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
20119 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
20121 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
20123 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
20124 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
20126 2017-06-26 Carl Love <cel@us.ibm.com>
20128 * config/rs6000/rs6000-c.c: Add support for built-in functions
20129 vector bool char vec_reve (vector bool char);
20130 vector signed char vec_reve (vector signed char);
20131 vector unsigned char vec_reve (vector unsigned char);
20132 vector bool int vec_reve (vector bool int);
20133 vector signed int vec_reve (vector signed int);
20134 vector unsigned int vec_reve (vector unsigned int);
20135 vector bool long long vec_reve (vector bool long long);
20136 vector signed long long vec_reve (vector signed long long);
20137 vector unsigned long long vec_reve (vector unsigned long long);
20138 vector bool short vec_reve (vector bool short);
20139 vector signed short vec_reve (vector signed short);
20140 vector double vec_reve (vector double);
20141 vector float vec_reve (vector float);
20142 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
20143 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
20144 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
20145 (altivec_vreve): New pattern.
20146 * config/rs6000/altivec.h (vec_reve): New define.
20147 * doc/extend.texi (vec_rev): Update the built-in documentation file
20148 for the new built-in functions.
20150 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20152 PR tree-optimization/71815
20153 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
20155 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
20157 (slsr_process_phi): Likewise.
20158 (replace_uncond_cands_and_profitable_phis): Don't replace a
20159 multiply candidate with a stride of 1 (copy or cast).
20160 (phi_incr_cost): Call uses_consumed_by_stmt rather than
20162 (lowest_cost_path): Likewise.
20163 (total_savings): Likewise.
20165 2017-06-26 Richard Biener <rguenther@suse.de>
20168 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
20169 Use def_builtin_pure for all gather builtins.
20171 2017-06-26 Richard Biener <rguenther@suse.de>
20173 PR tree-optimization/81203
20174 * tree-tailcall.c (find_tail_calls): Do not move stmts into
20175 non-dominating BBs.
20177 2017-06-26 Marek Polacek <polacek@redhat.com>
20180 * doc/invoke.texi: Document -Wmultistatement-macros.
20182 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
20184 * doc/sourcebuild.texi (ARM-specific attributes): Document new
20185 arm_neon_ok_no_float_abi effective target.
20187 2017-06-26 Richard Biener <rguenther@suse.de>
20189 PR tree-optimization/80928
20190 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
20191 (copy_bbs): Set BB_DUPLICATED flag early.
20192 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
20194 (execute_on_shrinking_pred): Likewise.
20195 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
20196 BB_DUPLICATED blocks.
20197 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
20198 iterate over all PHIs considering removal of *gsi.
20200 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
20202 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
20205 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
20207 * config/rs6000/rs6000-string.c: (expand_block_clear,
20208 do_load_for_compare, select_block_compare_mode,
20209 compute_current_alignment, expand_block_compare,
20210 expand_strncmp_align_check, expand_strn_compare,
20211 expand_block_move, rs6000_output_load_multiple)
20212 Move functions related to string/block move/compare
20213 to a separate file.
20214 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
20215 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
20216 for this function which is now used in two files.
20217 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
20218 * config.gcc: Add rs6000-string.o to extra_objs for
20219 targets powerpc*-*-* and rs6000*-*-*.
20221 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
20224 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
20225 32-bit, since indexed is not valid for DImode.
20226 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
20227 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
20228 (define_peephole2 for Altivec d-form load): Add 32-bit support.
20229 (define_peephole2 for Altivec d-form store): Likewise.
20232 * multiple_target.c (create_dispatcher_calls): Only create the
20233 dispatcher call if the function is the default clone of a
20234 versioned function.
20236 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
20238 PR middle-end/80902
20239 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
20240 a call, force the call to not be a tail call.
20242 2017-06-23 Jeff Law <law@redhat.com>
20244 * doc/contrib.texi: Add entry for Steven Pemberton's work on
20247 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
20249 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
20250 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
20251 handling for early expansion of vector shifts (sl,sr,sra,rl).
20252 (builtin_function_type): Add vector shift right instructions
20253 to the unsigned argument list.
20255 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
20257 rtl-optimizatoin/79286
20258 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
20259 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
20260 trap. PIC register plus a const unspec without offset can never trap.
20262 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
20264 * tree.h (builtin_structptr_type): New type.
20265 (builtin_structptr_types): Declare new array.
20266 * tree.c (builtin_structptr_types): New array.
20267 (free_lang_data, build_common_tree_nodes): Use it.
20269 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
20272 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
20275 2017-06-22 Matt Turner <mattst88@gmail.com>
20277 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
20278 Lake models to skylake case. Assume skylake for unknown
20279 models with clflushopt.
20281 2017-06-22 Jeff Law <law@redhat.com>
20283 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
20284 frame sizes that do not satisfy aarch64_uimm12_shift.
20286 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
20288 * profile-count.h (apply_probability,
20289 apply_scale, probability_in): Fix checks for zero.
20291 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20293 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
20294 * doc/cppdiropts.texi (-I @var{dir}): Document it.
20296 2016-06-22 Richard Biener <rguenther@suse.de>
20298 * tree-vect-loop.c (vect_model_reduction_cost): Handle
20299 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
20300 REDUC_MAX_EXPR support.
20301 (vectorizable_reduction): Likewise.
20302 (vect_create_epilog_for_reduction): Likewise.
20304 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
20306 * match.pd (A / (1 << B) -> A >> B): New.
20307 * generic-match-head.c: Include optabs-tree.h.
20308 * gimple-match-head.c: Likewise.
20309 * optabs-tree.h (target_supports_op_p): New.
20310 * optabs-tree.c (target_supports_op_p): New.
20312 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20314 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
20315 $gcc_cv_ld --help output.
20316 (gcc_cv_ld_demangle): Likewise.
20317 (gcc_cv_ld_eh_frame_hdr): Likewise.
20318 (gcc_cv_ld_pie): Likewise.
20319 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
20320 (gcc_cv_ld_buildid): Likewise.
20321 (gcc_cv_ld_sysroot): Likewise.
20322 (ld_bndplt_support): Likewise.
20323 (ld_pushpopstate_support): Likewise.
20324 * configure: Regenerate.
20325 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
20327 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20330 * config/i386/sse.md (round<mode>2): Renumber match_dup and
20331 operands indexes to avoid gap between operands and match_dups.
20333 2017-06-21 Andrew Pinski <apinski@cavium.com>
20335 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
20336 Increment Arith_shift and Arith_shift_reg by 1.
20337 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
20339 * config/aarch64/aarch64.c (thunderx_tunings): Enable
20340 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
20341 (aarch64_strip_extend): Add new argument and test for it.
20342 (aarch64_cheap_mult_shift_p): New function.
20343 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
20344 add a cost if it is true.
20345 Update calls to aarch64_strip_extend.
20346 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
20348 2017-06-21 Andrew Pinski <apinski@cavium.com>
20350 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
20352 (thunderxt88): Likewise.
20353 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
20354 (thunderx_prefetch_tune): New variable.
20355 (thunderx2t99_prefetch_tune): Update for the correct values.
20356 (thunderxt88_tunings): New variable.
20357 (thunderx_tunings): Use thunderx_prefetch_tune instead of
20358 generic_prefetch_tune.
20359 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
20361 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20363 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
20364 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
20365 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
20366 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
20367 (aarch64_atomic_cas<mode>, GPI): Likewise.
20369 2017-06-21 Martin Liska <mliska@suse.cz>
20371 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
20372 statements on cold and hot labels.
20373 * predict.c (tree_estimate_probability_bb): Remove the
20374 prediction from this place.
20376 2017-06-21 Martin Liska <mliska@suse.cz>
20378 PR tree-optimization/79489
20379 * gimplify.c (maybe_add_early_return_predict_stmt): New
20381 (gimplify_return_expr): Call the function.
20382 * predict.c (tree_estimate_probability_bb): Remove handling
20384 * predict.def: Update comment about early return predictor.
20385 * gimple-predict.h (is_gimple_predict): New function.
20386 * predict.def: Change default value of early return to 66.
20387 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
20389 * passes.def: Put pass_strip_predict_hints to the beginning of
20392 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
20394 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
20395 FUNCTION_DECL declarations.
20396 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
20398 (dwaf2out_decl): Likewise.
20399 * godump.c (go_early_global_decl): Skip call to the real debug hook
20400 for FUNCTION_DECL declarations.
20401 * passes.c (rest_of_decl_compilation): Skip call to the
20402 early_global_decl debug hook for FUNCTION_DECL declarations, unless
20403 -fdump-go-spec is passed.
20405 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20407 * config/i386/i386.c (struct builtin_isa): New field pure_p.
20408 Reorder for compactness.
20409 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
20410 (def_builtin_pure, def_builtin_pure2): New functions.
20411 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
20413 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
20415 * match.pd (nop_convert): New predicate.
20416 ((A +- CST1) +- CST2): Allow some NOP conversions.
20418 2017-06-21 Jakub Jelinek <jakub@redhat.com>
20421 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
20422 with ctors/dtors if GOVD_SHARED is set.
20424 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20426 * config/aarch64/aarch64.md (movti_aarch64):
20427 Emit mov rather than orr.
20428 (movtf_aarch64): Likewise.
20429 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
20430 Emit mov rather than orr.
20432 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20434 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
20435 Swap alternatives, make integer dup more expensive.
20437 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
20439 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20440 Return true for non-tls symbols.
20442 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
20444 * config/aarch64/aarch64-cores.def (cortex-a55): New.
20445 (cortex-a75): Likewise.
20446 (cortex-a75.cortex-a55): Likewise.
20447 * config/aarch64/aarch64-tune.md: Regenerate.
20448 * doc/invoke.texi (-mtune): Document new values for -mtune.
20450 2017-06-21 Tom de Vries <tom@codesourcery.com>
20452 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
20453 stack_size feature.
20454 (Effective-Target Keywords, Other attributes): Suggest using
20455 dg-add-options stack_size feature to get stack limit in stack_size
20456 effective target documentation.
20458 2017-06-21 Julian Brown <julian@codesourcery.com>
20459 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20461 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
20462 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
20463 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
20465 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
20466 attribute type list for neon_multiply.
20467 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
20468 attribute type list for neon_multiply.
20469 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
20470 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
20471 attribute type list for neon_multiply.
20472 * config/arm/types.md (crypto_pmull): Add.
20473 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
20474 attribute type list.
20476 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
20478 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
20481 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20483 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
20484 to make sure not to dereference a NULL cost_classes_ptr pointer.
20486 2017-06-20 Carl Love <cel@us.ibm.com>
20488 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20489 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
20490 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
20491 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
20492 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
20493 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
20494 VMULOSW): New enum "unspec" values.
20495 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
20496 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
20497 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
20498 altivec_vmulosw): New patterns.
20499 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
20500 VMULOSW): Add definitions.
20502 2017-06-20 Julia Koval <julia.koval@intel.com>
20504 * config/i386/i386.c: Fix rounding expand for new pattern.
20505 * config/i386/subst.md: Fix pattern (parallel -> unspec).
20507 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20509 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
20510 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
20512 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20514 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
20517 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
20519 * config/aarch64/aarch64-cores.def: Rearrange to sort by
20520 architecture, then by implementer ID.
20521 * config/aarch64/aarch64-tune.md: Regenerate.
20523 2017-06-20 Richard Biener <rguenther@suse.de>
20525 PR middle-end/81097
20526 * fold-const.c (split_tree): Fold to type before negating.
20528 2017-06-20 David Malcolm <dmalcolm@redhat.com>
20530 * diagnostic-show-locus.c
20531 (selftest::test_fixit_deletion_affecting_newline): New function.
20532 (selftest::diagnostic_show_locus_c_tests): Call it.
20534 2017-06-20 Andreas Schwab <schwab@suse.de>
20537 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
20540 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20542 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
20543 __ARM_FEATURE_COPROC according to support.
20545 2017-06-20 Jakub Jelinek <jakub@redhat.com>
20547 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20548 Rewritten to avoid overflow for > 32-bit pointers.
20551 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
20552 by removing enum keyword.
20553 (ubsan_type_descriptor): Likewise. Formatting fix.
20556 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
20557 splitter): Require TARGET_SSE2 in the condition.
20559 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
20562 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
20563 for doing vector set of SFmode on ISA 3.0.
20564 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
20565 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
20567 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
20568 SFmode value into a V4SF variable that was extracted from another
20569 V4SF variable without converting the element to double precision
20570 and back to single precision vector format.
20571 (vsx_insert_extract_v4sf_p9_2): Likewise.
20573 2017-06-19 Jakub Jelinek <jakub@redhat.com>
20575 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
20576 in UWHI to avoid undefined overflow.
20579 * ubsan.h (enum ubsan_encode_value_phase): New.
20580 (ubsan_encode_value): Change second argument to
20581 enum ubsan_encode_value_phase with default value of
20582 UBSAN_ENCODE_VALUE_GENERIC.
20583 * ubsan.c (ubsan_encode_value): Change second argument to
20584 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
20585 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
20586 create_tmp_var_raw instead of create_tmp_var and use a
20588 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
20589 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
20590 ubsan_encode_value callers.
20593 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
20594 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
20595 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
20597 2017-06-19 Richard Biener <rguenther@suse.de>
20599 PR middle-end/81118
20600 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
20601 estimates if we changed anything.
20603 2017-06-19 Richard Biener <rguenther@suse.de>
20605 PR tree-optimization/80887
20606 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
20607 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
20608 simplified lookups, then reset mprts_hook.
20609 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
20611 (try_to_simplify): Likewise.
20613 2017-06-19 Martin Liska <mliska@suse.cz>
20616 * gimplify.c (gimplify_switch_expr):
20617 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
20619 2017-06-19 Martin Liska <mliska@suse.cz>
20621 * doc/install.texi: Document that PGO runs in 4 stages.
20623 2017-06-19 Martin Liska <mliska@suse.cz>
20626 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
20627 to dispatcher function name.
20628 * multiple_target.c (replace_function_decl): New function.
20629 (create_dispatcher_calls): Redirect both edges and references.
20631 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
20633 * profile-count.c (profile_count::dump): Dump quality.
20634 (profile_count::differs_from_p): Update for unsigned val.
20635 * profile-count.h (profile_count_quality): New enum.
20636 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
20638 2017-06-19 Richard Biener <rguenther@suse.de>
20640 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
20641 struct function as arg.
20642 (estimate_numbers_of_iterations): Export overload with loop arg.
20643 (free_numbers_of_iterations_estimates_loop): Use an overload of
20644 free_numbers_of_iterations_estimates instead.
20645 * tree-cfg.c (remove_bb): Adjust.
20646 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
20647 * tree-parloops.c (gen_parallel_loop): Likewise.
20648 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
20650 (tree_unroll_loops_completely): Likewise.
20651 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
20652 Use an overload instead and export.
20653 (estimated_loop_iterations): Adjust.
20654 (max_loop_iterations): Likewise.
20655 (likely_max_loop_iterations): Likewise.
20656 (estimate_numbers_of_iterations): Take struct function as arg
20658 (loop_exits_before_overflow): Adjust.
20659 (free_numbers_of_iterations_estimates_loop): Use an overload.
20660 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
20661 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
20663 2017-06-19 Richard Biener <rguenther@suse.de>
20666 * ipa-prop.c (find_constructor_constant_at_offset): Handle
20667 RANGE_EXPR conservatively.
20669 2017-06-16 Carl Love <cel@us.ibm.com>
20671 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
20672 definitions for vec_float, vec_float2, vec_floato,
20673 vec_floate built-ins.
20674 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
20675 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
20677 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
20678 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
20679 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
20680 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
20681 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
20682 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
20683 vec_floato): Add builtin defines.
20684 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
20685 Update the built-in documentation file for the new built-in
20688 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20690 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
20691 (mthumb): Mark as the negative of -marm.
20693 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20695 * doc/invoke.texi (ARM Options, -mcpu): Document supported
20697 (ARM Options, -mtune): Document that this accepts the same
20698 extension options as -mcpu.
20699 (ARM Options, -mfpu): Document addition of -mfpu=auto.
20701 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20703 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
20704 permitted extensions.
20706 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20708 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
20709 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
20710 (armv8-m.main): Add option +nodsp.
20711 * config/arm/arm-cpu-cdata.h: Regenerated.
20713 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20715 * config/arm/t-fuchsia: New file.
20716 * config.gcc (arm*-*-fuchsia*): Use it.
20718 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20720 * config/arm/t-symbian: Rewrite for new option infrastructure.
20722 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20724 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
20725 (MULTILIB_REQUIRED): Likewise.
20727 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20729 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
20730 (MULTILIB_RESUE): Likewise.
20731 (MULTILIB_MATCHES): Likewise.
20732 (MULTLIB_REQUIRED): Likewise.
20734 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20736 * config/arm/t-rtems: Rewrite for new option framework.
20738 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20740 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
20741 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
20742 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
20743 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
20744 * config/arm/t-multilib: ... here.
20745 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
20746 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
20747 armv7-a and armv8*-a when A-profile libraries have not been built.
20748 * config/arm/t-rmprofile: Rewrite.
20750 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20752 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
20753 with a backslash. Remove the backslash after substituting unescaped
20755 * doc/fragments.texi (MULTILIB_REUSE): Document it.
20757 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20759 * config.gcc: (arm*-*-*): When building a-profile libraries, force
20760 the driver to pass through the default setting of -mfloat-abi.
20761 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
20763 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
20764 (all_feat_combs): New rule.
20765 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
20767 * config/arm/t-aprofile: Rewrite.
20769 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20771 * config/arm/arm.h (FPUTYPE_AUTO): Define.
20772 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
20773 fpu is not specified by the user/command-line.
20774 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
20775 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
20776 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
20777 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
20778 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
20779 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
20781 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20783 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
20784 * config/arm/t-arm-elf: Rewritten.
20786 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20788 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
20789 have some floating-point instructions.
20790 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
20791 (TARGET_MAYBE_HARD_FLOAT): New macro.
20792 * config/arm/arm-builtins.c (arm_init_builtins): Use
20793 TARGET_MAYBE_HARD_FLOAT.
20794 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
20796 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20798 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
20799 (configargs.h): Include it.
20800 (arm_print_hint_for_fpu_option): New function.
20801 (arm_parse_fpu_option): New function.
20802 (candidate_extension): New class.
20803 (arm_canon_for_multilib): New function.
20804 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
20805 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
20806 (ARCH_CANONICAL_SPECS): New macro.
20807 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
20809 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20811 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
20812 are set after handling multilib fragments. Set target_cpu_default2
20815 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20817 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
20819 (arm*-*-*): Set target_cpu_default2 to a quoted string.
20820 * config/arm/parsecpu.awk (check_cpu): Validate any extension
20822 (check_arch): Likewise.
20823 * config/arm/arm.c (arm_configure_build_target): Handle
20824 TARGET_CPU_DEFAULT being a string constant. Scan any feature
20825 options in the default.
20827 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20829 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
20830 when an option is an alias of another.
20831 * config/arm/parsecpu.awk (optalias): New parser token.
20832 (gen_comm_data): Mark non-alias options as such. Emit entries
20833 for extension aliases.
20834 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
20835 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
20836 (armv6kz, armv6zk, armv6t2): Likewise.
20837 (armv7): Make vfpv3-d16 an alias.
20838 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
20840 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
20841 Sort in canonical order.
20842 (armv8-a): Sort in canonical order.
20843 (armv8.1-a, armv8.2-a): Likewise.
20844 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
20846 (cortex-a9): Sort in canonical order.
20847 * config/arm/arm.c (selftests.h): Include it.
20848 (arm_test_cpu_arch_data): New function.
20849 (arm_run_self_tests): New function.
20850 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
20851 (targetm): Move declaration to the end of the file.
20852 * arm-cpu-cdata.h: Regenerated.
20854 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20856 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
20857 call to target_mode_check describing the type of option passed.
20858 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
20859 (arm_target_thumb_only): Use arm_parse_arch_option_name or
20860 arm_parse_cpu_option_name to match parameters against list of
20862 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
20863 arm_arch_core_flags data structure.
20864 * config/arm/arm-cpu_cdata.h: Regenerated.
20866 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20868 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
20870 (arm_print_hint_for_cpu_option): Likewise.
20871 (arm_print_hint_for_arch_option): Likewise.
20872 (arm_parse_cpu_option_name): Likewise.
20873 (arm_parse_arch_option_name): Likewise.
20874 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
20875 of entries in the all_fpus list.
20876 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
20877 (arm_parse_cpu_option_name): Declare.
20878 (arm_parse_arch_option_name): Declare.
20879 (arm_parse_option_features): Declare.
20880 (arm_intialize_isa): Declare.
20881 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
20883 (gen_comm_data): ... here. Make definitions non-static.
20884 * config/arm/arm-cpu-data.h: Regenerated.
20885 * config/arm/arm-cpu-cdata.h: Regenerated.
20887 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20889 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
20890 (cpu_arch_extension): New structure.
20891 (cpu_arch_option, arch_option, cpu_option): New structures.
20892 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
20893 architecture types.
20894 (gen_data): Generate new format data tables.
20895 * config/arm/arm.c (cpu_tune): New structure.
20896 (cpu_option, processors): Delete.
20897 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
20898 (arm_print_hint_for_cpu_option): ... this and ...
20899 (arm_print_hint_for_arch_option): ... this.
20900 (arm_parse_arch_cpu_name): Delete. Replace with ...
20901 (arm_parse_cpu_option_name): ... this and ...
20902 (arm_parse_arch_option_name): ... this.
20903 (arm_unrecognized_feature): Change type of target parameter to
20905 (arm_parse_arch_cpu_features): Delete. Replace with ...
20906 (arm_parse_option_features): ... this.
20907 (arm_configure_build_target): Rework to use new configuration data
20909 (arm_print_tune_info): Rework for new configuration data tables.
20910 * config/arm/arm-cpu-data.h: Regenerated.
20911 * config/arm/arm-cpu.h: Regenerated.
20913 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20915 * Makefile.in (OBJS): Move sbitmap.o from here ...
20916 (OBJS-libcommon): ... to here.
20918 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20920 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
20921 (ISA_ALL_CRYPTO): New macro.
20922 (ISA_ALL_SIMD): New macro
20923 (ISA_ALL_FP): New macro.
20924 * config/arm/arm.c (fpu_bitlist): Update initializer.
20925 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
20927 (arm9e): Add fpu. Add option for nofp
20928 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
20929 (arm926ej-s, arm1026ej-s): Likewise.
20930 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
20931 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
20932 neon-fp16, neon-vfpv4, nofp and nosimd.
20933 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
20934 (cortex-a8): Add fpu. Add option for nofp.
20935 (cortex-a9): Add fpu. Add options for nosimd and nofp.
20936 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
20937 (cortex-r4f): Add fpu.
20938 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
20939 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
20941 (cortex-r8): Likewise.
20942 (cortex-m4): Add fpu. Add option for nofp.
20943 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
20944 (cortex-a17.cortex-a7): Likewise.
20945 (cortex-a32): Add fpu. Add options for crypto and nofp.
20946 (cortex-a35, cortex-a53): Likewise.
20947 (cortex-a57): Add fpu. Add option for crypto.
20948 (cortex-a72, cortex-a73): Likewise.
20949 (exynos-m1): Likewise.
20950 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
20951 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
20952 (cortex-m33): Add fpu. Add option for nofp.
20953 * config/arm/arm-cpu-cdata.h: Regenerated
20954 * config/arm/arm-cpu-data.h: Regenerated.
20956 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20958 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
20959 (armv5te, armv5tej): Likewise.
20960 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
20961 (armv7): Add options fp and vfpv3-d16.
20962 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
20963 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
20965 (armv7ve): Likewise.
20966 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
20967 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
20968 (armv8-a): Add nocrypto option.
20969 (armv8.1-a, armv8.2-a): Likewise.
20970 (armv8-m.main): add options fp, fp.dp and nofp.
20972 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
20974 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
20976 (armv8-a+crc): Delete.
20977 (armv8.1-a): Add options simd, crypto and nofp.
20978 (armv8.2-a): Add options fp16, simd, crypto and nofp.
20979 (armv8.2-a+fp16): Delete.
20980 (armv8-m.main): Add option dsp.
20981 (armv8-m.main+dsp): Delete.
20982 (cortex-a8): Add fpu. Add nofp option.
20983 (cortex-a9): Add fpu. Add nofp and nosimd options.
20984 * config/arm/parsecpu.awk (gen_data): Generate option tables and
20985 link to main cpu and architecture data structures.
20986 (gen_comm_data): Only put isa attributes from the main architecture
20988 (option): New statement for architecture and CPU entries.
20989 * arm.c (struct cpu_option): New structure.
20990 (struct processors): Add entry for options.
20991 (arm_unrecognized_feature): New function.
20992 (arm_parse_arch_cpu_name): Ignore any characters after the first
20994 (arm_parse_arch_cpu_feature): New function.
20995 (arm_configure_build_target): Separate out any CPU and architecture
20996 features and parse separately. Don't error out if -mfpu=auto is
20997 used with only an architecture string.
20998 (arm_print_asm_arch_directives): New function.
20999 (arm_file_start): Call it.
21000 * config/arm/arm-cpu-cdata.h: Regenerated.
21001 * config/arm/arm-cpu-data.h: Likewise.
21002 * config/arm/arm-tables.opt: Likewise.
21004 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21006 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
21007 assembler when it is not -mfpu=auto.
21009 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21011 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
21012 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
21013 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
21014 (ASM_CPU_SPEC): Rewrite.
21015 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
21016 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
21017 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
21018 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
21019 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
21020 copied string is NUL-terminated. Also strip any characters prefixed
21022 (arm_rewrite_selected_arch): New function.
21023 (arm_rewrite_march): New function.
21025 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
21027 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
21028 (x_arm_cpu_string, x_arm_tune_string): Likewise.
21029 (march, mcpu, mtune): Convert to string-based options.
21030 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
21031 (arm_parse_arch_cpu_name): New function.
21032 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
21033 identify selected architecture or CPU.
21034 (arm_option_save): New function.
21035 (TARGET_OPTION_SAVE): Redefine.
21036 (arm_option_restore): Restore string options.
21037 (arm_option_print): Print string options.
21039 2017-06-16 Martin Sebor <msebor@redhat.com>
21041 PR tree-optimization/80933
21042 PR tree-optimization/80934
21043 * builtins.c (fold_builtin_3): Do not handle bcmp here.
21044 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
21045 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
21046 (gimple_fold_builtin): Call them.
21048 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21050 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
21051 as unlikely; update profile.
21053 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21055 * predict.c (force_edge_cold): Handle declaring edges impossible
21058 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21060 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
21062 (try_unroll_loop_completely): Fix reporting.
21064 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21066 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
21068 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
21071 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
21072 if given a non-constant argument for an intrinsic which requires a
21075 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21077 * profile.c (compare_freqs): New function.
21078 (branch_prob): Sort edge list.
21079 (find_spanning_tree): Assume that the list is priority sorted.
21081 2017-06-16 Richard Biener <rguenther@suse.de>
21083 PR tree-optimization/81090
21084 * passes.def (pass_record_bounds): Remove.
21085 * tree-pass.h (make_pass_record_bounds): Likewise.
21086 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
21087 make_pass_record_bounds): Likewise.
21088 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
21089 not free niter estimates at the beginning but at the end.
21090 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
21092 2017-06-16 Richard Biener <rguenther@suse.de>
21094 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
21095 initializer to workaround ICE in host GCC 4.8.
21097 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
21099 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
21101 (clone_inlined_nodes): Update.
21103 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21105 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
21106 prefetch settings, and enable prefetching by default at -O3.
21108 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21110 * config/aarch64/aarch64.c (aarch64_override_options_internal):
21111 Set flag_prefetch_loop_arrays according to tuning data.
21113 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21115 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
21116 New tune structure.
21117 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
21118 [Unrelated to main purpose of the patch] Place the pointer field last
21119 to enable type checking errors when tune structure are wrongly merged.
21120 * config/aarch64/aarch64.c (generic_prefetch_tune,)
21121 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
21122 (thunderx2t99_prefetch_tune): New tune constants.
21123 (tune_params *_tunings): Update all tunings (no functional change).
21124 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
21125 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
21126 from tunings structures.
21128 2017-06-16 Jakub Jelinek <jakub@redhat.com>
21131 * ubsan.c (instrument_null): Add T argument, use it instead
21132 of computing it based on IS_LHS.
21133 (instrument_object_size): Likewise.
21134 (pass_ubsan::execute): Adjust instrument_null and
21135 instrument_object_size callers to pass gimple_get_lhs or
21136 gimple_assign_rhs1 result to it. Use instrument_null instead of
21137 calling get_base_address and instrument_mem_ref. Handle
21138 aggregate call arguments for object-size sanitization.
21140 2017-06-16 Yury Gribov <tetra2005@gmail.com>
21142 PR tree-optimization/81089
21143 * tree-vrp.c (is_masked_range_test): Validate operands of
21146 2017-06-15 Martin Sebor <msebor@redhat.com>
21149 * dumpfile.c (dump_register): Avoid calling memset to initialize
21150 a class with a default ctor.
21151 * gcc.c (struct compiler): Remove const qualification.
21152 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
21153 * hash-table.h: Ditto.
21154 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
21156 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
21157 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
21159 * params.h (struct param_info): Make struct members non-const.
21160 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
21161 with default initialization.
21162 * vec.h (vec_copy_construct, vec_default_construct): New helper
21164 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
21165 with vec_copy_construct.
21166 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
21167 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
21168 * doc/invoke.texi (-Wclass-memaccess): Document.
21170 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21172 * emit-rtl.h (is_leaf): Update comment about local
21173 register allocator.
21175 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21178 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
21179 for a variable to have a section before checking if the section has a
21181 Set section to.persistent if persistent attribute is set.
21182 Warn if .persistent attribute is used on an automatic variable.
21184 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
21186 PR rtl-optimization/80474
21187 * reorg.c (update_block): Do not ignore instructions in a delay slot.
21189 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
21191 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
21194 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
21196 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
21197 (casesi): Emit bounds checking as RTL.
21198 (casesi_internal_mips16_<mode>): Remove bounds checking.
21200 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
21202 * config/xtensa/xtensa.c (xtensa_option_override): Append
21203 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
21204 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
21205 xtensa_doloop_hooks): Define unconditionally.
21206 (xtensa_reorg_loops): Only call reorg_loops in the presence of
21208 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
21209 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
21210 for it in xtensa_option_override.
21211 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
21212 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
21214 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
21216 * doc/cppopts.texi: Document '-' special value to -MF.
21218 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
21220 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
21221 (cortex_a53_fconst): Likewise.
21222 (cortex_a53_fpmul): Likewise.
21223 (cortex_a53_f_load_64): Likewise.
21224 (cortex_a53_f_load_many): Likewise.
21225 (cortex_a53_advsimd_alu): Likewise.
21226 (cortex_a53_advsimd_alu_q): Likewise.
21227 (cortex_a53_advsimd_mul): Likewise.
21228 (cortex_a53_advsimd_mul_q): Likewise.
21229 (fpmac bypass): Add new bypass for fpmac-fpmac case.
21230 Add missing fmul, r2f_cvt and fconst cases.
21232 2017-06-14 Richard Biener <rguenther@suse.de>
21234 PR middle-end/81088
21235 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
21237 (fold_binary_loc): When associating do not treat pre-existing
21238 TREE_OVERFLOW on literal constants as a reason to allow
21239 TREE_OVERFLOW on associated literal constants.
21241 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21243 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
21244 (MASK_FEATURES): New macro.
21245 * config/sparc/sparc.c (sparc_option_override): Remove the special
21246 handling of -mfpu and generalize it to all MASK_FEATURES switches.
21248 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21250 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
21251 a division of 0 if non-call exceptions are enabled.
21253 2017-06-14 Andrew Pinski <apinski@cavium.com>
21254 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21257 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
21258 Improve vector initialization code gen for only variable case.
21260 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
21262 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
21264 2017-06-14 Richard Biener <rguenther@suse.de>
21266 PR tree-optimization/81083
21267 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
21270 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21272 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
21273 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
21274 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
21275 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
21276 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
21277 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
21279 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21281 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
21282 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
21284 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21286 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
21288 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21290 * config/rs6000/t-rtems: Don't handle SPE.
21292 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21294 * config/rs6000/t-linux: Don't handle SPE.
21296 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21298 * config/rs6000/eabispe.h: Delete file.
21300 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21302 * config/rs6000/t-spe: Delete file.
21304 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21306 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
21307 (rs6000_legitimate_offset_address_p): Return false for anything in
21308 V2SImode or V2SFmode.
21310 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
21312 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
21313 except V2SF and V2SI. Rearrange the vector modes, and add comments.
21314 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
21316 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
21317 (rs6000_legitimate_offset_address_p): Ditto.
21318 (rs6000_emit_move): Ditto.
21319 (rs6000_init_builtins): Remove V4HI_type_node.
21321 2017-06-13 Martin Liska <mliska@suse.cz>
21324 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
21325 (gate_asan): Likewise.
21326 * asan.h (asan_no_sanitize_address_p): Remove the function.
21327 (sanitize_flags_p): New function.
21328 * builtins.def: Fix coding style.
21329 * common.opt: Use renamed enum value.
21330 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
21331 * doc/extend.texi: Document no_sanitize attribute.
21332 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
21333 to SANITIZE_UNDEFINED_NONDEFAULT.
21334 * gcc.c (sanitize_spec_function): Use the renamed enum value.
21335 * gimple-fold.c (optimize_atomic_compare_exchange_p):
21336 Use sanitize_flags_p.
21337 * gimplify.c (gimplify_function_tree): Likewise.
21338 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
21339 * opts.c (parse_no_sanitize_attribute): New function.
21340 (common_handle_option): Use renamed enum value.
21341 * opts.h (parse_no_sanitize_attribute): Declare.
21342 * tree.c (sanitize_flags_p): New function.
21343 * tree.h: Declared here.
21344 * tsan.c: Use sanitize_flags_p.
21345 * ubsan.c (ubsan_expand_null_ifn): Likewise.
21346 (instrument_mem_ref): Likewise.
21347 (instrument_bool_enum_load): Likewise.
21348 (do_ubsan_in_current_function): Remove the function.
21349 (pass_ubsan::execute): Use sanitize_flags_p.
21350 * ubsan.h: Remove do_ubsan_in_current_function
21351 * tree-cfg.c (print_no_sanitize_attr_value): New function.
21352 (dump_function_to_file): Use it here.
21354 2017-06-13 Martin Jambor <mjambor@suse.cz>
21356 PR tree-optimization/80803
21357 PR tree-optimization/81063
21358 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
21359 (propagate_subaccesses_across_link): Enqueue subtree whenever
21360 necessary instead of relying on the caller.
21362 2017-06-13 Martin Jambor <mjambor@suse.cz>
21364 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
21365 that have a first_link.
21366 (sort_and_splice_var_accesses): Do not check first_link before
21368 (subtree_mark_written_and_enqueue): Likewise.
21369 (propagate_all_subaccesses): Likewise and do not stop at first
21370 parent with a first_link.
21372 2017-06-13 Martin Jambor <mjambor@suse.cz>
21374 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
21377 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21379 * match.pd: New pattern.
21381 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21383 * tree-vrp.c (is_masked_range_test): New function.
21384 (register_edge_assert_for): Determine ranges for
21387 2017-06-13 Yury Gribov <tetra2005@gmail.com>
21389 PR tree-optimization/67328
21390 * fold-const.c (maskable_range_p): New function.
21391 (build_range_check): Generate bittests if possible.
21393 2017-06-13 Martin Liska <mliska@suse.cz>
21395 * gimple-pretty-print.c (dump_probability): Add new argument.
21396 (dump_edge_probability): Dump both probability and count.
21397 (dump_gimple_label): Likewise.
21398 (dump_gimple_bb_header): Likewise.
21400 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
21403 * config/avr/avr-devices.c: Fix indentation.
21404 * config/avr/gen-avr-mmcu-specs.c: Dito.
21406 2017-06-13 Richard Biener <rguenther@suse.de>
21408 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
21409 instead get vector type from stmt_info.
21410 (vectorizable_reduction): Adjust. Remove dead code.
21412 2017-06-13 Richard Biener <rguenther@suse.de>
21414 PR middle-end/81065
21415 * fold-const.c (extract_muldiv_1): Remove bogus distribution
21416 case of C * (x * C2 + C3).
21417 (fold_addr_of_array_ref_difference): Properly fold index difference.
21419 2017-06-12 David S. Miller <davem@davemloft.net>
21422 * config/sparc/sparc.md (return expander): Emit frame blockage if
21423 function uses alloca.
21425 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
21427 * combine.c (make_field_assignment): Check len rather than the mode
21428 precision when calling force_to_mode.
21430 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
21432 Support multilibs and devices that see flash in RAM address range.
21435 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
21436 (avr_mcu_t) <flash_pm_offset>: New field.
21437 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
21438 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
21439 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
21440 (AVR_TINY_PM_OFFSET): Remove macro.
21441 * config/avr/avr.opt (-mshort-calls): New option.
21442 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
21443 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
21444 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
21445 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
21446 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
21447 instead of avr_arch->have_jmp_call.
21448 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
21449 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
21450 avr_arch->flash_pm_offset to define.
21451 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
21452 new field flash_pm_offset. Add entry for avrxmega3.
21453 (avr_texinfo): Add entry for avrxmega3.
21454 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
21455 attiny212, attiny214,
21456 attiny412, attiny414, attiny416, attiny417,
21457 attiny814, attiny816, attiny817,
21458 attiny1614, attiny1616, attiny1617,
21459 attiny3214, attiny3216, attiny3217.
21460 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
21461 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
21462 (avr_print_operand_address) [AVR_TINY]: Same.
21463 (avr_asm_init_sections) <readonly_data_section>: Only patch
21464 callback if avr_arch->flash_pm_offset = 0.
21465 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
21466 for rodata if avr_arch->flash_pm_offset != 0.
21467 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
21468 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
21469 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
21470 (m_options): Append opt_rcall.
21471 (m_dirnames): Append dir_rcall.
21472 * config/avr/t-multilib: Regenerate.
21474 * configure.ac [target=avr]: Check whether avrxmega3 default
21475 linker description file works as needed.
21476 * configure: Regenerate.
21477 * doc/avr-mmcu.texi: Regenerate.
21478 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
21479 <__AVR_ARCH__>: Document avrxmega3 and 103.
21480 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
21481 <__AVR_SHORT_CALLS__>: Document it.
21482 <__AVR_PM_BASE_ADDRESS__>: Document it.
21483 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
21484 (AVR Variable Attributes) <progmem>: Document this is
21485 not needed for avrxmega3.
21486 (AVR Named Address Spaces) <__flash>: Dito.
21488 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
21490 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
21492 2017-06-12 Doug Rupp <rupp@adacore.com>
21494 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
21495 Append vxworks-stdint.h to the tm_file list.
21496 * config/vxworks-stdint.h: New file.
21498 2017-06-12 Martin Liska <mliska@suse.cz>
21500 PR tree-optimization/81041
21501 * tree-profile.c (gimple_gen_ic_func_profiler):
21502 Create an extra BB in profile-generate
21503 (gimple_gen_time_profiler): Likewise.
21505 2017-06-12 Jakub Jelinek <jakub@redhat.com>
21507 PR tree-optimization/81003
21508 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
21509 (update_range_test): Use it instead of force_gimple_operand_gsi.
21511 2017-06-12 Richard Biener <rguenther@suse.de>
21513 PR tree-optimization/81053
21514 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
21515 with backedge value not defined in loop. Simplify def stmt
21518 2017-06-11 Tom de Vries <tom@codesourcery.com>
21521 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
21523 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
21524 nvptx_cannot_force_const_mem.
21526 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21528 * opts.c (finish_options): Move test for flag_split_stack after
21529 it has been initialized.
21531 2017-06-11 Jason Merrill <jason@redhat.com>
21533 * tree.h (id_equal): New.
21534 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
21535 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
21536 instead of strcmp of IDENTIFIER_POINTER.
21538 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21540 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
21541 (mark_all_inlined_calls_cdtor): Fix formating.
21542 (inline_transform): Rescale profile before inlining.
21544 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21546 * cgraph.h (cgraph_edge::clone): Update prototype.
21547 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
21548 (cgraph_node::create_clone): Update.
21549 (cgraph_node::create_version_clone): Update.
21550 * tree-inline.c (copy_bb): Update.
21551 (expand_call_inline): Update.
21553 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21555 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
21556 factored out from ...
21557 (rs6000_emit_prologue): ... here.
21559 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
21561 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
21562 factored out from ...
21563 (rs6000_emit_prologue): ... here.
21565 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21567 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
21569 (handle_missing_profiles): Fix computation of tp_first_run.
21570 (counts_to_freqs): Do not touch freqs when count is 0.
21572 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
21574 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
21577 2017-06-10 Tom de Vries <tom@codesourcery.com>
21579 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
21580 attributes): Document signal effective target.
21582 2017-06-10 Tom de Vries <tom@codesourcery.com>
21584 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
21585 Document effective target stack_size.
21587 2017-06-09 David Malcolm <dmalcolm@redhat.com>
21589 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
21590 to the edit_context if they can be auto-applied.
21592 2017-06-9 Ian Lance Taylor <iant@golang.org>
21594 * opts.c (finish_options): If -fsplit-stack, disable implicit
21595 -forder-blocks-and-partition.
21596 * doc/invoke.texi (Optimize Options): Document that when using
21597 -fsplit-stack -forder-blocks-and-partition is not implicitly
21600 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21602 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
21603 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
21604 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
21605 * builtins.def (abort, trap, unreachable): Declare cold.
21606 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
21607 * tree-core.h (ECF_COLD): New.
21608 * tree.c (set_call_expr_flags): Handle ECF_COLD.
21609 (build_common_builtin_nodes): Mark unreachable and abort as cold.
21611 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21613 * predict.c (unlikely_executed_stmt_p): Cleanup.
21615 2017-06-09 Richard Biener <rguenther@suse.de>
21617 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
21618 model if the ref is always written to.
21620 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21622 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
21624 2017-06-09 Tamar Christina <tamar.christina@arm.com>
21626 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
21629 2017-06-09 Tom de Vries <tom@codesourcery.com>
21632 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
21633 "target cannot support label values" when encountering LABEL_REF.
21635 2017-06-09 Martin Liska <mliska@suse.cz>
21637 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
21638 (gimple_gen_ic_func_profiler): Emit direct comparison
21639 of __gcov_indirect_call_callee with NULL.
21640 (gimple_gen_time_profiler): Change probability from
21641 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
21643 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21645 * profile.c (edge_gcov_counts): Turn to pointer.
21646 (compute_branch_probabilities, compute_branch_probabilities): Update.
21647 (branch_prob): Do not clear edge_gcov_count.
21648 * profile.h (edge_gcov_counts): Turn to pointer.
21649 (edge_gcov_count): Update.
21651 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
21653 * gimple.h (gimple_check_failed): Mark cold.
21655 2017-06-09 Richard Biener <rguenther@suse.de>
21657 PR tree-optimization/66623
21658 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
21659 refactor check_reduction into two parts, properly computing
21660 whether we have to check reduction validity for outer loop
21663 2017-06-09 Richard Biener <rguenther@suse.de>
21665 PR tree-optimization/79483
21666 * graphite-scop-detection.c (order): New global.
21667 (get_order): Compute bb to order mapping that satisfies code
21668 generation constraints.
21669 (cmp_pbbs): New helper.
21670 (build_scops): Start domwalk at entry block, sort generated
21673 2017-06-09 Richard Biener <rguenther@suse.de>
21675 PR middle-end/81007
21676 * ipa-polymorphic-call.c
21677 (ipa_polymorphic_call_context::restrict_to_inner_class):
21678 Skip FIELD_DECLs with error_mark_node type.
21679 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
21682 2017-06-09 Martin Liska <mliska@suse.cz>
21684 * predict.c (struct branch_predictor): New struct.
21685 (test_prediction_value_range): New test.
21686 (predict_c_tests): New function.
21687 * selftest-run-tests.c (selftest::run_tests): Run the function.
21688 * selftest.h: Declare new tests.
21690 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
21693 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
21694 gen_add3_insn did not fail.
21695 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
21696 r0, construct that number in a temporary reg and add that reg to r0.
21697 If asked to put the result in r0 as well, fail.
21699 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
21701 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21702 for early expansion of vec_eqv.
21704 2017-06-08 Jakub Jelinek <jakub@redhat.com>
21706 PR middle-end/81005
21707 * ubsan.c (instrument_null): Avoid pointless code temporary.
21708 (pass_ubsan::execute): Instrument aggregate arguments of calls.
21710 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
21714 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
21717 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
21720 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21722 * predict.c (unlikely_executed_edge_p): Move ahead.
21723 (probably_never_executed_edge_p): Use it.
21725 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21727 PR middle-end/79988
21728 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
21729 gimple_call_builtin_p call.
21731 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21733 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
21734 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
21735 rtl_check_failed_type2, rtl_check_failed_code1,
21736 rtl_check_failed_code2, rtl_check_failed_code_mode,
21737 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
21738 rtvec_check_failed_bounds, rtl_check_failed_flag,
21739 _fatal_insn_not_found, _fatal_insn): Likewise.
21740 * tree.h (tree_contains_struct_check_failed,
21741 tree_check_failed, tree_not_check_failed,
21742 tree_class_check_failed, tree_range_check_failed,
21743 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
21744 tree_vec_elt_check_failed, phi_node_elt_check_failed,
21745 tree_operand_check_failed, omp_clause_check_failed,
21746 omp_clause_operand_check_failed, omp_clause_range_check_failed):
21749 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21751 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
21752 flag_branch_probabilities.
21753 * ipa-inline.c (edge_badness): Likewise.
21754 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
21755 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
21756 * predict.c (maybe_hot_frequency_p): Likewise.
21757 (probably_never_executed): Likewise.
21758 * sched-ebb.c (schedule_ebbs): Likewise.
21759 * sched-rgn.c (find_single_block_region): Likewise.
21760 * tracer.c (tail_duplicate): Likewise.
21762 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21764 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
21765 longer requires x_flag_profile_use.
21767 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21769 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
21770 instead of flag_reorder_blocks_and_partition.
21771 * dbxout.c (dbxout_function_end): Likewise.
21772 * dwarf2out.c (gen_subprogram_die): Likewise.
21773 * haifa-sched.c (sched_create_recovery_edges): Likewise.
21774 * hw-doloop.c (reorg_loops): Likewise.
21775 * varasm.c (assemble_start_function,
21776 assemble_end_function): Likewise.
21777 (decide_function_section): Do not check for
21778 flag_reorder_blocks_and_partition.
21780 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
21782 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
21784 (chkp_get_hard_register_fake_addr_expr): Ditto.
21785 (chkp_build_addr_expr): Add check for hard reg case.
21786 (chkp_parse_array_and_component_ref): Ditto.
21787 (chkp_find_bounds_1): Ditto.
21788 (chkp_process_stmt): Don't generate bounds store for
21791 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21793 * predict.c (maybe_hot_bb_p): Do not check profile status.
21794 (maybe_hot_edge_p): Likewise.
21795 (probably_never_executed): Check for zero counts even if profile
21797 (unlikely_executed_edge_p): New function.
21798 (unlikely_executed_stmt_p): New function.
21799 (unlikely_executed_bb_p): New function.
21800 (set_even_probabilities): Use unlikely predicates.
21801 (combine_predictions_for_bb): Likewise.
21802 (predict_paths_for_bb): Likewise.
21803 (predict_paths_leading_to_edge): Likewise.
21804 (determine_unlikely_bbs): New function.
21805 (estimate_bb_frequencies): Use it.
21806 (compute_function_frequency): Use zero counts even if profile is
21808 * profile-count.h: Fix typo.
21810 2017-08-08 Julia Koval <julia.koval@intel.com>
21812 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
21813 _mm512_mask_cvtsepi16_storeu_epi8,
21814 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
21815 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
21816 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
21817 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
21818 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
21819 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
21820 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
21821 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
21822 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
21823 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
21824 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
21825 __builtin_ia32_pmovuswb256mem_mask,
21826 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
21827 __builtin_ia32_pmovwb512mem_mask): New builtins.
21829 2017-08-08 Julia Koval <julia.koval@intel.com>
21831 PR target/73350,80862
21832 * config/i386/subst.md (round): Fix round pattern.
21833 * config/i386/i386.c (ix86_erase_embedded_rounding):
21834 Fix erasing rounding for the fixed pattern.
21836 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
21838 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
21840 2017-06-08 Martin Liska <mliska@suse.cz>
21842 PR gcov-profile/80911
21843 * gcov.c (block_info::block_info): New constructor.
21845 2017-06-07 Carl Love <cel@us.ibm.com>
21847 * config/rs6000/rs6000-c: The return type of the following
21848 built-in functions was implemented as int not long long. Fix sign
21849 of return value for the unsigned version of vec_mulo and vec_mule.
21850 vector unsigned long long vec_bperm (vector unsigned long long,
21851 vector unsigned char)
21852 vector signed long long vec_mule (vector signed int,
21854 vector unsigned long long vec_mule (vector unsigned int,
21855 vector unsigned int)
21856 vector signed long long vec_mulo (vector signed int,
21858 vector unsigned long long vec_mulo (vector unsigned int,
21859 vector unsigned int)
21860 * doc/extend.texi: Fix the documentation for the built-in
21863 2017-06-07 Carl Love <cel@us.ibm.com>
21866 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
21869 2017-06-07 Carl Love <cel@us.ibm.com>
21871 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
21872 support, Generate doublehv for signed int/float for BE case only.
21874 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
21876 * doc/invoke.texi (mcx16): Rewrite.
21878 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21880 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
21881 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
21882 *mov<mode>_softfloat, and an anonymous splitter): Use
21883 nonimmediate_operand instead of rs6000_nonimmediate_operand.
21885 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21887 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
21889 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21890 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
21891 (rs6000_debug_reg_global): Adjust.
21892 (rs6000_init_hard_regno_mode_ok): Adjust.
21893 (rs6000_dbx_register_number): Adjust.
21894 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
21895 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21896 Remove SPE_ACC and SPEFSCR.
21897 (REG_ALLOC_ORDER): Ditto.
21898 (FRAME_POINTER_REGNUM): Change to 111.
21899 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
21900 (REG_CLASS_NAMES): Ditto.
21901 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
21902 (REGISTER_NAMES): Ditto.
21903 (ADDITIONAL_REG_NAMES): Ditto.
21904 (rs6000_reg_names): Ditto.
21905 * config/rs6000/rs6000.md: Renumber some register number
21908 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21910 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
21912 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
21913 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
21915 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
21916 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
21917 Delete the SPE high registers.
21918 (REG_ALLOC_ORDER): Ditto.
21919 (enum reg_class): Remove SPE_HIGH_REGS.
21920 (REG_CLASS_NAMES): Ditto.
21921 (REG_CLASS_CONTENTS): Delete the SPE high registers.
21922 (REGISTER_NAMES): Ditto.
21923 (rs6000_reg_names): Ditto.
21924 * doc/tm.texi.in: Remove SPE as example.
21925 * doc/tm.texi: Regenerate.
21927 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21929 * config/rs6000/8540.md (ppc8540_brinc): Delete.
21930 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
21931 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
21932 * config/rs6000/rs6000.md (type): Remove "brinc".
21934 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21936 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
21937 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
21938 * config/rs6000/linuxspe.h: Delete file.
21939 * config/rs6000/rs6000.md: Don't include spe.md.
21940 * config/rs6000/spe.h: Delete file.
21941 * config/rs6000/spe.md: Delete file.
21942 * config/rs6000/t-rs6000: Remove spe.md.
21944 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21946 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
21947 (reg_or_none500mem_operand): Delete.
21948 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
21949 instead of reg_or_none500mem_operand.
21951 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21953 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21954 handling of SPE flags.
21955 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
21957 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
21959 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
21961 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
21962 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
21963 paired_divv2sf3): Similar.
21964 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
21965 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
21966 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
21968 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
21969 Rename the paired_* instruction patterns.
21970 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
21972 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
21973 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
21974 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
21975 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
21976 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
21977 PAIRED_VECTOR_MODE.
21978 (struct machine_function): Delete field spe_insn_chain_scanned_p.
21979 (spe_func_has_64bit_regs_p): Delete.
21980 (spe_expand_predicate_builtin): Delete.
21981 (spe_expand_evsel_builtin): Delete.
21982 (TARGET_DWARF_REGISTER_SPAN): Do not define.
21983 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
21984 (invalid_e500_subreg): Delete.
21985 (rs6000_legitimize_address): Always force_reg op2 as well, for
21986 paired single memory accesses.
21987 (rs6000_member_type_forces_blk): Delete.
21988 (rs6000_spe_function_arg): Delete.
21989 (rs6000_expand_unop_builtin): Delete SPE handling.
21990 (rs6000_expand_binop_builtin): Ditto.
21991 (spe_expand_stv_builtin): Delete.
21992 (bdesc_2arg_spe): Delete.
21993 (spe_expand_builtin): Delete.
21994 (spe_expand_predicate_builtin): Delete.
21995 (spe_expand_evsel_builtin): Delete.
21996 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
21997 (spe_init_builtins): Delete.
21998 (spe_func_has_64bit_regs_p): Delete.
21999 (savres_routine_name): Delete "info" parameter. Adjust callers.
22000 (rs6000_emit_stack_reset): Ditto.
22001 (rs6000_dwarf_register_span): Delete.
22002 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
22003 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
22004 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
22006 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
22008 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
22009 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
22010 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
22011 mulv2sf3, divv2sf3): Delete expanders.
22013 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22015 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
22017 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22019 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
22020 * config/rs6000/rs6000.c: Ditto.
22022 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22024 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
22025 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
22026 comparison_operator.
22028 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22030 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
22031 * config/rs6000/rs6000.opt: Ditto.
22032 * config/rs6000/t-rtems: Ditto.
22034 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22036 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
22037 TARGET_E500_SINGLE by 0, simplify.
22038 * config/rs6000/rs6000.c: Ditto.
22039 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
22040 (spe_build_register_parallel): Delete.
22041 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
22042 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
22043 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
22044 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
22045 (E500_CONVERT): Delete.
22046 * config/rs6000/spe.md: Remove many patterns and all define_constants.
22048 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
22050 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
22051 * config/rs6000/dfp.md: Ditto.
22052 (negdd2, *negdd2_fpr): Merge.
22053 (absdd2, *absdd2_fpr): Merge.
22054 (negtd2, *negtd2_fpr): Merge.
22055 (abstd2, *abstd2_fpr): Merge.
22056 * config/rs6000/e500.h: Delete file.
22057 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
22058 TARGET_FPRS by 1 and simplify.
22059 * config/rs6000/rs6000-c.c: Ditto.
22060 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
22061 TARGET_DF_SPE by 0.
22062 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
22064 * config/rs6000/rs6000.md: Ditto.
22065 (floatdidf2, *floatdidf2_fpr): Merge.
22066 (move_from_CR_gt_bit): Delete.
22067 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
22068 (E500_CR_IOR_COMPARE): Delete.
22069 (All patterns that require !TARGET_FPRS): Delete.
22070 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
22072 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22074 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
22076 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22078 * graphds.c (add_edge): Intitialize edge's attached data.
22079 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
22080 pointer parameter. Call pointed function on each edge during
22081 graph traversing. Skip traversing the edge when the function
22083 (graphds_dfs, graphds_scc): Ditto.
22084 (for_each_edge): New parameter. Pass the new parameter to callback
22086 * graphds.h (skip_edge_callback): New function pointer type.
22087 (graphds_dfs, graphds_scc): New function pointer parameter.
22088 (graphds_edge_callback, for_each_edge): New parameter.
22090 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22092 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
22093 out code checking if runtime alias check is possible to below ...
22094 Call the new function.
22095 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
22096 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
22098 2017-06-07 Marek Polacek <polacek@redhat.com>
22101 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
22102 TYPE_OVERFLOW_WRAPS check.
22104 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22106 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
22107 if versioning is required.
22108 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
22109 peeling with the check for versioning.
22111 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22113 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
22114 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
22115 Set true to new parameter if new ssa variable is defined.
22116 (vect_gen_vector_loop_niters): Refactor. Set range information
22117 for the new vector loop bound variable.
22118 (vect_do_peeling): Ditto.
22120 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22122 * tree-affine.c (ssa.h): Include header file.
22123 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
22124 has wrapping overflow behavior.
22126 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22128 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
22130 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22132 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
22133 (tree_to_aff_combination): ... here.
22135 2017-06-07 Bin Cheng <bin.cheng@arm.com>
22137 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
22138 reg_pressure model function.
22139 (ivopts_global_cost_for_size): Delete.
22140 (determine_set_costs, iv_ca_recount_cost): Call new model function
22141 ivopts_estimate_reg_pressure.
22143 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22145 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
22146 expensive than udiv. Remove floating point cases from mod.
22148 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22150 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
22151 Increase idiv cost.
22153 2017-06-07 Tamar Christina <tamar.christina@arm.com>
22155 * config/aarch64/aarch64.md
22156 (copysignsf3): Fix mask generation.
22158 2017-06-07 Jakub Jelinek <jakub@redhat.com>
22160 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
22162 (class dump_manager): Add register_dumps method.
22163 * dumpfile.c: Include langhooks.h.
22164 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
22165 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
22166 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
22167 (dump_manager::dump_register): Start with 512 entries instead of 32.
22168 (dump_manager::register_dumps): New method.
22169 * toplev.c (general_init): Instead of invoking register_dumps
22170 langhook, invoke register_dumps method on the dump manager.
22171 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
22174 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
22176 * doc/md.texi: Clarify the restrictions on a define_insn condition.
22177 Say that # requires an associated define_split to exist, and that
22178 the define_split must be suitable for use after register allocation.
22180 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22182 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
22183 (compute_outgoing_frequencies): Also initialize zero counts.
22184 (find_many_sub_basic_blocks): Do not produce uninitialized profile
22185 around loops; preserve more of profile when nothing changes.
22187 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
22189 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
22191 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
22192 * config/arm/arm-cpu-cdata.h: Regenerate.
22193 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
22194 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
22195 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
22196 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
22197 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
22199 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
22201 * config/arm/t-rmprofile: Likewise.
22202 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
22204 2017-06-06 David S. Miller <davem@davemloft.net>
22207 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
22208 blockage if function uses alloca.
22210 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22212 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
22213 New "uid" fields to hold pretty-print IDs of group and ref.
22214 Memory references are now identified as <group_id>:<ref_id>
22215 instead of using [random] addresses.
22216 (dump_mem_details): Simplify, no functional change.
22217 (dump_mem_ref): Simplify and make output more concise.
22218 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
22219 (find_or_create_group): Initialize group uid.
22220 (record_ref): Initialize ref uid. Improve debug output.
22221 (prune_group_by_reuse, should_issue_prefetch_p,)
22222 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
22223 (mark_nontemporal_store, determine_loop_nest_reuse):
22224 Improve debug output.
22226 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
22228 * dbgcnt.def (prefetch): New debug counter.
22229 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
22230 (schedule_prefetches): Stop issueing prefetches if debug counter
22233 2017-06-06 Tom de Vries <tom@codesourcery.com>
22235 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
22236 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
22238 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22240 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
22241 Use aarch64_reg_or_zero predicate for operand 4.
22242 (aarch64_compare_and_swap<mode> define_insn_and_split):
22243 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
22244 (aarch64_store_exclusive<mode>): Likewise for operand 2.
22246 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
22248 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
22249 (arm_compute_save_core_reg_mask): This.
22250 (thumb1_compute_save_reg_mask): Rename into ...
22251 (thumb1_compute_save_core_reg_mask): This.
22252 (arm_compute_save_reg0_reg12_mask): Adapt comment.
22253 (arm_compute_frame_layout): Likewise.
22255 2017-06-06 Richard Biener <rguenther@suse.de>
22257 PR tree-optimization/80974
22258 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
22259 keep or clear leaders SSA info.
22261 2017-06-06 Tom de Vries <tom@codesourcery.com>
22263 * config/nvptx/nvptx.c (split_mode_p): New function.
22264 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
22266 2017-06-06 Tom de Vries <tom@codesourcery.com>
22268 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
22270 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22273 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
22276 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22278 * shrink-wrap.c (handle_simple_exit): Update profile.
22279 (try_shrink_wrapping): Upate profile.
22281 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22283 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
22284 (tree_guess_outgoing_edge_probabilities): New.
22285 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
22286 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
22288 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22290 * ipa-split.c (split_function): Initialize return bb profile.
22292 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
22294 * profile.c (compute_branch_probabilities): Also initialize
22295 EXIT_BLOCK profile.
22297 2017-06-06 Richard Biener <rguenther@suse.de>
22299 PR tree-optimization/80928
22300 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
22301 (vect_analyze_loop_operations): Properly guard analysis for
22303 (vect_transform_loop): Likewise.
22304 (vect_analyze_loop_2): Also reset SLP type on PHIs.
22305 (vect_model_induction_cost): Do not cost for pure SLP.
22306 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
22307 of induction in inner loop vectorization.
22308 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
22309 (vect_get_and_check_slp_defs): Handle vect_induction_def.
22310 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
22312 (vect_analyze_slp_cost_1): Cost induction.
22313 (vect_detect_hybrid_slp_stmts): Handle PHIs.
22314 (vect_get_slp_vect_defs): Likewise.
22315 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
22316 (vect_transform_stmt): Handle SLP reductions.
22317 * tree-vectorizer.h (vectorizable_induction): Adjust.
22319 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22321 * config/rs6000/rs6000.c (make_resolver_func): Update
22322 init_lowered_empty_function call.
22324 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
22326 * doc/invoke.texi: Document the -fprofile-abs-path option.
22327 * common.opt (fprofile-abs-path): New option.
22328 * gcov-io.h (gcov_write_filename): Declare.
22329 * gcov-io.c (gcov_write_filename): New function.
22330 * coverage.c (coverage_begin_function): Use gcov_write_filename.
22331 * profile.c (output_location): Likewise.
22333 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22335 * shring-wrap.c: Revert accidental commit.
22337 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
22339 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
22341 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22343 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
22345 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
22346 profile in callgraph edge.
22347 * profile-count.h (apply_probability): If THIS is 0, then result is 0
22348 (apply_scale): Likewise.
22349 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
22350 Also scale profile when inlining function with zero profile.
22351 (initialize_cfun): Update exit block profile even when it is zero.
22352 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
22353 when profile is read.
22355 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
22357 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
22358 (CLONE_*): New constants to define the processors we can generate
22359 code for with the target_clone attribute.
22360 (rs6000_clone_map): New array to identify which clone processors
22361 the current program is running on.
22362 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
22363 target_clone attribute.
22364 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
22365 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
22366 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
22367 (cpu_expand_builtin): Add support for target_clone attribute.
22368 (rs6000_valid_attribute_p): Allow "default" attribute.
22369 (get_decl_name): New debug function to simplify printing the
22370 current function name in debugging statements.
22371 (rs6000_clone_priority): New functions to support the target_clone
22372 attribute, and be able to generate code to switch between ISA 2.05
22373 through ISA 3.0 (power6 through power9).
22374 (rs6000_compare_version_priority): Likewise.
22375 (rs6000_get_function_versions_dispatcher): Likewise.
22376 (make_resolver_func): Likewise.
22377 (add_condition_to_bb): Likewise.
22378 (dispatch_function_versions): Likewise.
22379 (rs6000_generate_version_dispatcher_body): Likewise.
22380 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
22381 (fusion_gpr_load_p): Fix a spacing issue.
22382 * doc/extend.texi (Common Function Attributes): Document that the
22383 PowerPC supports the target_clone attribute.
22385 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
22387 * config/arm/arm.h: explain F symbol found in description of ARM
22388 register allocation in its legend.
22390 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22392 * config/mips/frame-header-opt.c: Include profile-count.h.
22393 * config/riscv/riscv.c: Include profile-count.h
22395 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
22397 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
22399 (sm_set_flag_if_changed): Add bbs field.
22400 (execute_sm_if_changed_flag_set): Pass BBS.
22401 (execute_sm): Update.
22403 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22405 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
22408 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22410 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
22411 (peephole2): New peephole2 to emit the above.
22412 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
22414 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22416 * config/aarch64/aarch64.c (define_peephole2 above
22417 *sub_<shift>_<mode>): New peephole.
22419 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22421 * config/i386/i386.c (make_resolver_func): Update.
22422 * Makefile.in: Add profile-count.h and profile-count.o
22423 * auto-profile.c (afdo_indirect_call): Update to new API.
22424 (afdo_set_bb_count): Update.
22425 (afdo_propagate_edge): Update.
22426 (afdo_propagate_circuit): Update.
22427 (afdo_calculate_branch_prob): Update.
22428 (afdo_annotate_cfg): Update.
22429 * basic-block.h: Include profile-count.h
22430 (struct edge_def): Turn count to profile_count.
22431 (struct basic_block_def): Likewie.
22432 (REG_BR_PROB_BASE): Move to profile-count.h
22433 (RDIV): Move to profile-count.h
22434 * bb-reorder.c (max_entry_count): Turn to profile_count.
22435 (find_traces): Update.
22436 (rotate_loop):Update.
22437 (connect_traces):Update.
22438 (sanitize_hot_paths):Update.
22439 * bt-load.c (migrate_btr_defs): Update.
22440 * cfg.c (RDIV): Remove.
22441 (init_flow): Use alloc_block.
22442 (alloc_block): Uninitialize count.
22443 (unchecked_make_edge): Uninitialize count.
22444 (check_bb_profile): Update.
22445 (dump_edge_info): Update.
22446 (dump_bb_info): Update.
22447 (update_bb_profile_for_threading): Update.
22448 (scale_bbs_frequencies_int): Update.
22449 (scale_bbs_frequencies_gcov_type): Update.
22450 (scale_bbs_frequencies_profile_count): New.
22451 * cfg.h (update_bb_profile_for_threading): Update.
22452 (scale_bbs_frequencies_profile_count): Declare.
22453 * cfgbuild.c (compute_outgoing_frequencies): Update.
22454 (find_many_sub_basic_blocks): Update.
22455 * cfgcleanup.c (try_forward_edges): Update.
22456 (try_crossjump_to_edge): Update.
22457 * cfgexpand.c (expand_gimple_tailcall): Update.
22458 (construct_exit_block): Update.
22459 * cfghooks.c (verify_flow_info): Update.
22460 (dump_bb_for_graph): Update.
22461 (split_edge): Update.
22462 (make_forwarder_block): Update.
22463 (duplicate_block): Update.
22464 (account_profile_record): Update.
22465 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
22466 (get_estimated_loop_iterations): Update.
22467 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
22468 (single_likely_exit): Update.
22469 * cfgloopmanip.c (scale_loop_profile): Update.
22471 (set_zero_probability): Update.
22472 (lv_adjust_loop_entry_edge): Update.
22473 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
22474 (purge_dead_edges): Update.
22475 (rtl_account_profile_record): Update.
22476 * cgraph.c (cgraph_node::create): Uninitialize count.
22477 (symbol_table::create_edge): Uninitialize count.
22478 (cgraph_update_edges_for_call_stmt_node): Update.
22479 (cgraph_edge::dump_edge_flags): Update.
22480 (cgraph_node::dump): Update.
22481 (cgraph_edge::maybe_hot_p): Update.
22482 * cgraph.h: Include profile-count.h
22483 (create_clone), create_edge, create_indirect_edge): Update.
22484 (cgraph_node): Turn count to profile_count.
22485 (cgraph_edge0: Likewise.
22486 (make_speculative, clone): Update.
22487 (create_edge): Update.
22488 (init_lowered_empty_function): Update.
22489 * cgraphclones.c (cgraph_edge::clone): Update.
22490 (duplicate_thunk_for_node): Update.
22491 (cgraph_node::create_clone): Update.
22492 * cgraphunit.c (cgraph_node::analyze): Update.
22493 (cgraph_node::expand_thunk): Update.
22494 * final.c (dump_basic_block_info): Update.
22495 * gimple-streamer-in.c (input_bb): Update.
22496 * gimple-streamer-out.c (output_bb): Update.
22497 * graphite.c (print_global_statistics): Update.
22498 (print_graphite_scop_statistics): Update.
22499 * hsa-brig.c: Include basic-block.h.
22500 * hsa-dump.c: Include basic-block.h.
22501 * hsa-gen.c (T sum_slice): Update.
22502 (convert_switch_statements):Update.
22503 * hsa-regalloc.c: Include basic-block.h.
22504 * ipa-chkp.c (chkp_produce_thunks): Update.
22505 * ipa-cp.c (struct caller_statistics): Update.
22506 (init_caller_stats): Update.
22507 (gather_caller_stats): Update.
22508 (ipcp_cloning_candidate_p): Update.
22509 (good_cloning_opportunity_p): Update.
22510 (get_info_about_necessary_edges): Update.
22511 (dump_profile_updates): Update.
22512 (update_profiling_info): Update.
22513 (update_specialized_profile): Update.
22514 (perhaps_add_new_callers): Update.
22515 (decide_about_value): Update.
22516 (ipa_cp_c_finalize): Update.
22517 * ipa-devirt.c (struct odr_type_warn_count): Update.
22518 (struct decl_warn_count): Update.
22519 (struct final_warning_record): Update.
22520 (possible_polymorphic_call_targets): Update.
22521 (ipa_devirt): Update.
22522 * ipa-fnsummary.c (redirect_to_unreachable): Update.
22523 * ipa-icf.c (sem_function::merge): Update.
22524 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
22525 * ipa-inline.c (compute_uninlined_call_time): Update.
22526 (compute_inlined_call_time): Update.
22527 (want_inline_small_function_p): Update.
22528 (want_inline_self_recursive_call_p): Update.
22529 (edge_badness): Update.
22530 (lookup_recursive_calls): Update.
22531 (recursive_inlining): Update.
22532 (inline_small_functions): Update.
22533 (dump_overall_stats): Update.
22534 (dump_inline_stats): Update.
22535 * ipa-profile.c (ipa_profile_generate_summary): Update.
22536 (ipa_propagate_frequency): Update.
22537 (ipa_profile): Update.
22538 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22539 * ipa-utils.c (ipa_merge_profiles): Update.
22540 * loop-doloop.c (doloop_modify): Update.
22541 * loop-unroll.c (report_unroll): Update.
22542 (unroll_loop_runtime_iterations): Update.
22543 * lto-cgraph.c (lto_output_edge): Update.
22544 (lto_output_node): Update.
22545 (input_node): Update.
22546 (input_edge): Update.
22547 (merge_profile_summaries): Update.
22548 * lto-streamer-in.c (input_cfg): Update.
22549 * lto-streamer-out.c (output_cfg): Update.
22550 * mcf.c (create_fixup_graph): Update.
22551 (adjust_cfg_counts): Update.
22552 (sum_edge_counts): Update.
22553 * modulo-sched.c (sms_schedule): Update.
22554 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
22555 * predict.c (maybe_hot_count_p): Update.
22556 (probably_never_executed): Update.
22557 (dump_prediction): Update.
22558 (combine_predictions_for_bb): Update.
22559 (propagate_freq): Update.
22560 (handle_missing_profiles): Update.
22561 (counts_to_freqs): Update.
22562 (rebuild_frequencies): Update.
22563 (force_edge_cold): Update.
22564 * predict.h: Include profile-count.h
22565 (maybe_hot_count_p, counts_to_freqs): UPdate.
22566 * print-rtl-function.c: Do not include cfg.h
22567 * print-rtl.c: Include basic-block.h
22568 * profile-count.c: New file.
22569 * profile-count.h: New file.
22570 * profile.c (is_edge_inconsistent): Update.
22571 (correct_negative_edge_counts): Update.
22572 (is_inconsistent): Update.
22573 (set_bb_counts): Update.
22574 (read_profile_edge_counts): Update.
22575 (compute_frequency_overlap): Update.
22576 (compute_branch_probabilities): Update; Initialize and deinitialize
22578 (branch_prob): Update.
22579 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
22580 (edge_gcov_count): New.
22581 (bb_gcov_count): New.
22582 * shrink-wrap.c (try_shrink_wrapping): Update.
22583 * tracer.c (better_p): Update.
22584 * trans-mem.c (expand_transaction): Update.
22585 (ipa_tm_insert_irr_call): Update.
22586 (ipa_tm_insert_gettmclone_call): Update.
22587 * tree-call-cdce.c: Update.
22588 * tree-cfg.c (gimple_duplicate_sese_region): Update.
22589 (gimple_duplicate_sese_tail): Update.
22590 (gimple_account_profile_record): Update.
22591 (execute_fixup_cfg): Update.
22592 * tree-inline.c (copy_bb): Update.
22593 (copy_edges_for_bb): Update.
22594 (initialize_cfun): Update.
22595 (freqs_to_counts): Update.
22596 (copy_cfg_body): Update.
22597 (expand_call_inline): Update.
22598 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
22599 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
22600 (try_unroll_loop_completely): Update.
22601 (try_peel_loop): Update.
22602 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
22603 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
22604 * tree-ssa-loop-split.c (connect_loops): Update.
22605 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
22606 * tree-ssa-reassoc.c (branch_fixup): Update.
22607 * tree-ssa-tail-merge.c (replace_block_by): Update.
22608 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
22609 (compute_path_counts): Update.
22610 (update_profile): Update.
22611 (recompute_probabilities): Update.
22612 (update_joiner_offpath_counts): Update.
22613 (estimated_freqs_path): Update.
22614 (freqs_to_counts_path): Update.
22615 (clear_counts_path): Update.
22616 (ssa_fix_duplicate_block_edges): Update.
22617 (duplicate_thread_path): Update.
22618 * tree-switch-conversion.c (case_bit_test_cmp): Update.
22619 (struct switch_conv_info): Update.
22620 * tree-tailcall.c (decrease_profile): Update.
22621 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
22622 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
22623 * value-prof.c (check_counter): Update.
22624 (gimple_divmod_fixed_value): Update.
22625 (gimple_mod_pow2): Update.
22626 (gimple_mod_subtract): Update.
22627 (gimple_ic_transform): Update.
22628 (gimple_stringop_fixed_value): Update.
22629 * value-prof.h (gimple_ic): Update.
22631 2017-06-02 Carl Love <cel@us.ibm.com>
22633 * config/rs6000/rs6000-c: Add support for built-in functions
22634 vector double vec_doublee (vector signed int);
22635 vector double vec_doublee (vector unsigned int);
22636 vector double vec_doublee (vector float);
22637 vector double vec_doubleh (vector signed int);
22638 vector double vec_doubleh (vector unsigned int);
22639 vector double vec_doubleh (vector float);
22640 vector double vec_doublel (vector signed int);
22641 vector double vec_doublel (vector unsigned int);
22642 vector double vec_doublel (vector float);
22643 vector double vec_doubleo (vector signed int);
22644 vector double vec_doubleo (vector unsigned int);
22645 vector double vec_doubleo (vector float);.
22646 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
22647 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
22649 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
22650 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
22651 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
22653 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
22654 vec_doublel, vec_doubleh.
22655 * doc/extend.texi: Update the built-in documentation file for the
22656 new built-in functions.
22658 2017-06-02 David Malcolm <dmalcolm@redhat.com>
22661 * ipa-inline-analysis.c (free_growth_caches): Set
22662 edge_removal_hook_holder to NULL after removing it.
22664 2017-06-02 Sudakshina Das <sudi.das@arm.com>
22666 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
22667 comparision with zero.
22669 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22670 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22671 for early expansion of vec_min and vec_max builtins.
22672 (builtin_function_type): Add min/max unsigned variants to those
22673 identified as having unsigned arguments.
22675 2017-06-02 Olivier Hainque <hainque@adacore.com>
22677 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
22679 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22681 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
22682 Use VALL_F16 iterator rather than VALL.
22684 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22686 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
22687 Emit CBNZ inside loop when doing a strong exchange and comparing
22688 against zero. Generate the CC flags after the loop.
22690 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
22692 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
22693 (dl_section_ref): New.
22694 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
22695 On AIX, append an expression to subtract the size of the
22696 section length to dl_section_ref.
22698 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
22700 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
22701 for early expansion of vector absolute builtins.
22703 2017-06-02 Richard Biener <rguenther@suse.de>
22705 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
22706 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
22708 2017-06-02 Richard Biener <rguenther@suse.de>
22710 PR tree-optimization/80948
22711 * tree-tailcall.c (find_tail_calls): Track stmts to move in
22712 stmt order as well.
22714 2017-06-02 Richard Biener <rguenther@suse.de>
22716 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
22718 * tree-vect-stmts.c (process_use): Do not mark backedge defs
22719 for inductions as relevant.
22721 2017-06-02 Richard Biener <rguenther@suse.de>
22723 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
22724 (vectorizable_induction): ... this. Remove dead code.
22726 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
22728 * builtins. (expand_builtin_alloca): Remove second parameter and
22729 infer its value from the first parameter instead.
22730 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
22732 2017-06-02 Jakub Jelinek <jakub@redhat.com>
22734 PR rtl-optimization/80903
22735 * loop-doloop.c (add_test): Unshare sequence.
22737 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
22739 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
22741 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
22743 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
22745 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
22746 xlogue_layout::get_instance, logue_layout::xlogue_layout,
22747 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
22748 (xlogue_layout::get_stub_rtx): Make static.
22749 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
22750 (xlogue_layout::compute_stub_managed_regs): Rename to...
22751 (xlogue_layout::count_stub_managed_regs): ...this.
22752 (xlogue_layout::is_stub_managed_reg): New function.
22753 (xlogue_layout::m_stub_names): Rename to...
22754 (xlogue_layout::s_stub_names): ...this, make static.
22755 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
22756 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
22757 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
22758 xlogue_layout::s_stub_names): Instantiate statics.
22759 (stub_managed_regs): Remove.
22760 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
22761 (disable_call_ms2sysv_xlogues): Rename to...
22762 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
22763 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
22765 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
22766 change after reload_completed.
22767 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
22769 (ix86_expand_prologue): Likewise.
22770 (ix86_expand_epilogue): Likewise.
22771 (ix86_expand_split_stack_prologue): Likewise.
22772 (ix86_compute_frame_layout): Remove frame parameter ...
22773 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
22774 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
22776 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
22777 (ix86_frame): Move from here ...
22778 * config/i386/i386.h (ix86_frame): ... to here.
22779 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
22780 complete ix86_frame data structure instead. Remove some_ld_name.
22782 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22784 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
22785 symbols that hold a DECL_VALUE_EXPR.
22787 2017-06-01 Martin Jambor <mjambor@suse.cz>
22789 PR tree-optimization/80898
22790 * tree-sra.c (process_subtree_disqualification): Removed.
22791 (disqualify_candidate): Do not acll
22792 process_subtree_disqualification.
22793 (subtree_mark_written_and_enqueue): New function.
22794 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
22795 RHS has been disqualified and re-queue LHS if necessary. Apart
22796 from that, ignore disqualified RHS.
22798 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22800 * config/s390/s390.c (s390_emit_epilogue): Disable early return
22801 address fetch for z10 or later.
22803 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22805 * config/arc/arc.md (tst_movb): Add guard when splitting.
22807 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22809 * config/arc/arc.c (arc_can_eliminate): Test against
22810 arc_frame_pointer_needed.
22812 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22814 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
22815 to prevent store reordering.
22816 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
22817 (type): Add block type.
22818 (stack_tie): Define special instruction to be used in
22821 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22823 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
22824 constraint. It is not valid for the pattern.
22825 (noncommutative_binary_comparison): Likewise.
22827 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22829 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
22832 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22834 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
22835 be used by the reg-alloc.
22837 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22839 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
22840 reg-alloc when having mul64 or mul32x16 instructions.
22841 (mulsidi3): Likewise.
22842 (umulsidi3): Likewise.
22843 (mulsi32x16): New pattern.
22844 (mulsi64): Likewise.
22845 (mulsidi64): Likewise.
22846 (umulsidi64): Likewise.
22847 (MUL32x16_REG): Define.
22848 (mul64_600): Use MUL32x16_REG.
22849 (mac64_600): Likewise.
22850 (umul64_600): Likewise.
22851 (umac64_600): Likewise.
22853 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
22855 * config/arc/arc.md (mulsi3_700): Make it commutative.
22857 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
22859 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
22861 (*sign_extendsidi2_insn): Likewise for movstosw.
22863 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
22865 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
22866 the type of the input discriminant value. Convert the
22867 discriminant value of signedness vary.
22869 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
22871 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
22872 Add to -Wall section.
22874 2017-06-01 Richard Biener <rguenther@suse.de>
22876 PR middle-end/66313
22877 * fold-const.c (fold_plusminus_mult_expr): If the factored
22878 factor may be zero use a wrapping type for the inner operation.
22879 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
22880 and handle moved defs.
22881 (process_assignment): Properly guard the unary op case. Return a
22882 tri-state indicating that moving the stmt before the call may allow
22883 to continue. Pass through to_move.
22884 (find_tail_calls): Handle moving unrelated defs before
22887 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
22890 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
22891 splitter result in the canonical way.
22893 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
22895 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
22896 also for 32bit target. Update insn attributes.
22897 (zero-extendsidi2 splitter): Allow all registers for operand 1.
22899 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
22901 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
22902 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
22903 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
22904 (_mm_maskz_min_ss): New intrinsics.
22906 2017-05-31 Martin Liska <mliska@suse.cz>
22908 * tree-vect-loop.c (vect_create_epilog_for_reduction):
22909 Change comment style to one we normally use.
22910 (vectorizable_reduction): Likewise.
22911 (vectorizable_induction): Likewise.
22912 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
22913 (vectorizable_call): Likewise.
22914 (vectorizable_simd_clone_call): Likewise.
22915 (vectorizable_conversion): Likewise.
22916 (vectorizable_assignment): Likewise.
22917 (vectorizable_shift): Likewise.
22918 (vectorizable_operation): Likewise.
22919 (vectorizable_store): Likewise.
22920 (vectorizable_load): Likewise.
22921 * tree-vectorizer.h: Likewise.
22923 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
22925 * passes.c (emergency_dump_function): New.
22926 * tree-pass.h (emergency_dump_function): Declare.
22927 * plugin.c (plugins_internal_error_function): Remove.
22928 * plugin.h (plugins_internal_error_function): Remove declaration.
22929 * toplev.c (internal_error_function): New static function. Use it...
22930 (general_init): ...here.
22932 2017-05-31 Graham Markall <graham.markall@embecosm.com>
22934 * config/arc/arc.c (arc_print_operand): Handle constant operands.
22935 (arc_rtx_costs): Add costs for new patterns.
22936 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
22937 * config/arc/predicates.md: Add _1_2_3_operand predicate.
22939 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22941 * tree-ssa-strlen.c (get_next_strinfo): New function.
22942 (get_stridx_plus_constant): Use it.
22943 (zero_length_string): Likewise.
22944 (adjust_related_strinfos): Likewise.
22945 (adjust_last_stmt): Likewise.
22947 2017-05-31 Richard Biener <rguenther@suse.de>
22950 * config/i386/i386.c (ix86_expand_builtin): Remove assert
22951 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
22953 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
22955 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
22956 loop_vinfo argument and use of dependence distance vectors.
22957 Check instead whether the two references differ only in their
22958 initial value and assume that they have the same alignment if the
22959 difference is a multiple of the vector alignment.
22960 (vect_analyze_data_refs_alignment): Update call accordingly.
22962 2017-05-31 Martin Liska <mliska@suse.cz>
22965 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
22967 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22969 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
22970 (create_intersect_range_checks): Move from ...
22971 * tree-data-ref.c (create_intersect_range_checks_index)
22972 (create_intersect_range_checks): ... to here.
22973 (create_runtime_alias_checks): New function factored from ...
22974 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
22975 here. Call above function.
22976 * tree-data-ref.h (create_runtime_alias_checks): New function.
22978 2017-05-31 Bin Cheng <bin.cheng@arm.com>
22980 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
22981 segment length for dr_b and compute it in wide_int.
22983 2017-05-31 Richard Biener <rguenther@suse.de>
22985 PR tree-optimization/80906
22986 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
22987 and pass through iv_map.
22988 (copy_bb_and_scalar_dependences): Adjust.
22989 (translate_pending_phi_nodes): Likewise.
22990 (copy_loop_close_phi_args): Handle code-generating IVs instead
22993 2017-05-30 David Malcolm <dmalcolm@redhat.com>
22995 * diagnostic-color.c (color_dict): Add "type-diff".
22996 (parse_gcc_colors): Update comment.
22997 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
22998 -fdiagnostics-show-template-tree and -fno-elide-type.
22999 (GCC_COLORS): Add type-diff to example.
23001 (-fdiagnostics-show-template-tree): New.
23002 (-fno-elide-type): New.
23003 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
23004 the pp_format_decoder callback. Call any m_format_postprocessor's
23006 (pretty_printer::pretty_printer): Initialize
23007 m_format_postprocessor.
23008 (pretty_printer::~pretty_printer): Delete any
23009 m_format_postprocessor.
23010 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
23011 (class format_postprocessor): New class.
23012 (struct pretty_printer::format_decoder): Document the new parameters.
23013 (struct pretty_printer::m_format_postprocessor): New field.
23014 * tree-diagnostic.c (default_tree_printer): Update for new
23015 bool and const char ** params.
23016 * tree-diagnostic.h (default_tree_printer): Likewise.
23018 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
23020 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
23021 (lwa_operand): Delete rs6000_gen_cell_microcode test.
23022 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
23023 rs6000_gen_cell_microcode code.
23024 (rs6000_final_prescan_insn): Delete.
23025 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
23026 "warn-cell-microcode" entries.
23027 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
23028 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
23029 throughout. Change cc_reg_not_micro_cr0_operand to
23030 cc_reg_not_cr0_operand throughout.
23031 (*extendhi<mode>2_noload): Delete.
23032 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
23033 (mwarn-cell-microcode): Delete.
23034 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
23035 -mgen-cell-microcode and -mwarn-cell-microcode.
23037 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
23040 * config/i386/constraints.md (Yd): New constraint.
23042 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
23043 and (?Yd, r) alternatives. Update insn attributes.
23044 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
23045 and (?*Yd, r) alternatives. Update insn attributes.
23046 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
23048 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
23050 * gimplify.c (gimplify_modify_expr): Don't create a
23051 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
23054 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
23056 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
23058 2017-05-30 Richard Biener <rguenther@suse.de>
23060 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
23061 and reduc_def fields.
23062 (STMT_VINFO_REDUC_TYPE): New define.
23063 (STMT_VINFO_REDUC_DEF): Likewise.
23064 (vect_force_simple_reduction): Adjust prototype.
23065 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
23066 (vect_is_simple_reduction): Remove check_reduction argument.
23067 (vect_force_simple_reduction): Adjust and set
23068 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23069 (vectorizable_reduction): Do not re-do reduction analysis
23070 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
23071 * tree-parloops.c (gather_scalar_reductions): Adjust.
23073 2017-05-30 Richard Biener <rguenther@suse.de>
23075 PR middle-end/80901
23076 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
23079 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
23081 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
23082 Introduce unknown_misalignment parameter and remove vf.
23083 (vect_peeling_hash_get_lowest_cost):
23084 Pass unknown_misalignment parameter.
23085 (vect_enhance_data_refs_alignment):
23086 Fix unsupportable data ref treatment.
23088 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23090 * tree-vect-data-refs.c (vect_get_data_access_cost):
23091 Workaround for SLP handling.
23092 (vect_enhance_data_refs_alignment):
23093 Compute costs for doing no peeling at all, compare to the best
23094 peeling costs so far and avoid peeling if cheaper.
23096 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23098 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
23099 Return peeling info and set costs to zero for unlimited cost
23101 (vect_enhance_data_refs_alignment): Also inspect all datarefs
23102 with unknown misalignment. Compute and costs for unknown
23103 misalignment, compare them to the costs for known misalignment
23104 and choose the cheapest for peeling.
23106 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23108 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
23109 (vect_get_peeling_costs_all_drs): Create function.
23110 (vect_peeling_hash_get_lowest_cost):
23111 Use vect_get_peeling_costs_all_drs.
23112 (vect_peeling_supportable): Create function.
23113 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
23115 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
23117 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
23118 DR_HAS_NEGATIVE_STEP.
23119 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
23120 (vect_enhance_data_refs_alignment): Use.
23121 (vect_duplicate_ssa_name_ptr_info): Use.
23122 * tree-vectorizer.h (dr_misalignment): Use.
23123 (known_alignment_for_access_p): Use.
23125 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
23128 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
23130 (has_section_name): New function.
23132 2017-05-30 Martin Liska <mliska@suse.cz>
23135 * auto-profile.c (get_function_decl_from_block): Fix
23138 2017-05-30 Richard Biener <rguenther@suse.de>
23140 PR middle-end/80876
23141 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
23143 2017-05-30 Martin Liska <mliska@suse.cz>
23145 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
23146 * dumpfile.h (struct dump_file_info): Remove ctors.
23148 2017-05-30 Martin Liska <mliska@suse.cz>
23150 * predict.def: Fix GNU coding style.
23152 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23154 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23155 Mark 'to' argument with ATTRIBUTE_UNUSED.
23157 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
23159 * config/xtensa/xtensa.c (xtensa_emit_call): Use
23160 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
23161 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
23164 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
23166 * doc/install.texi (Options specification): Restore entry of
23167 --enable-sjlj-exceptions.
23169 2017-05-27 Michael Eager <eager@eagercon.com>
23172 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
23174 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
23176 * config/microblaze/microblaze.h
23177 (FIXED_REGISTERS): Update in macro.
23178 (CALL_USED_REGISTERS): Update in macro.
23180 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
23182 * doc/install.texi: Add links to macOS binary distributions.
23184 2017-05-27 Jakub Jelinek <jakub@redhat.com>
23188 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23190 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23192 2017-05-26 Martin Liska <mliska@suse.cz>
23194 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
23196 2017-05-26 Martin Liska <mliska@suse.cz>
23198 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
23199 always leading ';; '.
23200 (dump_bb_info): Likewise.
23201 (brief_dump_cfg): Likewise.
23202 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
23203 * dumpfile.c: Remove usage of TDF_VERBOSE.
23204 * dumpfile.h (enum dump_kind): Likewise.
23205 (dump_gimple_bb_header): Do not use TDF_COMMENT.
23206 * print-tree.c (debug_verbose): Remove.
23207 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
23208 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
23209 * tree-diagnostic.c (default_tree_printer): Replace
23210 TDF_DIAGNOSTIC with TDF_SLIM.
23212 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23214 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
23215 in parameter loop, rather than loop_vinfo.
23216 (create_intersect_range_checks): Ditto.
23217 (vect_create_cond_for_alias_checks): Update call to above functions.
23219 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23221 PR tree-optimization/80815
23222 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
23223 for merging runtime alias checks. Handle negative DR_STEPs.
23225 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23227 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
23229 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
23230 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
23231 out code pruning runtime alias checks.
23232 * tree-data-ref.c (prune_runtime_alias_test_list): New function
23233 factored out from above.
23234 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23236 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
23238 (prune_runtime_alias_test_list): New decalaration.
23240 2017-05-26 Bin Cheng <bin.cheng@arm.com>
23242 * tree-vect-data-refs.c (compare_tree): Rename and move ...
23243 * tree-data-ref.c (data_ref_compare_tree): ... to here.
23244 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
23245 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
23246 (operator==, comp_dr_with_seg_len_pair): Ditto.
23247 (vect_prune_runtime_alias_test_list): Ditto.
23249 2017-05-26 Martin Liska <mliska@suse.cz>
23252 * params.def: Bound partial-inlining-entry-probability param.
23254 2017-05-26 Marek Polacek <polacek@redhat.com>
23257 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
23260 2017-05-26 Richard Biener <rguenther@suse.de>
23262 PR tree-optimization/80842
23263 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
23266 2017-05-26 Richard Biener <rguenther@suse.de>
23268 PR tree-optimization/80844
23269 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
23271 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
23273 * doc/md.texi (Machine Constraints): Update x86 family
23274 machine constraints section to match 'config/i386/constraints.md'.
23276 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
23278 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
23280 2017-05-25 Nathan Sidwell <nathan@acm.org>
23282 * doc/invoke.texi (--enable-languages): Update documentation.
23284 2017-05-25 Martin Liska <mliska@suse.cz>
23286 * dumpfile.c: Add TDF_FOLDING.
23287 * dumpfile.h (enum dump_kind): Likewise.
23288 * genmatch.c (dt_simplify::gen_1): Use it.
23290 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23292 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
23294 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23296 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
23297 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
23299 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
23301 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
23302 * match.pd (X == C): Rewrite it here.
23303 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
23304 with_certain_nonzero_bits2): New predicates.
23305 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
23307 2017-05-24 Nathan Sidwell <nathan@acm.org>
23309 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
23312 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
23315 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23317 * config/powerpcspe: New port. Files are copied from the rs6000
23318 port, with "rs6000" in filenames replaced by "powerpcspe".
23320 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
23322 PR rtl-optimization/80754
23323 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
23325 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
23327 * config/sparc/sparc.md (length): Return the correct value for -mflat
23328 sibcalls to match output_sibcall.
23330 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
23334 * config/rs6000/rs6000.c (struct machine_function): Add new field
23336 (rs6000_get_separate_components): Init that field, use it.
23337 (rs6000_components_for_bb): Use the field.
23339 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
23341 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
23343 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
23345 PR middle-end/80823
23346 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
23348 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23351 * config/s390/s390.c (s390_check_qrst_address): Check incoming
23352 address against address_operand predicate.
23353 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
23355 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
23357 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
23358 parameters passed indirectly.
23360 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
23362 * config/i386/i386.md (*movdi_internal): Remove SSE4
23363 alternative 18 (?r, *v). Update insn attributes.
23364 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
23365 Update insn attributes.
23366 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
23367 Update insn attributes.
23368 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
23369 alternative 1 (r, v). Remove isa attribute.
23370 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
23371 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
23372 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
23374 2017-05-23 Tom de Vries <tom@codesourcery.com>
23376 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
23379 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23381 * cgraphunit.c (symbol_table::process_new_functions): Update.
23382 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
23383 (inline_generate_summary): Rename to ...
23384 (ipa_fn_summary_generate): ... this one.
23385 (inline_read_summary): Rename to ...
23386 (ipa_fn_summary_read): ... this one.
23387 (inline_write_summary): Rename to ...
23388 (ipa_fn_summary_write): ... this one.
23389 (inline_free_summary): Rename to ...
23390 (ipa_free_fn_summary): ... this one.
23391 (pass_data_local_fn_summary, pass_local_fn_summary,
23392 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
23393 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
23394 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
23395 make_pass_ipa_fn_summary): New.
23396 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
23397 inline_write_summary, inline_free_summary): Remove.
23398 (ipa_free_fn_summary) : New.
23399 * ipa-inline.c (ipa_inline): Update.
23400 (pass_ipa_inline): Do not generate summaries.
23401 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
23403 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
23404 and add pass_ipa_fn_summary.
23405 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
23407 (make_pass_inline_parameters): Remove.
23409 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
23411 * omp-low.c (struct omp_context): Remove "default_kind" member.
23414 * omp-offload.c (execute_oacc_device_lower): Remove the
23415 parallelism dimensions function attributes for unparallelized
23416 OpenACC kernels constructs.
23418 2017-05-23 Martin Liska <mliska@suse.cz>
23420 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
23422 (cgraph_edge::make_speculative): Likewise.
23423 (cgraph_edge::resolve_speculation): Likewise.
23424 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
23425 (cgraph_node::dump): Likewise.
23426 * cgraph.h: Likewise.
23427 * cgraphunit.c (analyze_functions): Likewise.
23428 (symbol_table::compile): Likewise.
23429 * ipa-cp.c (print_all_lattices): Likewise.
23430 (determine_versionability): Likewise.
23431 (initialize_node_lattices): Likewise.
23432 (ipcp_verify_propagated_values): Likewise.
23433 (estimate_local_effects): Likewise.
23434 (update_profiling_info): Likewise.
23435 (create_specialized_node): Likewise.
23436 (perhaps_add_new_callers): Likewise.
23437 (decide_about_value): Likewise.
23438 (decide_whether_version_node): Likewise.
23439 (identify_dead_nodes): Likewise.
23440 (ipcp_store_bits_results): Likewise.
23441 * ipa-devirt.c (dump_targets): Likewise.
23442 (ipa_devirt): Likewise.
23443 * ipa-icf.c (sem_item::dump): Likewise.
23444 (sem_function::equals): Likewise.
23445 (sem_variable::equals): Likewise.
23446 (sem_item_optimizer::read_section): Likewise.
23447 (sem_item_optimizer::execute): Likewise.
23448 (congruence_class::dump): Likewise.
23449 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
23450 (dump_inline_summary): Likewise.
23451 (estimate_node_size_and_time): Likewise.
23452 (inline_analyze_function): Likewise.
23453 * ipa-inline-transform.c (inline_call): Likewise.
23454 * ipa-inline.c (report_inline_failed_reason): Likewise.
23455 (want_early_inline_function_p): Likewise.
23456 (edge_badness): Likewise.
23457 (update_edge_key): Likewise.
23458 (inline_small_functions): Likewise.
23459 * ipa-profile.c (ipa_profile): Likewise.
23460 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
23461 (ipa_make_edge_direct_to_target): Likewise.
23462 (remove_described_reference): Likewise.
23463 (ipa_impossible_devirt_target): Likewise.
23464 (propagate_controlled_uses): Likewise.
23465 (ipa_print_node_params): Likewise.
23466 (ipcp_transform_function): Likewise.
23467 * ipa-pure-const.c (pure_const_read_summary): Likewise.
23468 (propagate_pure_const): Likewise.
23469 * ipa-reference.c (generate_summary): Likewise.
23470 (read_write_all_from_decl): Likewise.
23471 (propagate): Likewise.
23472 (ipa_reference_read_optimization_summary): Likewise.
23473 * ipa-utils.c (ipa_merge_profiles): Likewise.
23474 * ipa.c (walk_polymorphic_call_targets): Likewise.
23475 (symbol_table::remove_unreachable_nodes): Likewise.
23476 (ipa_single_use): Likewise.
23477 * passes.c (execute_todo): Likewise.
23478 * predict.c (drop_profile): Likewise.
23479 * symtab.c (symtab_node::get_dump_name): New function.
23480 (symtab_node::dump_name): Likewise.
23481 (symtab_node::dump_asm_name): Likewise.
23482 (symtab_node::dump_references): Likewise.
23483 (symtab_node::dump_referring): Likewise.
23484 (symtab_node::dump_base): Likewise.
23485 (symtab_node::debug_symtab): Likewise.
23486 * tree-sra.c (convert_callers_for_node): Likewise.
23487 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
23488 * value-prof.c (init_node_map): Likewise.
23490 2017-05-23 Martin Liska <mliska@suse.cz>
23492 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
23493 and symtab_node::debug_symtab to symbol_table::debug.
23494 * cgraphunit.c (analyze_functions): Use the renamed function.
23495 (symbol_table::compile): Likewise.
23496 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
23497 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
23498 * passes.c (execute_todo): Likewise.
23499 * symtab.c (symbol_table::dump): New function.
23500 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
23502 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23504 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
23505 that nonconst implies exec.
23507 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
23509 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23510 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23511 (inline_edge_summary_vec): Turn into ...
23512 (ipa_call_summaries): ... this one.
23513 (redirect_to_unreachable, edge_set_predicate,
23514 evaluate_properties_for_edge, inline_summary_alloc,
23515 reset_ipa_call_summary, reset_inline_summary,
23516 inline_summary_t::duplicate): Update.
23517 (inline_edge_duplication_hook): Turn to ...
23518 (ipa_call_summary_t::duplicate): ... this one.
23519 (inline_edge_removal_hook): Turn to ...
23520 (ipa_call_summary_t::remove): ... this one.
23521 (dump_inline_edge_summary): Turn to ...
23522 (dump_ipa_call_summary): ... this one.
23523 (estimate_function_body_sizes): Update.
23524 (inline_update_callee_summaries): Update.
23525 (remap_edge_change_prob): Update.
23526 (remap_edge_summaries): Update.
23527 (inline_merge_summary): Update.
23528 (do_estimate_edge_time): Update.
23529 (inline_generate_summary): Update.
23530 (inline_read_section): Update.
23531 (inline_read_summary): Update.
23532 (inline_free_summary): Update.
23533 * ipa-inline.c (can_inline_edge_p): Update.
23534 (compute_inlined_call_time): Update.
23535 (want_inline_small_function_p): Update.
23536 (edge_badness): Update.
23537 (early_inliner): Update.
23538 * ipa-inline.h (inline_edge_summary): Turn to ...
23539 (ipa_call_summary): ... this one.
23540 (ipa_call_summary_t): New class.
23541 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23542 (ipa_call_summaries): New.
23543 (inline_edge_summary): Remove.
23544 (estimate_edge_growth): Update.
23545 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23546 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23547 * ipa-split.c (execute_split_functions): Update.
23548 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23550 2017-05-23 Tom de Vries <tom@codesourcery.com>
23552 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23553 attributes): Document rdrand effective target.
23555 2017-05-23 Tom de Vries <tom@codesourcery.com>
23557 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
23558 attributes): Sort alphabetically.
23560 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
23562 * config/avr/genmultilib.awk: Use gsub instead of gensub.
23564 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
23567 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
23568 V2DF/V2DI splat into two separate patterns, one that handles
23569 registers, and the other that only handles memory. Drop support
23570 for splatting from a GPR on ISA 2.07 and then splitting the
23571 splat into direct move and splat.
23572 (vsx_splat_<mode>_reg): Likewise.
23573 (vsx_splat_<mode>_mem): Likewise.
23575 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
23577 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
23579 2017-05-22 Jakub Jelinek <jakub@redhat.com>
23581 PR middle-end/80809
23582 * omp-low.c (finish_taskreg_remap): New function.
23583 (finish_taskreg_scan): If unit size of ctx->record_type
23584 is non-constant, unshare the size expression and replace
23585 decls in it with possible outer var refs.
23587 PR middle-end/80809
23588 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
23589 GOVD_SHARED rather than GOVD_PRIVATE with it.
23590 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
23591 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
23593 PR middle-end/80853
23594 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
23595 as last argument to build_outer_var_ref for pointer bases of array
23596 section reductions.
23598 2017-05-19 Martin Sebor <msebor@redhat.com>
23600 * print-tree.c (print_node): Print DECL_READ_P flag.
23602 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23604 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
23605 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
23606 * cgraph.c: Likewise.
23607 * cgraphunit.c: Likewise.
23608 * gengtype.c: Likewise.
23609 * ipa-cp.c: Likewise.
23610 * ipa-devirt.c: Likewise.
23611 * ipa-icf.c: Likewise.
23612 * ipa-predicate.c: Likewise.
23613 * ipa-profile.c: Likewise.
23614 * ipa-prop.c: Likewise.
23615 * ipa-split.c: Likewise.
23617 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
23618 edge_predicate_pool, dump_inline_hints,
23619 inline_summary::account_size_time, redirect_to_unreachable,
23620 edge_set_predicate, set_hint_predicate,
23621 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
23622 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
23623 inline_summary_t::remove, remap_hint_predicate_after_duplication,
23624 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
23625 ipa_call_summary_t::remove, initialize_growth_caches,
23626 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
23627 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
23628 mark_modified, unmodified_parm_1, unmodified_parm,
23629 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
23630 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
23631 compute_bb_predicates, will_be_nonconstant_expr_predicate,
23632 will_be_nonconstant_predicate, record_modified_bb_info,
23633 get_minimal_bb, record_modified, param_change_prob,
23634 phi_result_unknown_predicate, predicate_for_phi_result,
23635 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
23636 estimate_function_body_sizes, compute_inline_parameters,
23637 compute_inline_parameters_for_curren, pass_data_inline_parameters,
23638 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
23639 inline_update_callee_summaries, remap_edge_change_prob,
23640 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
23641 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
23642 inline_analyze_function, inline_summary_t::insert,
23643 inline_generate_summary, read_ipa_call_summary, inline_read_section,
23644 inline_read_summary, write_ipa_call_summary, inline_write_summary,
23645 inline_free_summary): Move to ipa-fnsummary.h
23646 (predicate_t): Remove.
23647 * ipa-fnsummary.c: New file.
23648 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
23649 (enum inline_hints_vals, inline_hints, agg_position_info,
23650 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
23651 inline_summaries, ipa_call_summary, ipa_call_summary_t,
23652 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
23653 dump_inline_summary, dump_inline_hints, inline_generate_summary,
23654 inline_read_summary, inline_write_summary, inline_free_summary,
23655 inline_analyze_function, initialize_inline_failed,
23656 inline_merge_summary, inline_update_overall_summary,
23657 compute_inline_parameters): Move to ipa-fnsummary.h
23658 * ipa-fnsummary.h: New file.
23659 * ipa-inline-transform.h: Include ipa-inline.h.
23660 * ipa-inline.c: LIkewise.
23662 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23664 * ipa-inline.c (edge_badness): Use inlined_time instead of
23665 inline_summaries->get.
23667 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23669 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
23671 2017-05-22 Nathan Sidwell <nathan@acm.org>
23673 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
23674 (fdump-lang): Document 'raw' option.
23675 * dumpfile.h (TDI_tu): Delete.
23676 * dumpfile.c (dump_files): Remove translation-unit.
23677 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
23679 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
23681 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
23682 command option from $(AWK) call.
23683 * config/avr/genmultilib.awk: Simplify and rewrite so that it
23684 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
23685 [FORMAT]: Remove handling of variable.
23686 * config/avr/t-multilib: Regenerate.
23688 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23690 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
23692 (dump_inline_summary): Do not print self_time.
23693 (estimate_function_body_sizes): Do not set self_time.
23694 (compute_inline_parameters): Likewise.
23695 (inline_read_section, inline_write_summary): Do not stream self_time.
23696 * ipa-inline.h (inline_summary): Drop self_time.
23698 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
23700 * ipa-inline-analysis.c (account_size_time): Rename to ...
23701 (inline_summary::account_size_time): ... this one.
23702 (reset_ipa_call_summary): Turn to ...
23703 (ipa_call_summary::reset): ... this one.
23704 (reset_inline_summary): Turn to ...
23705 (inline_summary::reset): ... this one.
23706 (inline_summary_t::remove): Update.
23707 (inline_summary_t::duplicate): Update.
23708 (ipa_call_summary_t::remove): Update.
23709 (dump_inline_summary): Update.
23710 (estimate_function_body_sizes): Update.
23711 (compute_inline_parameters): Update.
23712 (estimate_node_size_and_time): Update.
23713 (inline_merge_summary): Update.
23714 (inline_update_overall_summary): Update.
23715 (inline_read_section): Update.
23716 (inline_write_summary): Update.
23717 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
23718 add account_size_time and reset member functions.
23719 (ipa_call_summary): Add reset function.
23720 * ipa-predicate.h (predicate::operator &): Constify.
23722 2017-05-22 Richard Biener <rguenther@suse.de>
23724 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
23726 2017-05-19 Jason Merrill <jason@redhat.com>
23728 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
23730 2017-05-19 Marek Polacek <polacek@redhat.com>
23733 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
23734 TYPE_OVERFLOW_WRAPS checks.
23736 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
23738 * tree-core.h (enum omp_clause_default_kind): Add
23739 "OMP_CLAUSE_DEFAULT_PRESENT".
23740 * tree-pretty-print.c (dump_omp_clause): Handle it.
23741 * gimplify.c (enum gimplify_omp_var_data): Add
23742 "GOVD_MAP_FORCE_PRESENT".
23743 (gimplify_adjust_omp_clauses_1): Map it to
23744 "GOMP_MAP_FORCE_PRESENT".
23745 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
23747 * gimplify.c (oacc_default_clause): Clarify.
23749 2017-05-19 Nathan Sidwell <nathan@acm.org>
23751 LANG_HOOK_REGISTER_DUMPS
23752 * toplev.c (general_init): Call register dump lang hook.
23753 * doc/invoke.texi: Document -fdump-lang option family.
23754 * dumpfile.c (dump_files): Remove class dump here.
23755 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
23756 * dumpfile.h (tree_dump_index): Remove TDI_class.
23757 * langhooks-def.h (lhd_register_dumps): Declare.
23758 (LANG_HOOKS_REGISTER_DUMPS): Define.
23759 (LANG_HOOKS_INITIALIZER): Add it.
23760 * langhooks.c (lhd_register_dumps): Define.
23761 * langhooks.h (struct lang_hooks): Add register_dumps.
23763 2017-05-19 Nathan Sidwell <nathan@acm.org>
23765 * context.h (context::set_passes): New.
23766 * context.c (context::context): Do not create pass manager.
23767 * toplev.c (general_init): Create pass manager here.
23769 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
23771 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
23772 use this splitter if two add or or instructions would also work for
23773 the constant we want to generate.
23775 2017-05-19 Richard Biener <rguenther@suse.de>
23778 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
23779 predicate evaluation.
23781 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23783 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
23785 * ipa-inline.c (want_inline_small_function_p): Do not cast to
23788 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
23790 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
23791 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
23792 (inline_edge_summary_vec): Turn into ...
23793 (ipa_call_summaries): ... this one.
23794 (redirect_to_unreachable, edge_set_predicate,
23795 evaluate_properties_for_edge, inline_summary_alloc,
23796 reset_ipa_call_summary, reset_inline_summary,
23797 inline_summary_t::duplicate): Update.
23798 (inline_edge_duplication_hook): Turn to ...
23799 (ipa_call_summary_t::duplicate): ... this one.
23800 (inline_edge_removal_hook): Turn to ...
23801 (ipa_call_summary_t::remove): ... this one.
23802 (dump_inline_edge_summary): Turn to ...
23803 (dump_ipa_call_summary): ... this one.
23804 (estimate_function_body_sizes): Update.
23805 (inline_update_callee_summaries): Update.
23806 (remap_edge_change_prob): Update.
23807 (remap_edge_summaries): Update.
23808 (inline_merge_summary): Update.
23809 (do_estimate_edge_time): Update.
23810 (inline_generate_summary): Update.
23811 (inline_read_section): Update.
23812 (inline_read_summary): Update.
23813 (inline_free_summary): Update.
23814 * ipa-inline.c (can_inline_edge_p): Update.
23815 (compute_inlined_call_time): Update.
23816 (want_inline_small_function_p): Update.
23817 (edge_badness): Update.
23818 (early_inliner): Update.
23819 * ipa-inline.h (inline_edge_summary): Turn to ...
23820 (ipa_call_summary): ... this one.
23821 (ipa_call_summary_t): New class.
23822 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
23823 (ipa_call_summaries): New.
23824 (inline_edge_summary): Remove.
23825 (estimate_edge_growth): Update.
23826 * ipa-profile.c (ipa_propagate_frequency_1): Update.
23827 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
23828 * ipa-split.c (execute_split_functions): Update.
23829 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
23831 2017-05-19 Richard Biener <rguenther@suse.de>
23833 PR middle-end/80764
23834 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
23836 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
23838 * config/rs6000/rs6000.c (struct machine_function): Add field
23839 fpr_is_wrapped_separately.
23840 (rs6000_get_separate_components): Use 64 components. Handle the
23841 new FPR components.
23842 (rs6000_components_for_bb): Handle the FPR components.
23843 (rs6000_emit_prologue_components): Handle the FPR components.
23844 (rs6000_emit_epilogue_components): Handle the FPR components.
23845 (rs6000_set_handled_components): Handle the FPR components.
23846 (rs6000_emit_prologue): Don't output prologue code for those FPRs
23847 that are already separately shrink-wrapped.
23848 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
23849 that are already separately shrink-wrapped.
23851 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
23854 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
23857 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
23858 (define_peephole2 for Altivec d-form load): Add peepholes to catch
23859 cases where the register allocator uses a move and an offsettable
23860 memory operation to/from a FPR register on ISA 2.06/2.07.
23861 (define_peephole2 for Altivec d-form store): Likewise.
23863 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
23866 * config/i386/mmx.md (*mov<mode>_internal): Enable
23867 alternatives 11, 12, 13 and 14 also for 32bit targets.
23868 Remove alternatives 15, 16, 17 and 18.
23869 * config/i386/sse.md (vec_concatv2di): Change
23870 alternative (!x, *y) to (x, ?!*Yn).
23872 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
23874 * dumpfile.h (enum dump_kind): Remove stray comma.
23876 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23878 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
23879 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
23880 predicate::num_conditions
23881 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
23882 (CHANGED): turn into predicate::changed.
23883 (agg_position_info): Move to ipa-predicate.h
23884 (add_condition, predicate::add_clause, predicate::operator &=,
23885 predicate::or_with, predicate::evaluate, predicate::probability,
23886 dump_condition, dump_clause, predicate::dump,
23887 predicate::remap_after_duplication, predicate::remap_after_inlining,
23888 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
23889 (evaluate_conditions_for_known_args): Update.
23890 (set_cond_stmt_execution_predicate): Update.
23891 * ipa-inline.h: Include ipa-predicate.h
23892 (condition, inline_param_summary, conditions, agg_position_info,
23893 predicate): Move to ipa-predicate.h
23894 * ipa-predicate.c: New file.
23895 * ipa-predicate.h: New file.
23897 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
23899 * final.c (leaf_function_p): Check we are not in a sequence.
23901 2017-05-18 Martin Liska <mliska@suse.cz>
23903 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
23904 * dumpfile.c (dump_register): Use new enum dump_kind.
23905 (get_dump_file_name): Likewise.
23906 (dump_enable_all): Likewise.
23907 (dump_switch_p_1): Likewise.
23908 (enable_rtl_dump_file): Remove usage of TDF_RTL.
23909 * dumpfile.h (enum dump_kind): New enum type.
23910 (struct dump_file_info): Create constructor and
23911 format fields and comments.
23912 * passes.c (pass_manager::register_one_dump_file):
23914 * statistics.c (statistics_early_init): Likewise.
23915 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
23916 TDF_TREE with TDF_SLIM.
23917 (gather_memory_references_ref): Likewise.
23919 2017-05-18 Martin Liska <mliska@suse.cz>
23921 * vec.h (struct vnull): Use it.
23923 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
23925 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
23926 (true_predicate, false_predicate, true_predicate_p,
23927 false_predicate_p): Remove.
23928 (single_cond_predicate, not_inlined_predicate): Turn to member function
23930 (add_condition): Update.
23931 (add_clause): Turn to...
23932 (predicate::add_clause): ... this one; update; allow passing NULL
23934 (and_predicates): Turn to ...
23935 (predicate::operator &=): ... this one.
23936 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
23937 (or_predicates): Turn to ...
23938 (predicate::or_with): ... this one.
23939 (evaluate_predicate): Turn to ...
23940 (predicate::evaluate): ... this one.
23941 (predicate_probability): Turn to ...
23942 (predicate::probability): ... this one.
23943 (dump_condition): Update.
23944 (dump_predicate): Turn to ...
23945 (predicate::dump): ... this one.
23946 (account_size_time): Update.
23947 (edge_set_predicate): Update.
23948 (set_hint_predicate): UPdate.
23949 (evaluate_conditions_for_known_args): Update.
23950 (evaluate_properties_for_edge): Update.
23951 (remap_predicate_after_duplication): Turn to...
23952 (predicate::remap_after_duplication): ... this one.
23953 (remap_hint_predicate_after_duplication): Update.
23954 (inline_summary_t::duplicate): UPdate.
23955 (dump_inline_edge_summary): Update.
23956 (dump_inline_summary): Update.
23957 (set_cond_stmt_execution_predicate): Update.
23958 (set_switch_stmt_execution_predicate): Update.
23959 (compute_bb_predicates): Update.
23960 (will_be_nonconstant_expr_predicate): Update.
23961 (will_be_nonconstant_predicate): Update.
23962 (phi_result_unknown_predicate): Update.
23963 (predicate_for_phi_result): Update.
23964 (array_index_predicate): Update.
23965 (estimate_function_body_sizes): Update.
23966 (estimate_node_size_and_time): Update.
23967 (estimate_ipcp_clone_size_and_time): Update.
23968 (remap_predicate): Rename to ...
23969 (predicate::remap_after_inlining): ... this one.
23970 (remap_hint_predicate): Update.
23971 (inline_merge_summary): Update.
23972 (inline_update_overall_summary): Update.
23973 (estimate_size_after_inlining): Update.
23974 (read_predicate): Rename to ...
23975 (predicate::stream_in): ... this one.
23976 (read_inline_edge_summary): Update.
23977 (write_predicate): Rename to ...
23978 (predicate::stream_out): ... this one.
23979 (write_inline_edge_summary): Update.
23980 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
23981 (clause_t): Turn to uint32_t
23982 (predicate): Turn to class; implement constructor and operators
23984 (size_time_entry): Update.
23985 (inline_summary): Update.
23986 (inline_edge_summary): Update.
23988 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
23990 * fold-const.c (fold_binary_loc): Move transformation...
23991 * match.pd (C - X CMP X): ... here.
23993 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
23995 * config/sparc/sparc.c (sparc_option_override): Set function
23996 alignment for -mcpu=niagara7 to 64 to match the I$ line.
23997 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
23999 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
24001 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
24003 2017-05-18 Marek Polacek <polacek@redhat.com>
24006 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
24007 (pass_ubsan::execute): Call gimple_assign_single_p instead of
24008 gimple_assign_load_p.
24010 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
24012 PR middle-end/80692
24013 * real.c (do_compare): Give decimal_do_compare preference over
24014 comparing just the signs.
24016 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
24018 * doc/md.texi (Canonicalization of Instructions): Describe the
24019 canonical form of instructions that inherently set a condition
24022 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
24024 PR middle-end/80775
24025 * tree-cfg.c: Move deletion of unreachable case statements to after
24026 the merging of consecutive case labels.
24028 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
24030 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
24031 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
24032 restoring of callee-saved registers.
24034 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
24036 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
24037 * config/visium/visium.c (single_set_and_flags): Likewise.
24038 * config/visium/visium.md (Substitutions): Likewise.
24040 2017-05-17 Martin Liska <mliska@suse.cz>
24042 * cfg.c: Introduce dump_flags_t type and
24043 use it instead of int type.
24045 * cfghooks.c: Likewise.
24046 * cfghooks.h (struct cfg_hooks): Likewise.
24047 * cfgrtl.c: Likewise.
24048 * cfgrtl.h: Likewise.
24049 * cgraph.c (cgraph_node::get_body): Likewise.
24050 * coretypes.h: Likewise.
24051 * domwalk.c: Likewise.
24052 * domwalk.h: Likewise.
24053 * dumpfile.c (struct dump_option_value_info): Likewise.
24054 (dump_enable_all): Likewise.
24055 (dump_switch_p_1): Likewise.
24056 (opt_info_switch_p): Likewise.
24057 * dumpfile.h (enum tree_dump_index): Likewise.
24058 (struct dump_file_info): Likewise.
24059 * genemit.c: Likewise.
24060 * generic-match-head.c: Likewise.
24061 * gengtype.c (open_base_files): Likewise.
24062 * gimple-pretty-print.c: Likewise.
24063 * gimple-pretty-print.h: Likewise.
24064 * graph.c (print_graph_cfg): Likewise.
24065 * graphite-scop-detection.c (dot_all_sese): Likewise.
24066 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
24067 * loop-unroll.c (report_unroll): Likewise.
24068 * passes.c (pass_manager::register_one_dump_file): Likewise.
24069 * print-tree.c: Likewise.
24070 * statistics.c: Likewise.
24071 * tree-cfg.c: Likewise.
24072 * tree-cfg.h: Likewise.
24073 * tree-dfa.c: Likewise.
24074 * tree-dfa.h: Likewise.
24075 * tree-dump.c (dump_function): Likewise.
24076 * tree-dump.h (struct dump_info): Likewise.
24077 * tree-pretty-print.c: Likewise.
24078 * tree-pretty-print.h: Likewise.
24079 * tree-ssa-live.c: Likewise.
24080 * tree-ssa-live.h: Likewise.
24081 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
24082 * tree-vect-loop.c: Likewise.
24083 * tree-vect-slp.c: Likewise.
24085 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
24086 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24088 PR tree-optimization/80457
24089 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
24090 of all arguments to a statement as scalar_to_vec operations.
24091 (vectorizable_call): Adjust call to vect_model_simple_cost for
24093 (vectorizable_conversion): Likewise.
24094 (vectorizable_assignment): Likewise.
24095 (vectorizable_shift): Likewise.
24096 (vectorizable_operation): Likewise.
24097 (vectorizable_comparison): Likewise.
24098 (vect_is_simple_cond): Record the def types for operands.
24099 (vectorizable_condition): Likewise, call vect_model_simple_cost.
24100 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
24101 for statement argument count.
24103 2017-05-16 Carl Love <cel@us.ibm.com>
24105 * config/rs6000/rs6000-c: Add support for built-in functions
24106 vector unsigned long long vec_bperm (vector unsigned long long,
24107 vector unsigned char)
24108 vector signed long long vec_mule (vector signed int,
24110 vector unsigned long long vec_mule (vector unsigned int,
24111 vector unsigned int)
24112 vector signed long long vec_mulo (vector signed int,
24114 vector unsigned long long vec_mulo (vector unsigned int,
24115 vector unsigned int)
24116 vector signed char vec_sldw (vector signed char,
24117 vector signed char,
24119 vector unsigned char vec_sldw (vector unsigned char,
24120 vector unsigned char,
24122 vector signed short vec_sldw (vector signed short,
24123 vector signed short,
24125 vector unsigned short vec_sldw (vector unsigned short,
24126 vector unsigned short,
24128 vector signed int vec_sldw (vector signed int,
24131 vector unsigned int vec_sldw (vector unsigned int,
24132 vector unsigned int,
24134 vector signed long long vec_sldw (vector signed long long,
24135 vector signed long long,
24137 vector unsigned long long vec_sldw (vector unsigned long long,
24138 vector unsigned long long,
24140 * config/rs6000/rs6000-c: Add support for built-in functions
24141 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
24142 * config/rs6000/altivec.h: Add defintion for vec_sldw.
24143 * doc/extend.texi: Update the built-in documentation for the
24144 new built-in functions.
24146 2017-05-16 Marek Polacek <polacek@redhat.com>
24150 * tree.c (save_expr): Don't fold the expression.
24152 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
24154 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
24155 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
24156 and (?*y,m). Update insn attributes.
24158 2017-05-16 Martin Liska <mliska@suse.cz>
24160 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
24161 flags argument of print_gimple_stmt, print_gimple_expr,
24162 print_generic_stmt and print_generic_expr.
24163 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
24164 * coretypes.h: Likewise.
24165 * except.c (dump_eh_tree): Likewise.
24166 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
24167 * gimple-pretty-print.h: Likewise.
24168 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
24169 (backprop::push_to_worklist): Likewise.
24170 (backprop::pop_from_worklist): Likewise.
24171 (backprop::process_use): Likewise.
24172 (backprop::intersect_uses): Likewise.
24173 (note_replacement): Likewise.
24174 * gimple-ssa-store-merging.c
24175 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
24176 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
24177 (pass_store_merging::execute): Likewise.
24178 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
24179 (ssa_base_cand_dump_callback): Likewise.
24180 (dump_incr_vec): Likewise.
24181 (replace_refs): Likewise.
24182 (replace_mult_candidate): Likewise.
24183 (create_add_on_incoming_edge): Likewise.
24184 (create_phi_basis): Likewise.
24185 (insert_initializers): Likewise.
24186 (all_phi_incrs_profitable): Likewise.
24187 (introduce_cast_before_cand): Likewise.
24188 (replace_one_candidate): Likewise.
24189 * gimplify.c (gimplify_expr): Likewise.
24190 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
24191 (set_rename): Likewise.
24192 (rename_uses): Likewise.
24193 (copy_loop_phi_nodes): Likewise.
24194 (add_close_phis_to_merge_points): Likewise.
24195 (copy_loop_close_phi_args): Likewise.
24196 (copy_cond_phi_args): Likewise.
24197 (graphite_copy_stmts_from_block): Likewise.
24198 (translate_pending_phi_nodes): Likewise.
24199 * graphite-poly.c (print_pdr): Likewise.
24200 (dump_gbb_cases): Likewise.
24201 (dump_gbb_conditions): Likewise.
24202 (print_scop_params): Likewise.
24203 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
24204 (build_cross_bb_scalars_use): Likewise.
24205 (gather_bbs::before_dom_children): Likewise.
24206 * hsa-dump.c (dump_hsa_immed): Likewise.
24207 * ipa-cp.c (print_ipcp_constant_value): Likewise.
24208 (get_replacement_map): Likewise.
24209 * ipa-inline-analysis.c (dump_condition): Likewise.
24210 (estimate_function_body_sizes): Likewise.
24211 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
24212 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
24213 * ipa-prop.c (ipa_dump_param): Likewise.
24214 (ipa_print_node_jump_functions_for_edge): Likewise.
24215 (ipa_modify_call_arguments): Likewise.
24216 (ipa_modify_expr): Likewise.
24217 (ipa_dump_param_adjustments): Likewise.
24218 (ipa_dump_agg_replacement_values): Likewise.
24219 (ipcp_modif_dom_walker::before_dom_children): Likewise.
24220 * ipa-pure-const.c (check_stmt): Likewise.
24221 (pass_nothrow::execute): Likewise.
24222 * ipa-split.c (execute_split_functions): Likewise.
24223 * omp-offload.c (dump_oacc_loop_part): Likewise.
24224 (dump_oacc_loop): Likewise.
24225 * trans-mem.c (tm_log_emit): Likewise.
24226 (tm_memopt_accumulate_memops): Likewise.
24227 (dump_tm_memopt_set): Likewise.
24228 (dump_tm_memopt_transform): Likewise.
24229 * tree-cfg.c (gimple_verify_flow_info): Likewise.
24230 (print_loop): Likewise.
24231 * tree-chkp-opt.c (chkp_print_addr): Likewise.
24232 (chkp_gather_checks_info): Likewise.
24233 (chkp_get_check_result): Likewise.
24234 (chkp_remove_check_if_pass): Likewise.
24235 (chkp_use_outer_bounds_if_possible): Likewise.
24236 (chkp_reduce_bounds_lifetime): Likewise.
24237 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
24238 (chkp_mark_completed_bounds): Likewise.
24239 (chkp_register_incomplete_bounds): Likewise.
24240 (chkp_mark_invalid_bounds): Likewise.
24241 (chkp_maybe_copy_and_register_bounds): Likewise.
24242 (chkp_build_returned_bound): Likewise.
24243 (chkp_get_bound_for_parm): Likewise.
24244 (chkp_build_bndldx): Likewise.
24245 (chkp_get_bounds_by_definition): Likewise.
24246 (chkp_generate_extern_var_bounds): Likewise.
24247 (chkp_get_bounds_for_decl_addr): Likewise.
24248 * tree-chrec.c (chrec_apply): Likewise.
24249 * tree-data-ref.c (dump_data_reference): Likewise.
24250 (dump_subscript): Likewise.
24251 (dump_data_dependence_relation): Likewise.
24252 (analyze_overlapping_iterations): Likewise.
24253 * tree-inline.c (expand_call_inline): Likewise.
24254 (tree_function_versioning): Likewise.
24255 * tree-into-ssa.c (dump_defs_stack): Likewise.
24256 (dump_currdefs): Likewise.
24257 (dump_names_replaced_by): Likewise.
24258 (dump_update_ssa): Likewise.
24259 (update_ssa): Likewise.
24260 * tree-object-size.c (pass_object_sizes::execute): Likewise.
24261 * tree-parloops.c (build_new_reduction): Likewise.
24262 (try_create_reduction_list): Likewise.
24263 (ref_conflicts_with_region): Likewise.
24264 (oacc_entry_exit_ok_1): Likewise.
24265 (oacc_entry_exit_single_gang): Likewise.
24266 * tree-pretty-print.h: Likewise.
24267 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
24268 (get_scalar_evolution): Likewise.
24269 (add_to_evolution): Likewise.
24270 (get_loop_exit_condition): Likewise.
24271 (analyze_evolution_in_loop): Likewise.
24272 (analyze_initial_condition): Likewise.
24273 (analyze_scalar_evolution): Likewise.
24274 (instantiate_scev): Likewise.
24275 (number_of_latch_executions): Likewise.
24276 (gather_chrec_stats): Likewise.
24277 (final_value_replacement_loop): Likewise.
24278 (scev_const_prop): Likewise.
24279 * tree-sra.c (dump_access): Likewise.
24280 (disqualify_candidate): Likewise.
24281 (create_access): Likewise.
24282 (reject): Likewise.
24283 (maybe_add_sra_candidate): Likewise.
24284 (create_access_replacement): Likewise.
24285 (analyze_access_subtree): Likewise.
24286 (analyze_all_variable_accesses): Likewise.
24287 (sra_modify_assign): Likewise.
24288 (initialize_constant_pool_replacements): Likewise.
24289 (find_param_candidates): Likewise.
24290 (decide_one_param_reduction): Likewise.
24291 (replace_removed_params_ssa_names): Likewise.
24292 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
24293 * tree-ssa-copy.c (dump_copy_of): Likewise.
24294 (copy_prop_visit_cond_stmt): Likewise.
24295 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
24296 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
24297 (record_equivalences_from_stmt): Likewise.
24298 * tree-ssa-dse.c (compute_trims): Likewise.
24299 (delete_dead_call): Likewise.
24300 (delete_dead_assignment): Likewise.
24301 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
24302 (forward_propagate_into_cond): Likewise.
24303 (pass_forwprop::execute): Likewise.
24304 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
24305 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
24307 (move_computations_worker): Likewise.
24308 (execute_sm): Likewise.
24309 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
24310 (remove_exits_and_undefined_stmts): Likewise.
24311 (remove_redundant_iv_tests): Likewise.
24312 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
24313 (adjust_iv_update_pos): Likewise.
24314 * tree-ssa-math-opts.c (bswap_replace): Likewise.
24315 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
24316 (value_replacement): Likewise.
24317 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
24318 * tree-ssa-pre.c (print_pre_expr): Likewise.
24319 (get_representative_for): Likewise.
24320 (create_expression_by_pieces): Likewise.
24321 (insert_into_preds_of_block): Likewise.
24322 (eliminate_insert): Likewise.
24323 (eliminate_dom_walker::before_dom_children): Likewise.
24324 (eliminate): Likewise.
24325 (remove_dead_inserted_code): Likewise.
24326 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
24327 * tree-ssa-reassoc.c (get_rank): Likewise.
24328 (eliminate_duplicate_pair): Likewise.
24329 (eliminate_plus_minus_pair): Likewise.
24330 (eliminate_not_pairs): Likewise.
24331 (undistribute_ops_list): Likewise.
24332 (eliminate_redundant_comparison): Likewise.
24333 (update_range_test): Likewise.
24334 (optimize_range_tests_var_bound): Likewise.
24335 (optimize_vec_cond_expr): Likewise.
24336 (rewrite_expr_tree): Likewise.
24337 (rewrite_expr_tree_parallel): Likewise.
24338 (linearize_expr): Likewise.
24339 (break_up_subtract): Likewise.
24340 (linearize_expr_tree): Likewise.
24341 (attempt_builtin_powi): Likewise.
24342 (attempt_builtin_copysign): Likewise.
24343 (transform_stmt_to_copy): Likewise.
24344 (transform_stmt_to_multiply): Likewise.
24345 (dump_ops_vector): Likewise.
24346 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
24347 (print_scc): Likewise.
24348 (set_ssa_val_to): Likewise.
24349 (visit_reference_op_store): Likewise.
24350 (visit_use): Likewise.
24351 (sccvn_dom_walker::before_dom_children): Likewise.
24352 (run_scc_vn): Likewise.
24353 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
24355 (expr_hash_elt::print): Likewise.
24356 (const_and_copies::pop_to_marker): Likewise.
24357 (const_and_copies::record_const_or_copy_raw): Likewise.
24358 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
24359 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
24360 (dump_predicates): Likewise.
24361 (find_uninit_use): Likewise.
24362 (warn_uninitialized_phi): Likewise.
24363 (pass_late_warn_uninitialized::execute): Likewise.
24364 * tree-ssa.c (verify_vssa): Likewise.
24365 (verify_ssa): Likewise.
24366 (maybe_optimize_var): Likewise.
24367 * tree-vrp.c (dump_value_range): Likewise.
24368 (dump_all_value_ranges): Likewise.
24369 (dump_asserts_for): Likewise.
24370 (register_edge_assert_for_2): Likewise.
24371 (vrp_visit_cond_stmt): Likewise.
24372 (vrp_visit_switch_stmt): Likewise.
24373 (vrp_visit_stmt): Likewise.
24374 (vrp_visit_phi_node): Likewise.
24375 (simplify_cond_using_ranges_1): Likewise.
24376 (fold_predicate_in): Likewise.
24377 (evrp_dom_walker::before_dom_children): Likewise.
24378 (evrp_dom_walker::push_value_range): Likewise.
24379 (evrp_dom_walker::pop_value_range): Likewise.
24380 (execute_early_vrp): Likewise.
24382 2017-05-16 Richard Biener <rguenther@suse.de>
24384 * dwarf2out.c (loc_list_from_tree_1): Do not create
24385 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
24387 2017-05-16 Richard Biener <rguenther@suse.de>
24389 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
24391 (note_variable_value_in_expr): If we resolved the decl ref
24392 do not push to the stack.
24394 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
24396 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
24397 operations in fast-math mode.
24398 (vaddq_f16): Likewise.
24399 (vmul_f16): Likewise.
24400 (vmulq_f16): Likewise.
24401 (vsub_f16): Likewise.
24402 (vsubq_f16): Likewise.
24403 * config/arm/neon.md (add<mode>3): New.
24405 (fma:<VH:mode>3): New. Also remove outdated comment.
24408 2017-05-16 Martin Liska <mliska@suse.cz>
24412 * ipa-devirt.c (warn_types_mismatch): Fix typo.
24413 (odr_types_equivalent_p): Likewise.
24415 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
24417 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
24419 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
24422 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
24423 non-interunit SSE move alternatives with '?'.
24424 (zero-extendsidi peephole2): New peephole to skip intermediate
24425 general register in SSE zero-extend sequence.
24427 2017-05-15 Jeff Law <law@redhat.com>
24429 * reorg.c (relax_delay_slots): Create a new variable to hold
24430 the temporary target rather than clobbering TARGET_LABEL.
24432 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
24433 missing argument to extract_bit_field call.
24434 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
24436 2017-05-15 Martin Liska <mliska@suse.cz>
24439 * gcc.c (process_command): Do not allow empty argument of -o option.
24441 2017-05-15 Renlin Li <renlin.li@arm.com>
24443 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
24444 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
24445 * config/aarch64/constraints.md (Usf): Add long call check.
24446 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
24447 (call_value): Likewise.
24448 (sibcall): Likewise.
24449 (sibcall_value): Likewise.
24451 (call_value_insn): New.
24452 (sibcall_insn): Update rtx pattern.
24453 (sibcall_value_insn): Likewise.
24454 (call_internal): Remove.
24455 (call_value_internal): Likewise.
24456 (sibcall_internal): Likewise.
24457 (sibcall_value_internal): Likewise.
24458 (call_reg): Likewise.
24459 (call_symbol): Likewise.
24460 (call_value_reg): Likewise.
24461 (call_value_symbol): Likewise.
24463 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
24466 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
24468 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
24470 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
24471 compatible with CCGOCmode and with CCZmode.
24473 2017-05-14 Martin Sebor <msebor@redhat.com>
24475 PR middle-end/77671
24476 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
24477 (gimple_fold_builtin_snprintf): Same.
24478 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
24479 (gimple_fold_builtin_snprintf): Same.
24480 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
24481 of character types.
24482 (is_call_safe): New function.
24483 (try_substitute_return_value): Call it.
24484 (try_simplify_call): New function.
24485 (pass_sprintf_length::handle_gimple_call): Call it.
24487 2017-05-14 Martin Sebor <msebor@redhat.com>
24489 PR middle-end/80669
24490 * builtins.c (expand_builtin_stpncpy): Simplify.
24492 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
24494 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
24495 * config/i386/i386.h
24496 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
24497 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
24498 (struct machine_function): Add new members call_ms2sysv,
24499 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
24500 (struct machine_frame_state): New fields sp_realigned and
24501 sp_realigned_offset.
24502 * config/i386/i386.c
24503 (enum xlogue_stub): New enum.
24504 (enum xlogue_stub_sets): New enum.
24505 (class xlogue_layout): New class.
24506 (struct ix86_frame): New fields stack_realign_allocate_offset,
24507 stack_realign_offset and outlined_save_offset. Modify comments to
24508 detail stack layout when using out-of-line stubs.
24509 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
24510 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
24511 -mcall-ms2sysv-xlogues.
24512 (stub_managed_regs): New static variable.
24513 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
24514 registers managed by out-of-line stub.
24515 (disable_call_ms2sysv_xlogues): New function.
24516 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
24517 m->call_ms2sysv when appropriate and compute frame layout for
24519 (sp_valid_at, fp_valid_at): New inline functions.
24520 (choose_basereg): New function.
24521 (choose_baseaddr): Add align parameter, use choose_basereg and modify
24523 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
24524 Use align parameter of choose_baseaddr to generated aligned SSE movs
24526 (pro_epilogue_adjust_stack): Modify to track
24527 machine_frame_state::sp_realigned.
24528 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
24529 (ix86_nsaved_sseregs): Likewise.
24530 (ix86_emit_save_regs): Likewise.
24531 (ix86_emit_save_regs_using_mov): Likewise.
24532 (ix86_emit_save_sse_regs_using_mov): Likewise.
24533 (get_scratch_register_on_entry): Likewise.
24534 (gen_frame_set): New function.
24535 (gen_frame_load): Likewise.
24536 (gen_frame_store): Likewise.
24537 (emit_outlined_ms2sysv_save): Likewise.
24538 (emit_outlined_ms2sysv_restore): Likewise.
24539 (ix86_expand_prologue): Modify stack re-alignment code and call
24540 emit_outlined_ms2sysv_save when appropriate.
24541 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
24542 parameter rtx_insn *insn, which allows the function to be used to only
24543 generate the notes.
24544 (ix86_expand_epilogue): Modify validity checks of frame and stack
24545 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
24546 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
24547 * config/i386/predicates.md
24548 (save_multiple): New predicate.
24549 (restore_multiple): Likewise.
24550 * config/i386/sse.md
24551 (save_multiple<mode>): New pattern.
24552 (save_multiple_realign<mode>): Likewise.
24553 (restore_multiple<mode>): Likewise.
24554 (restore_multiple_and_return<mode>): Likewise.
24555 (restore_multiple_leave_return<mode>): Likewise.
24556 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
24558 2017-05-14 Julia Koval <julia.koval@intel.com>
24560 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
24561 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
24562 (__builtin_ia32_xsetbv): New builtins.
24563 * config/i386/i386.c (ix86_expand_special_args_builtin):
24565 (ix86_expand_builtin): Special expand for new intrinsics.
24566 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
24567 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
24568 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
24570 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24572 * cfganal.c (inverted_post_order_compute): Change argument type
24574 * cfganal.h (inverted_post_order_compute): Adjust prototype.
24575 * df-core.c (rest_of_handle_df_initialize): Adjust.
24576 (rest_of_handle_df_finish): Likewise.
24577 (df_analyze_1): Likewise.
24578 (df_analyze): Likewise.
24579 (loop_inverted_post_order_compute): Change argument to be a vec *.
24580 (df_analyze_loop): Adjust.
24581 (df_get_n_blocks): Likewise.
24582 (df_get_postorder): Likewise.
24583 * df.h (struct df_d): Change field to be a vec.
24584 * lcm.c (compute_laterin): Adjust.
24585 (compute_available): Likewise.
24586 * lra-lives.c (lra_create_live_ranges_1): Likewise.
24587 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
24588 * tree-ssa-pre.c (compute_antic): Likewise.
24590 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24592 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
24593 (depth_first_search::depth_first_search): Change structure init
24594 function to this constructor.
24595 (depth_first_search::add_bb): Rename function to this member.
24596 (depth_first_search::execute): Likewise.
24597 (flow_dfs_compute_reverse_finish): Adjust.
24599 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24601 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
24602 (longest_simple_path): Likewise.
24603 * shrink-wrap.c (spread_components): Likewise.
24604 (disqualify_problematic_components): Likewise.
24605 (emit_common_heads_for_components): Likewise.
24606 (emit_common_tails_for_components): Likewise.
24607 (insert_prologue_epilogue_for_components): Likewise.
24609 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24611 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
24614 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24616 * df-core.c (df_set_blocks): Start using auto_bitmap.
24617 (df_compact_blocks): Likewise.
24618 * df-problems.c (df_rd_confluence_n): Likewise.
24619 * df-scan.c (df_insn_rescan_all): Likewise.
24620 (df_process_deferred_rescans): Likewise.
24621 (df_update_entry_block_defs): Likewise.
24622 (df_update_exit_block_uses): Likewise.
24623 (df_entry_block_bitmap_verify): Likewise.
24624 (df_exit_block_bitmap_verify): Likewise.
24625 (df_scan_verify): Likewise.
24626 * lra-constraints.c (lra_constraints): Likewise.
24627 (undo_optional_reloads): Likewise.
24628 (lra_undo_inheritance): Likewise.
24629 * lra-remat.c (calculate_gen_cands): Likewise.
24630 (do_remat): Likewise.
24631 * lra-spills.c (assign_spill_hard_regs): Likewise.
24632 (spill_pseudos): Likewise.
24633 * tree-ssa-pre.c (bitmap_set_and): Likewise.
24634 (bitmap_set_subtract_values): Likewise.
24636 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24638 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
24639 management with auto_bitmap.
24640 (fix_inter_tick): Likewise.
24641 (fix_recovery_deps): Likewise.
24642 * ira.c (add_store_equivs): Likewise.
24643 (find_moveable_pseudos): Likewise.
24644 (split_live_ranges_for_shrink_wrap): Likewise.
24645 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
24646 (rtx_reuse_manager::seen_def_p): Likewise.
24647 (rtx_reuse_manager::set_seen_def): Likewise.
24648 * print-rtl.h (class rtx_reuse_manager): Likewise.
24650 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24652 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
24654 (migrate_btr_def): Likewise.
24655 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
24656 * df-core.c (loop_post_order_compute): Likewise.
24657 (loop_inverted_post_order_compute): Likewise.
24658 * hsa-common.h: Likewise.
24659 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
24660 * init-regs.c (initialize_uninitialized_regs): Likewise.
24661 * ipa-inline.c (resolve_noninline_speculation): Likewise.
24662 (inline_small_functions): Likewise.
24663 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
24664 * ira.c (combine_and_move_insns): Likewise.
24665 (build_insn_chain): Likewise.
24666 * loop-invariant.c (find_invariants): Likewise.
24667 * lower-subreg.c (propagate_pseudo_copies): Likewise.
24668 * predict.c (tree_predict_by_opcode): Likewise.
24669 (predict_paths_leading_to): Likewise.
24670 (predict_paths_leading_to_edge): Likewise.
24671 (estimate_loops_at_level): Likewise.
24672 (estimate_loops): Likewise.
24673 * shrink-wrap.c (try_shrink_wrapping): Likewise.
24674 (spread_components): Likewise.
24675 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
24676 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
24677 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
24678 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
24679 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
24680 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
24681 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
24682 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
24683 (mark_threaded_blocks): Likewise.
24684 (thread_through_all_blocks): Likewise.
24685 * tree-ssa.c (verify_ssa): Likewise.
24686 (execute_update_addresses_taken): Likewise.
24687 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
24689 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24691 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
24693 (post_order_compute): Likewise.
24694 (inverted_post_order_compute): Likewise.
24695 (pre_and_rev_post_order_compute_fn): Likewise.
24697 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24699 * genrecog.c (int_set::int_set): Explicitly construct our
24700 auto_vec base class.
24701 * vec.h (auto_vec::auto_vec): New constructor.
24703 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24705 * bitmap.h (class auto_bitmap): New constructor taking
24706 bitmap_obstack * argument.
24708 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
24710 * bitmap.h (class auto_bitmap): Change type of m_bits to
24711 bitmap_head, and adjust ctor / dtor and member operators.
24713 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
24715 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
24716 when returned register mode doesn't match original mode.
24718 2017-05-12 Jeff Law <law@redhat.com>
24719 Jakub Jelinek <jakub@redhat.com>
24721 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
24722 we look for cc setter after the compare-elim changes.
24723 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
24724 within the vector to match what compare-elim now expects.
24725 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
24726 (xorsi3_flags, one_cmplsi2_flags): Likewise.
24728 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
24729 after the compare-elim changes.
24730 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
24731 the vector to match what compare-elim now expects.
24732 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
24733 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
24734 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
24735 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
24736 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
24738 * config/visium/visium.c (single_set_and_flags): Fix where
24739 we look for cc setter after the compare-elim changes.
24740 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
24741 with the vector to match what compare-elim now expects.
24742 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
24743 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
24744 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
24745 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
24746 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
24747 (neg<mode>2_insn_set_overflow): Likewise.
24749 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
24751 PR middle-end/79794
24752 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
24753 maybe_expand_insn call, set ops[0].target. If still set after call,
24754 set alt_rtl. Add extra arg to recursive calls.
24755 (extract_bit_field): Add alt_rtl argument. Pass to
24757 * expmed.h (extract_bit_field): Fix prototype.
24758 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
24759 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
24760 to extract_bit_field_calls.
24761 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
24762 Pass alt_rtl to extract_bit_field calls.
24763 * calls.c (store_unaligned_arguments_into_psuedos)
24764 load_register_parameters): Pass extra NULL to extract_bit_field calls.
24765 * optabs.c (maybe_legitimize_operand): Clear op->target when call
24767 * optabs.h (struct expand_operand): Add target bitfield.
24769 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24771 * compare-elim.c (try_eliminate_compare): Canonicalize
24772 operation with embedded compare to
24773 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
24774 (set (reg) (operation)].
24776 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
24778 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
24781 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
24782 cost of adding a carry flag for ADC instruction.
24783 [case MINUS]: Ignore the cost of subtracting a carry flag
24784 for SBB instruction.
24786 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
24788 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
24790 * config/rs6000/bmiintrin.h: New file.
24791 * config/rs6000/bmi2intrin.h: New file.
24792 * config/rs6000/x86intrin.h: New file.
24794 2017-05-12 Jeff Law <law@redhat.com>
24796 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
24799 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
24801 PR middle-end/80707
24802 * tree-cfg.c: Remove cfg edges of unreachable case statements.
24804 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24806 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24807 early expansion of vector divide builtins.
24808 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
24809 builtins identified as having unsigned arguments.
24811 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24813 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
24814 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
24815 expansion of vector logical operations (and, andc, or, xor,
24818 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24820 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
24821 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
24823 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
24825 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
24826 early GIMPLE expansion of vector multiplies.
24828 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24830 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
24831 TARGET_HAVE_MOVT conditional.
24832 (movt splitter): Likewise.
24834 2017-05-12 Richard Biener <rguenther@suse.de>
24836 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
24837 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
24838 Fold all stmts not inplace.
24840 2017-05-12 Richard Biener <rguenther@suse.de>
24842 PR tree-optimization/80713
24843 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
24844 inserted_exprs bit for not removed stmts.
24846 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
24848 PR middle-end/69921
24849 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
24850 parallelized" attribute for parallelized OpenACC kernels.
24851 * omp-offload.c (execute_oacc_device_lower): Use it.
24853 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
24854 Set "oacc kernels" attribute.
24855 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
24856 parameter. Adjust all users.
24857 (oacc_fn_attrib_kernels_p): Remove function.
24858 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
24859 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
24860 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
24861 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
24862 assert "oacc kernels" attribute is set.
24864 2017-05-11 Carl Love <cel@us.ibm.com>
24866 * config/rs6000/rs6000-c: Add support for built-in functions
24867 vector unsigned char vec_popcnt (vector signed char)
24868 vector unsigned char vec_popcnt (vector unsigned char)
24869 vector unsigned short vec_popcnt (vector signed short)
24870 vector unsigned short vec_popcnt (vector unsigned short)
24871 vector unsigned int vec_popcnt (vector signed int)
24872 vector unsigned int vec_popcnt (vector unsigned int)
24873 vector unsigned long long vec_popcnt (vector signed long long)
24874 vector unsigned long long vec_popcnt (vector unsigned long long)
24875 vector signed long long vec_slo (vector signed long long,
24876 vector signed char)
24877 vector signed long long vec_slo (vector signed long long,
24878 vector unsigned char)
24879 vector unsigned long long vec_slo (vector unsigned long long,
24880 vector signed char)
24881 vector unsigned long long vec_slo (vector unsigned long long,
24882 vector unsigned char)
24883 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
24884 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
24885 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
24886 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
24887 * doc/extend.texi: Update the built-in documentation file for the
24888 new built-in functions.
24890 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
24892 * attribs.h (sorted_attr_string): Move machine independent
24893 functions for target clone support from the i386 port to common
24894 code. Rename ix86_function_versions to common_function_versions.
24895 Rename make_name to make_unique_name.
24896 (common_function_versions): Likewise.
24897 (make_unique_name): Likewise.
24898 (make_dispatcher_decl): Likewise.
24899 (is_function_default_version): Likewise.
24900 * attribs.c (attr_strcmp): Likewise.
24901 (sorted_attr_string): Likewise.
24902 (common_function_versions): Likewise.
24903 (make_unique_name): Likewise.
24904 (make_dispatcher_decl): Likewise.
24905 (is_function_default_version): Likewise.
24906 * config/i386/i386.c (attr_strcmp): Likewise.
24907 (sorted_attr_string): Likewise.
24908 (ix86_function_versions): Likewise.
24909 (make_name): Likewise.
24910 (make_dispatcher_decl): Likewise.
24911 (is_function_default_version): Likewise.
24912 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
24914 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24917 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
24918 Account for direct move costs for vec_construct of integer
24921 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
24924 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
24925 (UNSPEC_STX_ATOMIC): Ditto.
24926 (loaddi_via_sse): New insn.
24927 (storedi_via_sse): Ditto.
24928 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
24929 Update corresponding peephole2 patterns.
24930 (atomic_storedi_fpu): Ditto.
24932 2017-05-11 Julia Koval <julia.koval@intel.com>
24934 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
24935 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
24937 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
24938 (__builtin_ia32_rsqrt14ss_mask): New builtins.
24939 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
24941 2017-05-11 Nathan Sidwell <nathan@acm.org>
24943 * graphite-poly.c: Include dumpfile.h.
24945 * dumpfle.h (dump_function): Declare here ...
24946 * tree-dump.h (dump_function): ... not here.
24947 * dumpfile.c: #include tree-cfg.h.
24948 (dump_function): Move here from ...
24949 * tree-dump.c (dump_function): ... here.
24950 * gimplify.c: #include splay-tree.h, not tree-dump.h.
24951 * graphite-poly.c: Don't include tree-dump.h.
24952 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
24953 * print-tree.c: Likewise.
24954 * stor-layout.c: Likewise.
24955 * tree-nested.c: Likewise.
24957 * dumpfile.c (dump_start): Use TDF_FLAGS.
24958 (dump_enable_all): Fix TDF_KIND check thinko.
24960 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
24962 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
24963 array entries to represent two legal parameterizations of the
24964 overloaded __builtin_cmpb function, as represented by the
24965 P6_OV_BUILTIN_CMPB constant.
24966 (altivec_resolve_overloaded_builtin): Add special case handling
24967 for the __builtin_cmpb function, as represented by the
24968 P6_OV_BUILTIN_CMPB constant.
24969 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
24970 (BU_P6_64BIT_2): New macro.
24971 (BU_P6_OVERLOAD_2): New macro
24972 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
24973 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
24974 (CMPB): Add overload support to represent both 32-bit and 64-bit
24975 compare-bytes function.
24976 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
24977 support for TARGET_CMPB.
24978 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
24979 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
24980 documentation of the __builtin_cmpb overloaded built-in function.
24982 2017-05-11 Richard Biener <rguenther@suse.de>
24984 PR tree-optimization/80705
24985 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
24986 bases are not vectorizable.
24988 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24990 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
24991 when counting register pressure.
24993 2017-05-11 Bin Cheng <bin.cheng@arm.com>
24995 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
24996 (iv_ca_more_deps): Renamed to ...
24997 (iv_ca_compare_deps): ... this.
24998 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
25000 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25002 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
25004 (determine_group_iv_costs): ... here.
25005 (find_inv_vars_cb): Record inv var if it's not recorded before.
25007 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25009 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
25010 (get_shiftadd_cost): Ditto.
25012 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25014 * tree-ssa-address.c: Include header file.
25015 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
25017 (add_to_parts): Refactor.
25018 (addr_to_parts): New parameter. Update use of move_hint_to_base.
25019 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
25022 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25024 PR tree-optimization/53090
25025 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
25027 (extract_cond_operands): Detect condition with IV on both sides
25028 and return COMP_IV_EXPR_2.
25029 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
25030 (rewrite_use_compare): Simplify by removing call to function
25031 extract_cond_operands.
25033 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25035 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
25036 (extract_cond_operands): Detect condition comparing against non-
25037 invariant bound and return appropriate enum value.
25038 (find_interesting_uses_cond): Update use of extract_cond_operands.
25039 Handle its return value accordingly.
25040 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
25042 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25044 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
25045 nonlinear iv_use computation in loop invariant sensitive way.
25047 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25049 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
25050 (find_iv_candidates): Call relate_compare_use_with_all_cands.
25052 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25054 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
25055 (dump_cand): Support iv_cand.inv_exprs.
25056 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
25058 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
25061 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25063 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
25065 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
25066 as local function. Include necessary header files.
25067 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
25069 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25071 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
25073 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25075 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
25076 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
25077 RSHIFT_EXPR and BIT_NOT_EXPR.
25079 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25081 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
25082 (adjust_setup_cost): New parameter supporting round up adjustment.
25083 (struct address_cost_data): Delete.
25084 (force_expr_to_var_cost): Don't bound cost with spill_cost.
25085 (split_address_cost, ptr_difference_cost): Delete.
25086 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
25087 (struct ainc_cost_data): New struct.
25088 (get_address_cost_ainc): New function.
25089 (get_address_cost, get_computation_cost): Reimplement.
25090 (determine_group_iv_cost_address): Record inv_expr for all uses of
25092 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
25093 (iv_ca_has_deps): Reimplemented to ...
25094 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
25096 (iv_ca_extend): Call iv_ca_more_deps.
25098 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25100 * tree-ssa-address.c (struct mem_address): Move to header file.
25101 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
25102 * tree-ssa-address.h (struct mem_address): Move from C file.
25103 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
25105 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25107 * tree-affine.h (aff_combination_type): New interface.
25108 (aff_combination_zero_p): Remove static.
25109 (aff_combination_const_p): New interface.
25110 (aff_combination_singleton_var_p): New interfaces.
25112 2017-05-11 Richard Biener <rguenther@suse.de>
25114 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
25115 Skip unreachable blocks and destinations.
25116 (eliminate): Move stmt removal and fixup ...
25117 (fini_eliminate): ... here. Skip inserted exprs.
25118 (pass_pre::execute): Move fini_pre after fini_eliminate.
25119 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
25120 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
25121 PRE to get rid of dead code that has invalid SSA form and
25122 split critical edges again.
25124 2017-05-11 Bin Cheng <bin.cheng@arm.com>
25126 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25128 2017-05-11 Richard Biener <rguenther@suse.de>
25130 * passes.c (execute_function_todo): Verify loops if they are
25131 said to be up-to-date.
25132 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
25133 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
25135 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
25138 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
25139 handle calling assemble_external ourself.
25142 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
25143 modes with zero size. Enhance comment.
25145 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25147 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
25148 built-ins for vec_xl and vec_xst with short and char pointer
25151 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25153 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
25154 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
25155 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
25156 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
25157 (_mm_maskz_min_round_ss): New intrinsics.
25158 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
25159 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
25160 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
25161 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
25162 (__builtin_ia32_minss_mask_round): New builtins.
25163 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25164 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25165 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
25167 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
25168 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
25170 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
25173 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
25175 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
25176 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
25177 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
25178 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
25179 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
25180 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
25181 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
25182 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25183 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25184 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
25185 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
25186 (__builtin_ia32_mulss_mask_round): New builtins.
25187 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25188 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25189 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
25191 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
25192 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25194 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25197 2017-05-10 Julia Koval <julia.koval@intel.com>
25199 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
25200 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
25201 (_mm256_setr_m128i): New intrinsics.
25203 2017-05-10 Julia Koval <julia.koval@intel.com>
25205 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
25206 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
25207 (_mm_maskz_rcp14_ss): New intrinsics.
25208 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
25209 (__builtin_ia32_rcp14ss_mask): New builtins.
25210 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
25212 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
25214 PR tree-optimization/51513
25215 * tree-cfg.c (gimple_seq_unreachable_p): New function.
25216 (assert_unreachable_fallthru_edge_p): Use it.
25217 (group_case_labels_stmt): Likewise.
25218 * tree-cfg.h: Prototype it.
25219 * stmt.c: Include cfghooks.h and tree-cfg.h.
25220 (emit_case_dispatch_table) <gap_label>: New local variable.
25221 Use it to fill dispatch table gaps.
25222 Test for default_label before updating probabilities.
25223 (expand_case) <default_label>: Remove unneeded initialization.
25224 Test for unreachable default case statement and remove its edge.
25225 Set default_label accordingly.
25226 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
25228 2017-05-10 Carl Love <cel@us.ibm.com>
25230 * config/rs6000/rs6000-c: Add support for built-in functions
25231 vector signed char vec_neg (vector signed char)
25232 vector signed short int vec_neg (vector short int)
25233 vector signed int vec_neg (vector signed int)
25234 vector signed long long vec_neg (vector signed long long)
25235 vector float vec_neg (vector float)
25236 vector double vec_neg (vector double)
25237 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
25239 * config/rs6000/altivec.h: Add define for vec_neg
25240 * doc/extend.texi: Update the built-in documentation for the
25241 new built-in functions.
25243 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25245 PR tree-optimization/77644
25246 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
25248 2017-05-10 Nathan Sidwell <nathan@acm.org>
25250 * dumpfile.h (TDI_lang_all): New.
25251 (TDF_KIND): New. Renumber others
25252 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
25254 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
25256 (get_dump_file_name): Adjust suffix generation.
25257 (dump_enable_all): Use TDF_KIND.
25258 * doc/invoke.texi (-fdump-lang-all): Document.
25260 * dumpfile.h: Tabify.
25262 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
25265 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
25266 Move member access before delete.
25268 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
25270 * tree-inline.c (expand_call_inline): Split block at stmt
25273 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
25276 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
25277 are now unused after splitting mov{sf,sd}_hardfloat.
25278 (f32_lr2): Likewise.
25279 (f32_lm): Likewise.
25280 (f32_lm2): Likewise.
25281 (f32_li): Likewise.
25282 (f32_li2): Likewise.
25283 (f32_lv): Likewise.
25284 (f32_sr): Likewise.
25285 (f32_sr2): Likewise.
25286 (f32_sm): Likewise.
25287 (f32_sm2): Likewise.
25288 (f32_si): Likewise.
25289 (f32_si2): Likewise.
25290 (f32_sv): Likewise.
25291 (f32_dm): Likewise.
25292 (f32_vsx): Likewise.
25293 (f32_av): Likewise.
25294 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
25295 For movsf, order stores so the VSX stores occur before the GPR
25296 store which encourages the register allocator to use a traditional
25297 FPR instead of a GPR. For movsd, order the stores so that the GPR
25298 store comes before the VSX stores to allow the power6 to work.
25299 This is due to the power6 not having a 32-bit integer store
25300 instruction from a FPR.
25301 (movsf_hardfloat): Likewise.
25302 (movsd_hardfloat): Likewise.
25304 2017-05-09 Martin Sebor <msebor@redhat.com>
25306 PR translation/80280
25307 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
25310 PR translation/80280
25311 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
25312 data member added in r247778.
25313 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
25315 2017-05-09 Nathan Sidwell <nathan@acm.org>
25317 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
25319 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
25322 2017-05-09 Marek Polacek <polacek@redhat.com>
25324 * doc/invoke.texi: Fix typo.
25326 2017-05-09 Richard Biener <rguenther@suse.de>
25328 * tree-vrp.c (vrp_val_is_max): Adjust comment.
25329 (vrp_val_is_min): Likewise.
25330 (set_value_range_to_value): Likewise.
25331 (set_value_range_to_nonnegative): Likewise.
25332 (gimple_assign_nonzero_p): Likewise.
25333 (gimple_stmt_nonzero_p): Likewise.
25334 (vrp_int_const_binop): Likewise. Remove unreachable case.
25335 (adjust_range_with_scev): Adjust comments.
25336 (compare_range_with_value): Likewise.
25337 (extract_range_from_phi_node): Likewise.
25338 (test_for_singularity): Likewise.
25340 2017-05-09 Richard Biener <rguenther@suse.de>
25342 * tree-vrp.c (get_single_symbol): Add assert that we don't
25343 get overflowed constants as invariant part.
25344 (compare_values_warnv): Add comment before the TREE_NO_WARNING
25345 checks. Use wi::cmp instead of recursing for integer constants.
25346 (compare_values): Just ignore whether we assumed undefined
25347 overflow instead of failing the compare.
25348 (extract_range_for_var_from_comparison_expr): Add comment before the
25349 TREE_NO_WARNING sets.
25350 (test_for_singularity): Likewise.
25351 (extract_range_from_comparison): Do not disable optimization
25352 when we assumed undefined overflow.
25353 (extract_range_basic): Remove init of unused var.
25355 2017-05-09 Richard Biener <rguenther@suse.de>
25357 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
25358 (extract_range_from_multiplicative_op_1): Adjust.
25359 (extract_range_from_binary_expr_1): Use int_const_binop.
25361 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
25364 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
25365 rs6000_store_data_bypass_p in seven define_bypass directives and
25366 in several comments.
25367 * config/rs6000/rs6000-protos.h: Add prototype for
25368 rs6000_store_data_bypass_p function.
25369 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
25370 function implements slightly different (rs6000-specific) semantics
25371 than store_data_bypass_p, returning false rather than aborting
25372 with assertion error when arguments do not satisfy the
25373 requirements of store data bypass.
25374 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
25375 rs6000_store_data_bypass_p.
25377 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
25379 * config/xtensa/xtensa-protos.h
25380 (xtensa_initial_elimination_offset): New declaration.
25381 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
25382 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
25383 macro definition, add case for FRAME_POINTER_REGNUM when
25384 FRAME_GROWS_DOWNWARD.
25385 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
25386 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
25387 xtensa_initial_elimination_offset.
25389 2017-05-08 Nathan Sidwell <nathan@acm.org>
25391 * doc/invoke.texi: Alphabetize -fdump options.
25393 2017-05-08 Martin Sebor <msebor@redhat.com>
25395 PR translation/80280
25396 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
25398 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
25400 * target.def (compute_frame_layout): New optional target hook.
25401 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
25402 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
25403 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
25405 * reload1.c (verify_initial_elim_offsets): Likewise.
25406 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
25407 (use_simple_return_p): Call arm_compute_frame_layout if needed.
25408 (arm_get_frame_offsets): Split up into this ...
25409 (arm_compute_frame_layout): ... and this function.
25411 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
25413 * config/aarch64/constraints.md (Usa): New constraint.
25414 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
25416 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
25418 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
25419 with_multilib_list after it has been checked.
25421 2017-05-08 Richard Biener <rguenther@suse.de>
25423 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
25424 (bitmap_set_subtract_values): Likewise.
25426 2017-05-08 Richard Biener <rguenther@suse.de>
25428 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
25429 (gimple_assign_nonzero): ... this and remove strict_overflow_p
25431 (gimple_stmt_nonzero_warnv_p): Rename to ...
25432 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
25434 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
25435 (extract_range_basic): Adjust, do not disable propagation on
25436 strict overflow sensitive simplification.
25437 (vrp_visit_cond_stmt): Likewise.
25439 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
25441 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
25442 body size unconditionally.
25444 2017-05-07 Jeff Law <law@redhat.com>
25447 2017-05-06 Jeff Law <law@redhat.com>
25448 PR tree-optimization/78496
25449 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25452 PR tree-optimization/78496
25453 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25454 (simplify_stmt_using_ranges): Call it.
25455 (vrp_dom_walker::before_dom_children): Extract equivalences
25456 from an ASSERT_EXPR with an equality comparison against a
25459 2017-05-06 Jeff Law <law@redhat.com>
25461 PR tree-optimization/78496
25462 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
25465 PR tree-optimization/78496
25466 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
25467 (simplify_stmt_using_ranges): Call it.
25468 (vrp_dom_walker::before_dom_children): Extract equivalences
25469 from an ASSERT_EXPR with an equality comparison against a
25472 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25474 * lra-constraints.c (lra_copy_reg_equiv): New function.
25475 (split_reg): Use it to copy equivalence information from the
25476 original register to the spill register.
25478 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
25480 PR rtl-optimization/75964
25481 * simplify-rtx.c (simplify_const_relational_operation): Remove
25482 invalid handling of comparisons of integer ABS.
25484 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
25486 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
25487 initialize to zero.
25488 (init_regs): Remove declaration.
25489 (function_arg_advance_32): Initialize error_p as boolean variable.
25491 2017-05-05 Nathan Sidwell <nathan@acm.org>
25493 * store-motion.c (remove_reachable_equiv_notes): Reformat long
25494 lines. Use for (;;).
25496 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25498 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
25499 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
25500 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
25501 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
25502 VF=2 that require versioning.
25504 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25506 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
25509 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25511 * diagnostic.h (diagnostic_override_option_index): Convert from
25512 macro to inline function.
25514 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25516 * diagnostic.c (last_module_changed_p): New function.
25517 (set_last_module): New function.
25518 (diagnostic_report_current_module): Convert macro usage to
25519 the above functions.
25520 * diagnostic.h (diagnostic_context::last_module): Strengthen
25521 from const line_map * to const line_map_ordinary *.
25522 (diagnostic_last_module_changed): Delete macro.
25523 (diagnostic_set_last_module): Delete macro.
25525 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25527 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
25528 with diagnostic_report_diagnostic.
25529 (diagnostic_n_impl_richloc): Likewise.
25530 * diagnostic.h (report_diagnostic): Delete macro.
25531 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
25532 with diagnostic_report_diagnostic.
25533 * substring-locations.c (format_warning_va): Likewise.
25535 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25537 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
25538 save/restor of format_spec. Move option-printing code to...
25539 (print_option_information): ...this new function, and
25540 reimplement by simply printing to the pretty_printer,
25541 rather than appending to the format string.
25543 2017-05-05 David Malcolm <dmalcolm@redhat.com>
25545 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
25546 handling logic into...
25547 (update_effective_level_from_pragmas): ...this new function.
25549 2017-05-04 Andrew Waterman <andrew@sifive.com>
25551 * config/riscv/riscv.opt (mstrict-align): New option.
25552 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
25553 (SLOW_UNALIGNED_ACCESS): Define.
25554 (riscv_slow_unaligned_access): Declare.
25555 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
25557 (riscv_slow_unaligned_access): New variable.
25558 (rocket_tune_info): Set slow_unaligned_access to true.
25559 (optimize_size_tune_info): Set slow_unaligned_access to false.
25560 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
25561 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
25562 (riscv_option_override): Set riscv_slow_unaligned_access.
25563 * doc/invoke.texi: Add -mstrict-align to RISC-V.
25565 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
25567 * config/riscv/riscv.md: Unify indentation.
25569 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
25574 * config/rs6000/rs6000.md (u code attribute): Add FIX and
25576 (extendsi<mode>2): Add support for doing sign extension via
25577 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
25578 don't have ISA 3.0 instructions.
25579 (extendsi<mode>2 splitter): Likewise.
25580 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
25581 generate the normal insns since SImode can now go in vector
25582 registers. Disallow the special UNSPECs needed for previous
25583 machines to hide SImode being used. Add new insns
25584 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
25585 (fix_trunc<mode>si2_stfiwx): Likewise.
25586 (fix_trunc<mode>si2_internal): Likewise.
25587 (fixuns_trunc<mode>si2): Likewise.
25588 (fixuns_trunc<mode>si2_stfiwx): Likewise.
25589 (fctiw<u>z_<mode>_smallint): Likewise.
25590 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
25591 of floating point to 32-bit integer from doing a direct move to
25592 the GPR registers to do a store.
25593 (fctiwz_<mode>): Break long line.
25595 2017-05-05 Bin Cheng <bin.cheng@arm.com>
25597 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
25598 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
25599 (addr_list, addr_offset_valid_p): New.
25600 (split_address_groups): Check offset validity with above function.
25601 (gt-tree-ssa-loop-ivopts.h): Include header file.
25603 2017-05-05 Nathan Sidwell <nathan@acm.org>
25605 * config.gcc (arm*-*-*): Add missing 'fi'.
25607 2017-05-05 Steve Ellcey <sellcey@cavium.com>
25609 * doc/invoke.texi (-fopt-info): Explicitly say order of options
25610 included in -fopt-info does not matter.
25611 * doc/optinfo.texi (-fopt-info): Fix description of default
25612 behavour. Explicitly say order of options included in -fopt-info
25615 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
25617 * config.gcc: Allow combinations of aprofile and rmprofile values for
25618 --with-multilib-list.
25619 * config/arm/t-multilib: New file.
25620 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
25621 variables. Remove setting of ISA and floating-point ABI in
25622 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
25623 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
25624 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
25625 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
25627 * config/arm/t-rmprofile: Likewise except for the matches changes.
25628 * doc/install.texi (--with-multilib-list): Document the combination of
25629 aprofile and rmprofile values and warn about pitfalls in doing that.
25631 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25633 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
25634 (movdi_aarch64): Likewise.
25636 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25638 PR tree-optimization/80632
25639 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
25641 (build_arrays): Initialize it for virtual phis.
25642 (fix_phi_nodes): Use it for virtual phis.
25644 PR tree-optimization/80558
25645 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
25646 [x, y] op z into [x op, y op z] for op & or | if conditions
25649 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
25650 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25653 * config/arm/arm.md (use_literal_pool): Remove.
25654 (64-bit immediate split): No longer takes cost into consideration
25655 if arm_disable_literal_pool is enabled.
25656 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
25657 used when arm_disable_literal_pool is enabled.
25658 (arm_max_const_double_inline_cost): Remove use of
25659 arm_disable_literal_pool.
25660 (push_minipool_fix): Add assert.
25661 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
25662 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
25663 (no_literal_pool_sf_immediate): New.
25665 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25667 PR tree-optimization/80613
25668 * tree-ssa-dce.c (propagate_necessity): Remove cases for
25669 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
25671 2017-05-05 Richard Biener <rguenther@suse.de>
25673 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
25675 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
25677 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
25678 of this flag from insn conditions due to removal from r247495.
25680 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
25682 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
25684 (arm_early_store_addr_dep_ptr): Likewise.
25685 * config/arm/aarch-common-protos.h
25686 (arm_early_load_addr_dep_ptr): Add prototype.
25687 (arm_early_store_addr_dep_ptr): Likewise.
25688 * config/arm/cortex-a53.md: Add new bypasses.
25690 2017-05-05 Jakub Jelinek <jakub@redhat.com>
25692 * tree.c (next_type_uid): Change type to unsigned.
25693 (type_hash_canon): Decrement back next_type_uid if
25694 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
25695 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
25698 2017-05-04 Martin Sebor <msebor@redhat.com>
25700 * builtins.c: Fix a trivial typo in a comment.
25702 PR middle-end/79234
25703 * builtins.c (check_sizes): Adjust to handle reading past the end.
25704 Avoid printing excessive upper bound of ranges. Use %E to print
25705 tree nodes instead of converting them to %wu.
25706 (expand_builtin_memchr): New function.
25707 (compute_dest_size): Rename...
25708 (compute_objsize): ...to this.
25709 (expand_builtin_memcpy): Adjust.
25710 (expand_builtin_mempcpy): Adjust.
25711 (expand_builtin_strcat): Adjust.
25712 (expand_builtin_strcpy): Adjust.
25713 (check_strncat_sizes): Adjust.
25714 (expand_builtin_strncat): Adjust.
25715 (expand_builtin_strncpy): Adjust and simplify.
25716 (expand_builtin_memset): Adjust.
25717 (expand_builtin_bzero): Adjust.
25718 (expand_builtin_memcmp): Adjust.
25719 (expand_builtin): Handle memcmp.
25720 (maybe_emit_chk_warning): Check strncat just once.
25722 2017-05-04 Martin Sebor <msebor@redhat.com>
25724 PR preprocessor/79214
25725 PR middle-end/79222
25726 PR middle-end/79223
25727 * builtins.c (check_sizes): Add inlining context and issue
25728 warnings even when -Wno-system-headers is set.
25729 (check_strncat_sizes): Same.
25730 (expand_builtin_strncat): Same.
25731 (expand_builtin_memmove): New function.
25732 (expand_builtin_stpncpy): Same.
25733 (expand_builtin): Handle memmove and stpncpy.
25735 2017-05-04 Bin Cheng <bin.cheng@arm.com>
25737 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
25738 which is not used any more.
25740 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25742 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
25744 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25746 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
25747 (cortexa53_tunings): Likewise.
25748 (cortexa57_tunings): Likewise.
25749 (cortexa72_tunings): Likewise.
25750 (cortexa73_tunings): Likewise.
25752 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25754 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
25755 Set loop alignment to 8.
25757 2017-05-04 Martin Sebor <msebor@redhat.com>
25759 PR translation/80280
25760 * builtins.c (expand_builtin_object_size): Add missing quoting to
25761 %D and like directives.
25762 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
25763 (hsa_type_for_tree_type): Same.
25764 (verify_function_arguments): Same.
25765 * symtab.c (symbol_table::change_decl_assembler_name): Same.
25766 * varasm.c (get_section): Same.
25769 2017-05-04 Martin Sebor <msebor@redhat.com>
25771 PR translation/80280
25772 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
25774 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
25776 * config/aarch64/aarch64.c (generic_addrcost_table):
25777 Change HI/TI mode setting.
25779 2017-05-04 Martin Jambor <mjambor@suse.cz>
25781 PR tree-optimization/80622
25782 * tree-sra.c (comes_initialized_p): New function.
25783 (build_accesses_from_assign): Only set write lazily when
25784 comes_initialized_p is false.
25785 (analyze_access_subtree): Use comes_initialized_p.
25786 (propagate_subaccesses_across_link): Assert !comes_initialized_p
25787 instead of testing for PARM_DECL.
25789 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25791 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
25792 constraint on operand 0 to allow more general addressing modes.
25793 Adjust output template.
25794 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
25796 * config/aarch64/aarch64-protos.h
25797 (aarch64_address_valid_for_prefetch_p): Declare prototype.
25798 * config/aarch64/constraints.md (Dp): New address constraint.
25799 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
25802 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
25804 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
25805 update use of estimate_ipcp_clone_size_and_time.
25806 (estimate_local_effects): Update use of
25807 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
25808 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
25809 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
25810 Return nonspecialized time.
25812 2017-05-04 Richard Biener <rguenther@suse.de>
25814 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
25815 for the last VUSE which def dominates the PHI. Directly call
25817 (get_continuation_for_phi_1): Remove.
25819 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
25821 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
25822 to explain the use of truncating division. Cap the number of
25823 iterations to the maximum given by nb_iterations_upper_bound,
25826 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25828 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
25829 * configure: Regenerate.
25830 * config.in: Regenerate.
25831 * config/i386/driver-mingw32.c: new file.
25832 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
25833 * config.host: Link driver-mingw32.o on MinGW host.
25834 * doc/install.texi: Document new --enable-mingw-wildcard configure
25837 2017-05-04 Marek Polacek <polacek@redhat.com>
25839 PR tree-optimization/80612
25840 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
25842 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25843 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
25845 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
25846 (movt splitter): Likewise.
25847 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
25848 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
25849 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
25850 block for Thumb-1 with MOVT.
25851 (thumb2_legitimate_address_p): Move code block ...
25852 (can_avoid_literal_pool_for_label_p): ... into this new function.
25853 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
25855 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
25856 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
25857 "M-profile targets with the MOVT instruction".
25859 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
25861 * config/arm/arm-builtins.c (arm_init_builtins): Rename
25862 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
25863 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
25865 2017-05-04 Martin Liska <mliska@suse.cz>
25867 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
25868 variable cond_code.
25870 2017-05-04 Richard Biener <rguenther@suse.de>
25872 * tree.c (array_at_struct_end_p): Handle arrays at struct
25873 end with flexarrays more conservatively. Refactor and treat
25874 arrays of arrays or aggregates more strict. Fix
25875 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
25876 * tree.c (array_at_struct_end_p): Adjust prototype.
25877 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
25878 * gimple-fold.c (get_range_strlen): Likewise.
25879 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
25881 2017-05-04 Richard Biener <rguenther@suse.de>
25883 PR tree-optimization/31130
25884 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
25886 (supports_overflow_infinity): Likewise.
25887 (is_negative_overflow_infinity): Likewise.
25888 (is_positive_overflow_infinity): Likewise.
25889 (is_overflow_infinity): Likewise.
25890 (stmt_overflow_infinity): Likewise.
25891 (overflow_infinity_range_p): Likewise.
25892 (usable_range_p): Remove as always returning true.
25893 (make_overflow_infinity): Remove.
25894 (negative_overflow_infinity): Likewise.
25895 (positive_overflow_infinity): Likewise.
25896 (avoid_overflow_infinity): Likewise.
25897 (set_value_range): Adjust accordingly.
25898 (set_value_range_to_nonnegative): Likewise, remove now unused
25899 overflow_infinity arg.
25900 (vrp_operand_equal_p): Adjust.
25901 (update_value_range): Likewise.
25902 (range_int_cst_singleton_p): Likewise.
25903 (operand_less_p): Likewise.
25904 (compare_values_warnv): Likewise.
25905 (extract_range_for_var_from_comparison_expr): Likewise.
25906 (vrp_int_const_binop): Likewise.
25907 (zero_nonzero_bits_from_vr): Likewise.
25908 (extract_range_from_multiplicative_op_1): Likewise.
25909 (extract_range_from_binary_expr_1): Likewise.
25910 (extract_range_from_unary_expr): Likewise.
25911 (extract_range_from_comparison): Likewise.
25912 (extract_range_basic): Likewise.
25913 (adjust_range_with_scev): Likewise.
25914 (compare_ranges): Likewise.
25915 (compare_range_with_value): Likewise.
25916 (dump_value_range): Likewise.
25917 (test_for_singularity): Likewise, remove strict_overflow_p parameter
25919 (simplify_cond_using_ranges): Adjust.
25921 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
25923 * brig-builtins.def: Added a builtin for class_f64.
25924 * builtin-types.def: Added a builtin type needed by class_f64.
25926 2017-05-03 Jason Merrill <jason@redhat.com>
25928 * timevar.def: Add TV_CONSTEXPR.
25930 2017-05-03 David Malcolm <dmalcolm@redhat.com>
25932 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
25934 2017-05-03 Martin Jambor <mjambor@suse.cz>
25936 * ipa-prop.c (ipa_update_after_lto_read): Removed.
25937 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
25938 * ipa-cp.c (ipcp_propagate_stage): Do not call
25939 ipa_update_after_lto_read.
25940 * ipa-inline.c (ipa_inline): Likewise.
25942 2017-05-03 Martin Jambor <mjambor@suse.cz>
25944 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
25945 tag. Added a default constructor and a destructor.
25946 (ipa_edge_args_sum_t): New class;
25947 (ipa_edge_args_sum): Declare.
25948 (ipa_edge_args_vector): Remove declaration.
25949 (IPA_EDGE_REF): Use ipa_edge_args_sum.
25950 (ipa_free_edge_args_substructures): Remove declaration.
25951 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
25952 (ipa_edge_args_info_available_for_edge_p): Likewise.
25953 * ipa-prop.c (ipa_edge_args_vector): Removed.
25954 (edge_removal_hook_holder): Likewise.
25955 (edge_duplication_hook_holder): Likewise.
25956 (ipa_edge_args_sum): New variable.
25957 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
25958 ipa_edge_args_vector.
25959 (ipa_free_edge_args_substructures): Likewise.
25960 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
25961 ipa_edge_args_vector.
25962 (ipa_edge_removal_hook): Turned into method
25963 ipa_edge_args_sum_t::remove.
25964 (ipa_edge_duplication_hook): Turned into method
25965 ipa_edge_args_sum_t::duplicate.
25966 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
25967 registering edge hooks.
25968 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
25969 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
25970 ipa_edge_args_sum instead of ipa_edge_args_vector.
25971 * ipa-profile.c (ipa_profile): Likewise.
25973 2017-05-03 Martin Jambor <mjambor@suse.cz>
25975 * symbol-summary.h (function_summary): New method exists.
25976 (function_summary::symtab_removal): Deallocate through release.
25977 (call_summary): New class.
25978 (gt_ggc_mx): New overload.
25979 (gt_pch_nx): Likewise.
25980 (gt_pch_nx): Likewise.
25982 2017-05-03 Jeff Law <law@redhat.com>
25984 PR tree-optimization/78496
25985 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
25986 from simplify_cond_using_ranges. Split off code to walk
25987 backwards through casts into ...
25988 (simplify_cond_using_ranges_2): New function.
25989 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
25990 (execute_vrp): After identifying jump threads, call
25991 simplify_cond_using_ranges_2.
25993 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
25996 * ipa-inline.h (inline_summary): Add ctor.
25997 (create_ggc): Do not use ggc_cleared_alloc.
25999 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
26000 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
26002 * gcc.c (handle_braces): Support escaping in switch matching
26004 * doc/invoke.texi (Spec Files): Document it.
26005 Remove superfluous @code markup in items.
26007 2017-05-03 David Malcolm <dmalcolm@redhat.com>
26009 * diagnostic-show-locus.c (struct column_range): New struct.
26010 (get_affected_columns): New function.
26011 (get_printed_columns): New function.
26012 (struct correction): New struct.
26013 (correction::ensure_capacity): New function.
26014 (correction::ensure_terminated): New function.
26015 (struct line_corrections): New struct.
26016 (line_corrections::~line_corrections): New dtor.
26017 (line_corrections::add_hint): New function.
26018 (layout::print_trailing_fixits): Reimplement in terms of the new
26020 (selftest::test_overlapped_fixit_printing): New function.
26021 (selftest::diagnostic_show_locus_c_tests): Call it.
26023 2017-05-03 Nathan Sidwell <nathan@acm.org>
26025 Canonicalize canonical type hashing
26026 * tree.h (type_hash_canon_hash): Declare.
26027 * tree.c (type_hash_list, attribute_hash_list): Move into
26028 type_hash_canon_hash.
26029 (build_type_attribute_qual_variant): Break out hash code calc into
26030 type_hash_canon_hash.
26031 (type_hash_canon_hash): New. Generic type hash computation.
26032 (build_range_type_1, build_array_type_1, build_function_type,
26033 build_method_type_directly, build_offset_type, build_complex_type,
26034 make_vector_type): Call it.
26036 2017-05-03 Richard Biener <rguenther@suse.de>
26038 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
26039 When all DRs have unknown misaligned do not always peel
26040 when there is a store but apply the same costing model as if
26041 there were only loads.
26043 2017-05-03 Richard Biener <rguenther@suse.de>
26046 PR tree-optimization/80492
26047 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26048 compare_base_decls returning dont-know properly.
26050 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
26052 * config/arm/iterators.md (CCSI): New mode iterator.
26053 (arch): New mode attribute.
26054 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
26055 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
26056 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
26057 code iterator for success result mode.
26058 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
26059 the corresponding new insn generators.
26061 2017-05-03 Bin Cheng <bin.cheng@arm.com>
26064 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26065 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26067 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
26069 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
26070 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
26071 (DDR_A): Wrap DDR argument in brackets.
26072 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
26073 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
26074 (DDR_REVERSED_P): Likewise.
26076 2017-05-03 Jakub Jelinek <jakub@redhat.com>
26078 PR tree-optimization/79472
26079 * tree-switch-conversion.c (struct switch_conv_info): Add
26080 contiguous_range and default_case_nonstandard fields.
26081 (collect_switch_conv_info): Compute contiguous_range and
26082 default_case_nonstandard fields, don't clear final_bb if
26083 contiguous_range and only the default case doesn't have the required
26085 (check_all_empty_except_final): Set default_case_nonstandard instead
26086 of failing if contiguous_range and the default case doesn't have empty
26088 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
26089 and only the default case doesn't have the required constants. Skip
26091 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
26092 if default_case_nonstandard.
26093 (build_constructors): Build constant 1 just once. Assert that default
26094 values aren't inserted in between cases if contiguous_range. Skip
26096 (build_arrays): Skip virtual phis.
26097 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
26098 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
26099 Handle virtual phis.
26100 (gen_inbound_check): Handle default_case_nonstandard case.
26101 (process_switch): Adjust check_final_bb caller. Call
26102 gather_default_values with the first non-default case instead of
26103 default case if default_case_nonstandard.
26105 2017-05-02 Nathan Sidwell <nathan@acm.org>
26107 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
26108 check. Fix formatting.
26110 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
26112 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
26113 errors when comparing specialized and unspecialized times.
26115 2017-05-02 David Malcolm <dmalcolm@redhat.com>
26117 * diagnostic-show-locus.c
26118 (layout::should_print_annotation_line_p): Make private.
26119 (layout::print_annotation_line): Make private.
26120 (layout::annotation_line_showed_range_p): Make private.
26121 (layout::show_ruler): Make private.
26122 (layout::print_source_line): Make private. Pass in line and
26123 line_width, rather than calling location_get_source_line. Drop
26125 (layout::print_leading_fixits): New method.
26126 (layout::print_any_fixits): Rename to...
26127 (layout::print_trailing_fixits): ...this, and make private.
26128 Don't print newline fixits.
26129 (diagnostic_show_locus): Move logic for printing one row into...
26130 (layout::print_line): ...this new function. Move the
26131 location_get_source_line call and error-handling from
26132 print_source_line to here. Call print_leading_fixits, and rename
26133 print_any_fixits to print_trailing_fixits.
26134 (selftest::test_fixit_insert_containing_newline): Update now that
26135 newlines are partially supported.
26136 (selftest::test_fixit_insert_containing_newline_2): New test.
26137 (selftest::test_fixit_replace_containing_newline): Update comments.
26138 (selftest::diagnostic_show_locus_c_tests): Call the new test.
26139 * edit-context.c (class added_line): New class.
26140 (class edited_line): Describe newline handling in comment.
26141 (edited_line::actually_edited_p): New method.
26142 (edited_line::print_content): Delete redundant decl.
26143 (edited_line::m_predecessors): New field.
26144 (edited_file::print_content): Call edited_line::print_content.
26145 (edited_file::print_diff): Update to support newlines.
26146 (edited_file::print_diff_hunk): Likewise.
26147 (edited_file::print_run_of_changed_lines): New function.
26148 (edited_file::print_diff_line): Convert to...
26149 (print_diff_line): ...this.
26150 (edited_file::get_effective_line_count): New function.
26151 (edited_line::edited_line): Initialize new field m_predecessors.
26152 (edited_line::~edited_line): Clean up m_predecessors.
26153 (edited_line::apply_fixit): Handle newlines.
26154 (edited_line::get_effective_line_count): New function.
26155 (edited_line::print_content): New function.
26156 (edited_line::print_diff_lines): New function.
26157 (selftest::test_applying_fixits_insert_containing_newline): New
26159 (selftest::test_applying_fixits_replace_containing_newline): New
26161 (selftest::insert_line): New function.
26162 (selftest::test_applying_fixits_multiple_lines): Add example of
26164 (selftest::edit_context_c_tests): Call the new tests.
26166 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26168 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
26169 parameter cand. Update dump information.
26170 (get_computation_cost): Update uses.
26172 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26174 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
26175 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
26176 (get_computation_at, rewrite_use_address): Update use of
26177 get_computation_aff.
26179 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26181 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
26182 (get_computation): Delete.
26183 (get_computation_cost): Implement like get_computation_cost_at.
26184 Use get_computation_at.
26185 (get_computation_cost_at): Delete.
26186 (rewrite_use_nonlinear_expr): Use get_computation_at.
26187 (rewrite_use_compare, remove_unused_ivs): Ditto.
26189 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26191 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
26193 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26195 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
26196 (ivopts_global_cost_for_size): Rename parameter and update uses.
26197 (iv_ca_recount_cost): Update uses.
26198 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
26199 candidates seperately in n_invs and n_cands.
26200 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
26202 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26204 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
26205 (find_inv_vars_cb): New.
26206 (find_depends): Renamed to ...
26207 (find_inv_vars): ... this.
26208 (add_candidate_1, force_var_cost): Call find_inv_vars.
26209 (split_address_cost, determine_group_iv_cost_cond): Ditto.
26211 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26213 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
26214 inv_vars. Add inv_exprs.
26215 (struct iv_cand): Rename depends_on to inv_vars.
26216 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
26217 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
26218 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
26219 (dump_cand): Dump inv_vars.
26220 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
26221 (record_invariant, find_depends, add_candidate_1): Ditto.
26222 (set_group_iv_cost, force_var_cost): Ditto.
26223 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
26224 (get_computation_cost_at, get_computation_cost): Ditto.
26225 (determine_group_iv_cost_generic): Ditto.
26226 (determine_group_iv_cost_address): Ditto.
26227 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
26228 (determine_group_iv_costs): Ditto.
26229 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
26230 (iv_ca_set_remove_invariants): Renamed to ...
26231 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
26232 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
26233 (iv_ca_set_add_invariants): Renamed to ...
26234 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
26235 (iv_ca_set_cp): Use iv_ca_set_add_invs.
26236 (iv_ca_has_deps): Support inv_vars and inv_exprs.
26237 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
26238 (create_new_ivs): Remove useless dump.
26240 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26242 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
26244 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
26245 (iv_ca_set_no_cp, create_new_iv): Ditto.
26247 2017-05-02 Bin Cheng <bin.cheng@arm.com>
26249 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
26251 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26253 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
26254 function tree_check2.
26256 2017-05-02 Martin Liska <mliska@suse.cz>
26258 * doc/gcov.texi: Add missing preposition.
26259 * gcov.c (function_info::function_info): Properly fill up
26260 all member variables.
26262 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26264 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
26266 2017-05-02 Tamar Christina <tamar.christina@arm.com>
26268 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
26270 2017-05-02 Martin Liska <mliska@suse.cz>
26273 * lto-streamer-in.c (lto_read_tree_1): Remove
26274 LTO_STREAMER_DEBUG.
26275 * lto-streamer.c (struct tree_hash_entry): Likewise.
26276 (struct tree_entry_hasher): Likewise.
26277 (tree_entry_hasher::hash): Likewise.
26278 (tree_entry_hasher::equal): Likewise.
26279 (lto_streamer_init): Likewise.
26280 (lto_orig_address_map): Likewise.
26281 (lto_orig_address_get): Likewise.
26282 (lto_orig_address_remove): Likewise.
26283 * lto-streamer.h: Likewise.
26284 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
26285 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
26287 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
26289 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
26290 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
26291 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
26292 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
26293 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
26294 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
26295 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
26296 (mm_maskz_sub_ss): New intrinsics.
26297 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26298 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
26299 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
26300 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
26301 (__builtin_ia32_subss_mask_round): New builtins.
26302 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
26303 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
26304 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
26306 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
26307 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
26309 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
26312 2017-05-02 Martin Jambor <mjambor@suse.cz>
26314 PR tree-optimization/78687
26315 * tree-sra.c (access): New field parent.
26316 (process_subtree_disqualification): New function.
26317 (disqualify_candidate): Call it.
26318 (build_accesses_from_assign): Reset write flag if creating an
26320 (build_access_subtree): Fill in parent field and also prpagate
26321 down grp_write flag.
26322 (create_artificial_child_access): New parameter set_grp_write, set
26323 grp_write to its value.
26324 (propagate_subaccesses_across_link): Also propagate grp_write flag
26326 (propagate_all_subaccesses): Push the closest parent back to work
26327 queue if add_access_to_work_queue returned true.
26329 2017-05-02 Richard Biener <rguenther@suse.de>
26331 * common.opt (fstrict-overflow): Alias negative to fwrapv.
26332 * doc/invoke.texi (fstrict-overflow): Remove all traces of
26333 -fstrict-overflow documentation.
26334 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
26335 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
26336 flag_strict_overflow.
26337 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
26338 * lto-opts.c (lto_write_options): Do not stream it.
26339 * lto-wrapper.c (merge_and_complain): Do not handle it.
26340 * opts.c (default_options_table): Do not set -fstrict-overflow.
26341 (finish_options): Likewise do not clear it when sanitizing.
26342 * simplify-rtx.c (simplify_const_relational_operation): Do not
26343 test flag_strict_overflow.
26345 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26347 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
26348 using enabled attribute.
26349 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
26350 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
26351 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
26352 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
26353 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
26354 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
26355 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
26356 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
26357 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
26358 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
26360 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
26362 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
26364 2017-05-02 Richard Biener <rguenther@suse.de>
26366 PR tree-optimization/80591
26368 2017-04-10 Richard Biener <rguenther@suse.de>
26370 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26373 2017-05-02 Richard Biener <rguenther@suse.de>
26375 PR tree-optimization/80549
26376 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
26377 (cleanup_tree_cfg_noloop): Create forwarders to known loop
26378 headers if they do not have a preheader.
26380 2017-05-02 Martin Liska <mliska@suse.cz>
26383 * common.opt: Fix typo.
26384 * doc/invoke.texi: Likewise.
26386 2017-05-01 Jan Beulich <jbeulich@suse.com>
26388 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
26389 swapping, add (x,x,m,x,n) alternative.
26391 2017-05-01 Nathan Sidwell <nathan@acm.org>
26393 * calls.c (combine_pending_stack_adjustment_and_call): Remove
26394 unnecessary unadjusted_alignment check.
26396 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
26399 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
26401 * gimplify.c (gimplify_cilk_detach): New function.
26402 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
26403 * tree-core.h: Document EXPR_CILK_SPAWN.
26404 * tree.h (EXPR_CILK_SPAWN): Define.
26406 2017-05-01 David Malcolm <dmalcolm@redhat.com>
26408 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
26409 to use new fixit_hint representation, using the "replace" logic.
26410 (get_line_span_for_fixit_hint): Likewise.
26411 (layout::print_any_fixits): Likewise.
26412 (selftest::test_one_liner_many_fixits): Rename to...
26413 (selftest::test_one_liner_many_fixits_1): ...this, and update
26414 comment and expected output to reflect that the multiple fix-it
26415 hints are now consolidated into one insertion.
26416 (selftest::test_one_liner_many_fixits_2): New test.
26417 (selftest::test_diagnostic_show_locus_one_liner): Update for
26419 (selftest::test_fixit_consolidation): Update for fix-it API
26421 * diagnostic.c (print_parseable_fixits): Likewise.
26422 * edit-context.c (edited_line::m_line_events): Convert from
26423 auto_vec <line_event *> to auto_vec <line_event>.
26424 (class line_event): Convert from abstract base class to a concrete
26425 class, taking over the role of replace_event.
26426 (class insert_event): Delete.
26427 (class replace_event): Rename to class line_event. Convert to
26429 (edit_context::add_fixits): Reimplement.
26430 (edit_context::apply_insert): Delete.
26431 (edit_context::apply_replace): Rename to...
26432 (edit_context::apply_fixit): ...this. Convert to half-open range.
26433 (edited_file::apply_insert): Delete.
26434 (edited_file::apply_replace): Rename to...
26435 (edited_file::apply_fixit): ...this.
26436 (edited_line::~edited_line): Drop deletion of events.
26437 (edited_line::apply_insert): Delete.
26438 (edited_line::apply_replace): Rename to...
26439 (edited_line::apply_fixit): ...this. Convert to half-open range.
26440 Update for change to type of m_line_events.
26441 * edit-context.h (edit_context::apply_insert): Delete.
26442 (edit_context::apply_replace): Rename to...
26443 (edit_context::apply_fixit): ...this.
26445 2017-05-01 Martin Sebor <msebor@redhat.com>
26447 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
26450 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
26453 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
26454 __get_cpuid_max returns 0.
26455 (__get_cpuid_count): Ditto.
26457 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
26459 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
26460 replacement expression is another instance of one of its arguments.
26462 2017-05-01 Jakub Jelinek <jakub@redhat.com>
26465 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
26466 check for stack push/pop autoinc.
26467 * config/i386/i386.c (ix86_agi_dependent): Return false
26468 if the only reason why modified_in_p returned true is that
26469 addr is SP based and set_insn is a push or pop.
26471 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26473 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
26476 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
26479 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
26480 (account_size_time): Use two predicates - exec_pred and
26482 (evaluate_conditions_for_known_args): Compute both clause and
26484 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
26485 (inline_summary_t::duplicate): Update.
26486 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
26488 (compute_inline_parameters): Likewise.
26489 (estimate_edge_size_and_time): Update caluclation of time.
26490 (estimate_node_size_and_time): Compute both time and nonspecialized
26492 (estimate_ipcp_clone_size_and_time): Update.
26493 (inline_merge_summary): Update.
26494 (do_estimate_edge_time): Update.
26495 (do_estimate_edge_size): Update.
26496 (do_estimate_edge_hints): Update.
26497 (inline_read_section, inline_write_summary): Stream both new predicates.
26498 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
26500 (compute_inlined_call_time): Cleanup.
26501 (big_speedup_p): Update.
26502 (edge_badness): Update.
26503 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
26504 (size_time_entry): Replace predicate by exec_predicate and
26505 nonconst_predicate.
26506 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
26507 (estimate_edge_time): Return also nonspec_time.
26508 (reset_edge_growth_cache): Update.
26510 2017-04-29 Jakub Jelinek <jakub@redhat.com>
26512 PR rtl-optimization/80491
26513 * ifcvt.c (noce_process_if_block): When looking for x setter
26514 with missing else_bb, don't check only the insn right before
26515 cond_earliest, but look for the last insn that x is modified in
26516 within the same bb.
26518 PR rtl-optimization/80491
26519 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
26521 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
26523 PR tree-optimization/80487
26524 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
26526 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26528 PR tree-optimization/79697
26529 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
26530 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
26531 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
26533 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
26534 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
26536 2017-04-28 Martin Sebor <msebor@redhat.com>
26538 PR tree-optimization/80523
26539 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
26540 (init_target_to_host_charmap, target_to_host, target_strtol10): New
26542 (maybe_warn, format_directive, parse_directive): Use new functions.
26543 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
26545 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
26547 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
26549 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
26551 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
26552 target_header_dir): Set correctly.
26553 * configure: Regenerated.
26554 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26555 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26556 instead of SYSTEM_HEADER_DIR.
26558 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
26560 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
26561 (estimate_local_effects): Likewise.
26562 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
26563 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
26564 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
26565 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
26566 do_estimate_edge_time, estimate_edge_time): Likewise.
26567 * ipa-inline-analysis.c (estimate_node_size_and_time,
26568 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
26569 (estimate_time_after_inlining): Remove.
26571 2017-04-28 Martin Liska <mliska@suse.cz>
26573 * doc/gcov.texi: Enhance documentation of gcov.
26575 2017-04-28 Martin Liska <mliska@suse.cz>
26577 * doc/gcov.texi: Sort options in alphabetic order.
26578 * doc/gcov-dump.texi: Likewise.
26579 * doc/gcov-tool.texi: Likewise.
26580 * gcov.c (print_usage): Likewise.
26581 * gcov-dump.c (print_usage): Likewise.
26582 * gcov-tool.c (print_merge_usage_message): Likewise.
26583 (print_rewrite_usage_message): Likewise.
26584 (print_overlap_usage_message): Likewise.
26586 2017-04-28 Martin Liska <mliska@suse.cz>
26588 PR gcov-profile/53915
26589 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
26591 2017-04-28 Martin Liska <mliska@suse.cz>
26593 PR gcov-profile/79891
26594 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
26595 is marked by compiler as living on a line.
26596 (get_cycles_count): Remove usage of the union.
26597 (output_intermediate_file): Likewise.
26598 (find_source): Fix GNU coding style.
26599 (accumulate_line_counts): Remove old non-all block mode.
26600 (output_lines): Remove usage of the union.
26601 * profile.c (output_location): Include all BBs, even if
26602 belonging to a same line (and file) as a previous BB.
26604 2017-04-28 Martin Liska <mliska@suse.cz>
26606 * gcov.c (process_args): Handle new argument 'w'.
26607 (read_graph_file): Assign ID to BBs.
26608 (output_branch_count): Display BB # if verbose flag is set.
26609 (output_lines): Likewise for arcs.
26610 (print_usage): Add '--verbose' option help.
26611 * doc/gcov.texi: Document --verbose (-w) option.
26613 2017-04-28 Martin Liska <mliska@suse.cz>
26615 * gcov.c (struct block_location_info): New struct.
26616 (process_file): Fill up the new structure.
26617 (read_graph_file): Replace usage of encoding by the newly added
26619 (add_line_counts): Likewise.
26620 (accumulate_line_counts): Remove usage of the union.
26621 (function_info::function_info): New function.
26622 (function_info::~function_info): Likewise.
26623 (process_file): Call delete instead of release_function.
26624 (release_function): Release the function.
26625 (release_structures): Call delete instead of release_function.
26626 (solve_flow_graph): Replace usage of num_blocks.
26627 (find_exception_blocks): Likewise.
26628 (output_lines): Fix GNU coding style.
26630 2017-04-28 Martin Liska <mliska@suse.cz>
26633 * coverage.c (coverage_remove_note_file): New function.
26634 * coverage.h: Declare the function.
26635 * toplev.c (finalize): Clean if an error has been seen.
26637 2017-04-28 Martin Liska <mliska@suse.cz>
26639 PR gcov-profile/80031
26640 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
26641 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
26642 * gcov.c (read_graph_file): Read just number of blocks.
26643 * profile.c (branch_prob): Do not stream 0 flags per a basic
26646 2017-04-28 Martin Liska <mliska@suse.cz>
26648 * gcov-dump.c (tag_*): Add new argument to declarations.
26649 (dump_gcov_file): Likewise.
26650 (tag_blocks): Add and use new argument depth.
26651 (tag_arcs): Likewise.
26652 (tag_lines): Likewise.
26653 (tag_counters): Likewise.
26654 (tag_summary): Likewise.
26655 (dump_working_sets): Use depth to do a proper indentation.
26657 2017-04-28 Jakub Jelinek <jakub@redhat.com>
26660 * cgraph.h (symtab_node::debug_symtab): No longer inline.
26661 * symtab.c (symtab_node::debug_symtab): Move definition here.
26663 2017-04-28 Richard Biener <rguenther@suse.de>
26665 * lto-streamer.h (LTO_major_version): Bump to 7.
26667 2017-04-28 Richard Biener <rguenther@suse.de>
26669 * tree-vrp.c (assert_info): New struct.
26670 (add_assert_info): New helper.
26671 (register_edge_assert_for_2): Refactor to add asserts to a vector
26673 (register_edge_assert_for_1): Likewise.
26674 (register_edge_assert_for): Likewise.
26675 (finish_register_edge_assert_for): New helper actually registering
26676 asserts where live on edge.
26677 (find_conditional_asserts): Adjust.
26678 (find_switch_asserts): Likewise.
26679 (evrp_dom_walker::try_find_new_range): Generalize.
26680 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
26682 2017-04-27 Marek Polacek <polacek@redhat.com>
26685 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
26686 arg10 and arg11 to itype.
26688 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
26690 * doc/extend.texi (Object Size Checking): Improve grammar.
26692 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
26695 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
26696 that the logic for permitting reciprocal estimates matches that
26699 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26702 * tree.c (type_cache_hasher::equal): Only compare
26703 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
26704 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
26705 non-aggregate element types.
26706 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
26707 about the flag on ARRAY_TYPEs in the comment, formatting fix.
26709 2017-04-27 Richard Biener <rguenther@suse.de>
26711 PR middle-end/80533
26712 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
26713 stripping ARRAY_REFs from MEM_EXPR make sure we're not
26714 keeping a reference to a trailing array.
26716 2017-04-27 Richard Biener <rguenther@suse.de>
26718 PR middle-end/80539
26719 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
26720 being in loop-closed SSA form conservatively.
26721 (chrec_fold_multiply_poly_poly): Likewise.
26723 2017-04-27 Tamar Christina <tamar.christina@arm.com>
26725 PR middle-end/79665
26726 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
26727 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
26729 2017-04-27 Jakub Jelinek <jakub@redhat.com>
26732 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
26733 (aarch64_function_arg_alignment): Return unsigned int again, but still
26734 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
26735 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26736 Don't emit -Wpsabi note.
26737 (aarch64_function_arg_boundary): Likewise.
26738 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26741 2017-04-26 Nathan Sidwell <nathan@acm.org>
26743 * tree.h (crc32_unsigned_n): Declare.
26744 (crc32_unsigned, crc32_unsigned): Make inline.
26745 * tree.c (crc32_unsigned_bits): Replace with ...
26746 (crc32_unsigned_n): ... this.
26747 (crc32_unsigned, crc32_byte): Remove.
26748 (crc32_string): Remove unnecessary braces.
26750 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26752 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
26753 * ipa-inline-analysis.c (MAX_TIME): Remove.
26754 (account_size_time): Use sreal for time.
26755 (dump_inline_summary): Update.
26756 (estimate_function_body_sizes): Update.
26757 (estimate_edge_size_and_time): Update.
26758 (estimate_calls_size_and_time): Update.
26759 (estimate_node_size_and_time): Update.
26760 (inline_merge_summary): Update.
26761 (inline_update_overall_summary): Update.
26762 (estimate_time_after_inlining): Update.
26763 (inline_read_section): Update.
26764 (inline_write_summary): Update.
26765 * ipa-inline.c (compute_uninlined_call_time): Update.
26766 (compute_inlined_call_time): Update.
26767 (recursive_inlining): Update.
26768 (inline_small_functions): Update.
26769 (dump_overall_stats): Update.
26770 * ipa-inline.h: Include sreal.h.
26771 (size_time_entry): Turn time to sreal.
26772 (inline_summary): Turn self_time nad time to sreal.
26774 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
26776 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
26778 (sreal::stream_out, sreal::stream_in): New.
26779 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
26781 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26783 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
26786 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
26789 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
26791 (dimode_scalar_chain::compute_convert_gain): Ditto.
26792 (dimode_scalar_chain::make_vector_copies): Ditto.
26793 (dimode_scalar_chain::convert_reg): Ditto.
26794 (dimode_scalar_chain::convert_insn): Ditto.
26795 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
26796 (VI248_AVX512BW_1): New mode iterator.
26797 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
26798 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
26801 2017-04-25 Martin Sebor <msebor@redhat.com>
26803 PR tree-optimization/80497
26804 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
26805 constants are representable in HOST_WIDE_INT.
26806 (parse_directive): Ditto.
26808 2017-04-25 Martin Sebor <msebor@redhat.com>
26811 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
26812 (new_zero_array): Adjust signature.
26813 (dom_info::dom_init): Used unsigned rather that size_t.
26814 (dom_info::dom_info): Same.
26816 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26817 Jakub Jelinek <jakub@redhat.com>
26820 * config/arm/arm.c: Include gimple.h.
26821 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26822 returns negative, increment ncrn only if it returned positive.
26823 (arm_needs_doubleword_align): Return int instead of bool,
26824 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
26825 members, but if there is any such non-FIELD_DECL
26826 > PARM_BOUNDARY aligned decl, return -1 instead of false.
26827 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
26828 returns negative, increment nregs only if it returned positive.
26829 (arm_setup_incoming_varargs): Likewise.
26830 (arm_function_arg_boundary): Emit -Wpsabi note if
26831 arm_needs_doubleword_align returns negative, return
26832 DOUBLEWORD_ALIGNMENT only if it returned positive.
26834 2017-04-25 Marek Polacek <polacek@redhat.com>
26837 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26838 first argument to type.
26840 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
26843 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
26844 type checks to test for compatibility instead of equality.
26846 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26847 Jakub Jelinek <jakub@redhat.com>
26850 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
26852 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
26853 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
26854 the alignment computation, but return their maximum in warn_alignment.
26855 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
26856 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
26858 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
26859 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
26862 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26864 * config/arc/simdext.md (dmpyh): Fix typo.
26866 2017-04-25 Richard Biener <rguenther@suse.de>
26868 PR tree-optimization/80492
26869 * alias.c (compare_base_decls): Handle registers with asm
26870 specification conservatively.
26871 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
26872 compare_base_decls returning dont-know properly.
26874 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26876 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
26877 (legitimate_offset_address_p): New function.
26878 (arc_legitimate_address_p): Use above function.
26880 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26882 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
26884 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26886 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
26887 ACCH registers whenever they are available.
26889 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26891 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
26892 double regs fix when not used.
26894 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26896 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
26898 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
26899 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
26901 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26903 * config/arc/arc.c (arc_output_addsi): Check for h-register class
26904 when emitting short ADD instructions.
26906 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
26908 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
26910 (cmpsi_cc_c_insn): Likewise.
26911 (cbranchsi4_scratch): Compute proper instruction length using
26912 compact_hreg_operand.
26913 * config/arc/predicates.md (compact_hreg_operand): New predicate.
26915 2017-04-25 Richard Biener <rguenther@suse.de>
26917 PR middle-end/80509
26918 * passes.c (pass_manager::pass_manager): Initialize
26919 m_name_to_pass_map.
26921 2017-04-25 Richard Biener <rguenther@suse.de>
26923 PR tree-optimization/79201
26924 * tree-ssa-sink.c (statement_sink_location): Handle calls.
26926 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26929 * config/s390/vector.md: Split MEM->GPR vector moves for
26930 non-s_operand addresses.
26932 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26935 * config/s390/predicates.md (reload_const_wide_int_operand): New
26937 * config/s390/s390.md ("movti"): Remove d/P alternative.
26938 ("movti_bigconst"): New pattern definition.
26940 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26943 * s390-protos.h (s390_expand_cs_hqi): Removed.
26944 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
26945 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
26946 modes as well as CCZ1mode and CCZmode.
26947 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
26948 signature of s390_emit_compare_and_swap.
26949 (s390_expand_cs_hqi): Likewise, make static.
26950 (s390_expand_cs_tdsi): Generate an explicit compare before trying
26951 compare-and-swap, in some cases.
26952 (s390_expand_cs): Wrapper function.
26953 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
26955 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
26956 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
26957 patterns for small and large integers. Forbid symref memory operands.
26958 Move expander to s390.c. Require cc register.
26959 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
26960 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
26961 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
26962 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
26963 symref memory operands. Remove CC mode and call s390_match_ccmode
26965 ("atomic_exchange<mode>"): Allow and implement all integer modes.
26967 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26969 * config/s390/s390.md (define_peephole2): New peephole to help
26970 combining the load-and-test pattern with volatile memory.
26972 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
26974 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
26975 with CCZmode for TARGET_Z196.
26977 2017-04-25 Jakub Jelinek <jakub@redhat.com>
26979 PR rtl-optimization/80501
26980 * combine.c (make_compound_operation_int): Set subreg_code to SET
26981 even for AND with mask of the sign bit of mode.
26983 PR rtl-optimization/80500
26984 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
26985 sum's initial value.
26987 2017-04-25 Julian Brown <julian@codesourcery.com>
26988 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26990 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
26992 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
26994 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
26996 2017-04-25 Julian Brown <julian@codesourcery.com>
26997 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26999 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
27000 (thunderx2t99_sha): New Reservation.
27002 2017-04-25 Julian Brown <julian@codesourcery.com>
27003 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27005 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
27006 type for 1-element load.
27008 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
27010 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
27012 2017-04-24 Martin Jambor <mjambor@suse.cz>
27014 PR tree-optimization/80293
27015 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
27016 char arrays not totally scalarizable if it is false.
27017 (analyze_all_variable_accesses): Pass correct value in the new
27018 parameter. Add a statistics counter.
27020 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
27022 PR middle-end/79931
27023 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
27025 2017-04-24 Richard Biener <rguenther@suse.de>
27027 PR tree-optimization/80494
27028 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
27029 out for complex types.
27031 2017-04-24 Richard Biener <rguenther@suse.de>
27033 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
27034 * tree-ssa-sccvn.c (print_scc): Print SCC size.
27035 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
27036 (DFS): Adjust and never fail.
27037 (sccvn_dom_walker::fail): Remove.
27038 (sccvn_dom_walker::before_dom_children): Adjust.
27039 (run_scc_vn): Likewise and never fail.
27040 * tree-ssa-pre.c (pass_pre::execute): Adjust.
27041 (pass_fre::execute): Likewise.
27043 2017-04-24 Richard Biener <rguenther@suse.de>
27045 PR tree-optimization/79725
27046 * tree-ssa-sink.c (statement_sink_location): Return whether
27047 failure reason was zero uses. Move that check later.
27048 (sink_code_in_bb): Deal with zero uses by removing the stmt
27051 2017-04-24 Richard Biener <rguenther@suse.de>
27054 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
27055 pointer-based references.
27057 2017-04-24 Richard Biener <rguenther@suse.de>
27060 * pass_manager.h (pass_manager::operator new): Remove.
27061 (pass_manager::operator delete): Likewise.
27062 * passes.c (pass_manager::operator new): Remove.
27063 (pass_manager::operator delete): Likewise.
27064 (pass_manager::pass_manager): Zero individual pass members.
27066 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
27069 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
27070 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
27071 Check "XEXP (src, 1)" operand here.
27072 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
27073 Check "XEXP (src, 1)" operand here.
27074 (dimode_scalar_chain::make_vector_copies): Detect count register
27075 of a shift instruction. Zero extend count register from QImode
27076 to DImode to satisfy vector shift pattern count operand predicate.
27077 Substitute vector shift count operand with a DImode copy.
27078 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
27081 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
27083 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
27084 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27085 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
27086 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
27087 (UNSPEC_NOREX_MEM): Remove definition.
27089 2017-04-21 Richard Biener <rguenther@suse.de>
27091 PR tree-optimization/79547
27092 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
27093 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
27094 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
27095 without any constraints.
27097 2017-04-21 Richard Biener <rguenther@suse.de>
27099 PR tree-optimization/78847
27100 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
27102 2017-04-21 Richard Biener <rguenther@suse.de>
27104 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
27105 (build_distinct_type_copy): Likewise.
27106 (build_variant_type_copy): Likewise.
27107 * tree.c (build_qualified_type): Pass down mem-stat info.
27108 (build_distinct_type_copy): Likewise.
27109 (build_variant_type_copy): Likewise.
27111 2017-04-21 Richard Biener <rguenther@suse.de>
27113 PR tree-optimization/80237
27114 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
27116 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
27117 for a simplified result.
27119 2016-04-21 Richard Biener <rguenther@suse.de>
27121 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
27122 sth as strict as a simple_iv but a chrec without symbols and an
27123 operand defined in the loop we are peeling (and not some subloop).
27124 (propagate_constants_for_unrolling): Propagate all constants.
27126 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27129 * config/i386/i386.c (print_reg): Remove assert for disalowed
27130 regno values, call output_operand_lossage instead.
27132 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
27135 * config/i386/constraints.md (Yc): New register constraint.
27136 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
27137 Use Yc constraint for alternative 2 of operand 0. Remove
27138 preferred_for_speed attribute.
27140 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27142 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
27143 lastprivate clauses in SIMT case.
27145 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
27147 * doc/invoke.texi (-Wextra-semi): Document new warning option.
27149 2017-04-20 Richard Biener <rguenther@suse.de>
27151 PR tree-optimization/57796
27152 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
27153 as N scalar stores.
27154 (vect_model_load_cost): Cost gathers as N scalar loads.
27156 2017-04-20 Richard Biener <rguenther@suse.de>
27158 * ggc-page.c (ggc_allocated_p): Rename to ...
27159 (safe_lookup_page_table_entry): ... this and return the lookup
27161 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
27163 2017-04-20 Richard Biener <rguenther@suse.de>
27165 PR tree-optimization/80453
27166 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
27167 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
27168 from the conditions.
27169 (vn_phi_eq): Pass them down.
27170 (vn_phi_lookup): Record them.
27171 (vn_phi_insert): Likewise.
27173 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
27175 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
27176 uninitialized variable warning to avoid buffer overrun.
27178 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
27181 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
27182 is suppressed for '{ 0 }' in C.
27184 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27186 * BASE-VER: Set to 8.0.0.
27188 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
27190 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
27191 priority .init_array and .fini_array section with SECTION_NOTYPE
27194 2017-04-20 Jakub Jelinek <jakub@redhat.com>
27196 PR middle-end/80423
27197 * tree.h (build_array_type): Add typeless_storage default argument.
27198 * tree.c (type_cache_hasher::equal): Also compare
27199 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
27200 (build_array_type): Add typeless_storage argument, set
27201 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
27203 (build_nonshared_array_type): Adjust build_array_type_1 caller.
27204 (build_array_type): Likewise. Add typeless_storage argument.
27206 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27207 Jakub Jelinek <jakub@redhat.com>
27209 PR tree-optimization/80426
27210 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
27211 operation on symbolic operands, also compute the overflow for the
27212 invariant part when the operation degenerates into a negation.
27214 2017-04-19 Jakub Jelinek <jakub@redhat.com>
27217 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
27218 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
27221 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
27223 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
27226 * config/avr/avr.c (tree.h): Include it.
27227 (cgraph.h): Include it.
27228 (avr_encode_section_info): Don't warn for uninitialized progmem
27229 variable if it's just an alias.
27231 2017-04-19 Richard Biener <rguenther@suse.de>
27234 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
27235 when needed by AutoPGO.
27237 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
27240 * doc/lto.texi: Remove an extra 'that'.
27242 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
27244 PR rtl-optimization/80429
27245 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
27246 are only used in debug insns.
27248 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
27249 Vladimir Makarov <vmakarov@redhat.com>
27251 * config/sparc/predicates.md (input_operand): Add comment. Return
27252 true for any memory operand when LRA is in progress.
27253 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
27255 2017-04-18 Jeff Law <law@redhat.com>
27258 * mips.md ({return,simple_return}_internal): Do not overwrite
27261 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27263 PR tree-optimization/80443
27264 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
27265 instead of adding 1, subtract -1 and similarly instead of subtracting
27268 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
27270 PR rtl-optimization/80357
27271 * haifa-sched.c (tmp_bitmap): New variable.
27272 (model_recompute): Handle duplicate use records.
27273 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
27274 (free_global_sched_pressure_data): Free it.
27276 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
27279 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
27280 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
27281 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
27282 instead of SYSTEM_HEADER_DIR.
27284 2017-04-18 Jeff Law <law@redhat.com>
27286 PR middle-end/80422
27287 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
27288 predecessors after walking up the insn chain.
27290 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27293 * dwarf2out.c (modified_type_die): Try harder not to emit internal
27294 sizetype type into debug info.
27296 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
27299 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
27300 unneeded test for TARGET_UPPER_REGS_SF.
27301 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
27303 2017-04-18 Jakub Jelinek <jakub@redhat.com>
27306 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
27307 instead of gsi_after_labels.
27309 2017-04-18 Jeff Law <law@redhat.com>
27311 * regcprop.c (maybe_mode_change): Avoid creating copies of the
27315 2017-04-13 Jeff Law <law@redhat.com>
27316 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27317 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27319 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
27322 * config/avr/avr.c (intl.h): Include it.
27323 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
27325 2017-04-18 Martin Liska <mliska@suse.cz>
27327 PR gcov-profile/78783
27328 * gcov-tool.c (gcov_output_files): Validate that destination
27329 file is either removed by the tool or by a user.
27331 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27332 Guy Benyei <guybe@mellanox.com>
27334 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
27335 block, and do not negate it, the stored id is already negative.
27337 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
27339 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
27341 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27344 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
27345 masks of options that should be turned off if the VSX vector
27346 options are turned off.
27347 (OTHER_P8_VECTOR_MASKS): Likewise.
27348 (OTHER_VSX_VECTOR_MASKS): Likewise.
27349 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
27350 rs6000_disable_incompatible_switches to validate no type switches
27352 (rs6000_incompatible_switch): New function to disallow turning on
27353 other vector options if -mno-vsx, -mno-power8-vector, or
27354 -mno-power9-vector are specified.
27356 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27358 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
27360 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27362 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
27363 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
27364 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
27365 (ARG_POINTER_CFA_OFFSET): Likewise.
27367 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
27369 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
27370 conditions to take advantage of various optimizations.
27372 2017-04-13 Jeff Law <law@redhat.com>
27374 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
27375 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
27376 (zero_extendsidi2_dext): Likewise.
27378 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27381 * fold-const.c (fold_ternary_loc): Revert
27382 use op0 instead of fold_convert_loc (loc, type, arg0) part of
27385 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
27387 PR rtl-optimization/80343
27388 * lra-remat.c (update_scratch_ops): Assign original hard reg to
27389 new scratch pseudo.
27391 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
27394 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
27395 to ubsan_encode_value.
27397 2017-04-13 Jeff Law <law@redhat.com>
27399 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
27400 appearing in DEBUG_INSNs.
27402 2017-04-13 Martin Liska <mliska@suse.cz>
27404 PR gcov-profile/80413
27405 * gcov-io.c (gcov_write_string): Copy to buffer just when
27406 allocated size is greater than zero.
27408 2017-04-13 Jakub Jelinek <jakub@redhat.com>
27411 * dwarf2out.c (decls_for_scope): Ignore declarations of
27412 current_function_decl in BLOCK_NONLOCALIZED_VARS.
27414 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
27417 * ipa-visibility.c (non_local_p): Fix typos.
27418 (localize_node): When localizing symbol in same comdat group,
27419 dissolve the group only when we know external symbols are going
27421 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
27423 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27425 PR tree-optimization/79390
27426 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
27427 order does not result in usable sequence, retry with reversed operand
27433 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
27434 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
27435 op0 instead of fold_convert_loc (loc, type, arg0).
27437 2017-04-12 Jeff Law <law@redhat.com>
27439 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
27440 has a delay slot in the generated code.
27442 * config/cris/cris.md (cris_preferred_reload_class): Return
27443 GENNONACR_REGS rather than GENERAL_REGS.
27445 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27448 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
27449 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
27450 signedness of the result type.
27452 2017-04-12 Richard Biener <rguenther@suse.de>
27453 Jeff Law <law@redhat.com>
27455 PR tree-optimization/80359
27456 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
27457 trim stores to TARGET_MEM_REFs.
27459 2017-04-12 Richard Biener <rguenther@suse.de>
27461 PR tree-optimization/79390
27462 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
27463 threading case even more.
27465 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
27468 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
27469 for quad_address_p for TImode, instead of just not indexed_address.
27471 2017-04-12 Richard Biener <rguenther@suse.de>
27472 Bernd Edlinger <bernd.edlinger@hotmail.de>
27474 PR middle-end/79671
27475 * alias.c (component_uses_parent_alias_set_from): Handle
27476 TYPE_TYPELESS_STORAGE.
27477 (get_alias_set): Likewise.
27478 * tree-core.h (tree_type_common): Add typeless_storage flag.
27479 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
27480 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
27481 for types containing members with TYPE_TYPELESS_STORAGE.
27482 (place_field): Likewise.
27483 (layout_type): Likewise for ARRAY_TYPE.
27484 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
27485 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
27486 TYPE_TYPELESS_STORAGE.
27487 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
27489 2017-04-12 Jakub Jelinek <jakub@redhat.com>
27492 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
27493 first argument to type.
27495 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27499 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
27500 CONST0_RTX (mode) rather than const0_rtx where appropriate.
27501 (rs6000_expand_binop_builtin): Likewise.
27502 (rs6000_expand_ternop_builtin): Likewise; also add missing
27503 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
27504 vshasigma built-ins.
27505 * doc/extend.texi: Document that vec_xxpermdi's third argument
27506 must be a constant.
27508 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
27510 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
27511 Use shift_const cost parameter when calculating gain of STV shifts.
27513 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
27515 PR rtl-optimization/70478
27516 * lra-constraints.c (process_alt_operands): Check memory for
27517 disfavoring memory insn operand.
27519 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27521 PR middle-end/80100
27522 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
27523 left shift in unsigned HOST_WIDE_INT type.
27525 PR rtl-optimization/80385
27526 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
27527 (not (neg X)) into (plus X -1) for complex or non-integral modes.
27530 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
27531 if they have any depend clauses.
27533 2017-04-11 Martin Liska <mliska@suse.cz>
27536 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
27537 * ipa-split.c (split_function): Create a local comdat symbol
27538 if caller is in a comdat group.
27540 2017-04-11 Martin Liska <mliska@suse.cz>
27543 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
27546 2017-04-11 Martin Sebor <msebor@redhat.com>
27548 PR middle-end/80364
27549 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
27550 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
27552 (directive::set_width, directive::set_precision, format_character):
27554 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
27557 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
27560 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
27561 conflict, set target->arch_name instead of target->cpu_name.
27563 2017-04-11 Richard Biener <rguenther@suse.de>
27565 PR tree-optimization/80374
27566 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
27567 build_zero_cst, remove fold_convertible_p check again.
27569 2017-04-11 Martin Liska <mliska@suse.cz>
27572 * ubsan.c (instrument_object_size): Do not instrument register
27575 2017-04-11 Jakub Jelinek <jakub@redhat.com>
27578 * config/i386/i386-builtin-types.def
27579 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
27580 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
27581 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
27582 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
27583 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
27584 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
27585 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
27586 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
27587 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
27588 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
27589 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
27590 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
27591 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
27592 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
27593 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
27594 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
27595 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
27596 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
27597 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
27598 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
27599 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
27600 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
27601 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
27602 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
27603 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
27604 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
27605 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
27606 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
27607 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
27608 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
27609 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
27610 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
27611 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
27612 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
27613 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
27614 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
27615 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
27616 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
27617 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
27618 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
27619 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
27620 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
27621 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
27622 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
27623 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
27624 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
27626 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
27627 flag to second_arg_count, handle 4 argument function type _COUNT
27628 aliases, handle second_arg_count on second argument rather than last.
27630 2017-04-10 Jeff Law <law@redhat.com>
27632 PR tree-optimization/80374
27633 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
27634 record anything if we can not convert integer_zero_node to the
27637 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27640 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27641 Enhance special handling given to the TARGET_P9_MINMAX option in
27642 relation to certain other options.
27644 2017-04-10 Bin Cheng <bin.cheng@arm.com>
27646 PR tree-optimization/80153
27647 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
27648 remove POINTER_PLUS_EXPR's base part directly, rather than through
27651 2017-04-10 Richard Biener <rguenther@suse.de>
27652 Bin Cheng <bin.cheng@arm.com>
27654 PR tree-optimization/80153
27655 * tree-affine.c (aff_combination_to_tree): Get base pointer from
27656 the first element of pointer type aff_tree. Build result expr in
27658 (add_elt_to_tree): Convert to type unconditionally. Remove other
27659 fold_convert calls.
27660 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
27661 (rewrite_use_nonlinear_expr): Check invariant using iv information.
27663 2017-04-10 Richard Biener <rguenther@suse.de>
27665 * tree-ssa-structalias.c (find_func_aliases): Properly handle
27668 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
27670 PR rtl-optimization/70478
27671 * lra-constraints.c (curr_small_class_check): New.
27672 (update_and_check_small_class_inputs): New.
27673 (process_alt_operands): Update curr_small_class_check. Disfavor
27674 alternative insn memory operands. Check available regs for small
27677 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
27680 * config/mips/mips.opt (-mvirt): Update description.
27681 * doc/invoke.texi (-mvirt): Likewise.
27683 2017-04-10 Richard Biener <rguenther@suse.de>
27685 PR middle-end/80362
27686 * fold-const.c (fold_binary_loc): Look at unstripped ops when
27687 looking for NEGATE_EXPR in -A / -B to A / B folding.
27689 2017-04-10 Martin Liska <mliska@suse.cz>
27691 PR gcov-profile/80224
27692 * gcov.c (print_usage): Fix usage string.
27693 (get_gcov_intermediate_filename): Remove.
27694 (output_gcov_file): Use both for normal and intermediate format.
27695 (generate_results): Do not initialize special file for
27696 intermediate format.
27698 2017-04-10 Richard Biener <rguenther@suse.de>
27700 PR tree-optimization/80304
27701 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
27704 2017-04-10 Nathan Sidwell <nathan@acm.org>
27707 * config/rs6000/rs6000.c (rs6000_vector_type): New.
27708 (rs6000_init_builtins): Use it.
27710 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27712 * config/arm/arm.md (<mrc>): Add mode to SET source.
27713 (<mrrc>): Likewise.
27715 2017-04-10 Richard Biener <rguenther@suse.de>
27717 PR middle-end/80344
27718 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
27720 2017-04-10 Jakub Jelinek <jakub@redhat.com>
27723 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
27724 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
27725 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
27726 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
27727 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
27728 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
27729 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
27730 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
27731 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
27732 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
27733 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
27734 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
27735 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
27736 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
27737 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
27738 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
27739 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
27740 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
27741 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
27742 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
27743 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
27744 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
27745 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
27747 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
27749 PR rtl-optimization/70478
27750 * lra-constraints.c: Reverse the last patch.
27752 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
27754 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
27755 Add comment for WCHAR_T.
27757 2017-04-08 Martin Liska <mliska@suse.cz>
27760 2017-04-07 Martin Liska <mliska@suse.cz>
27763 * ipa-split.c (split_function): Add function part to a same comdat
27766 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27769 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
27771 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
27773 * rs6000/rs6000.c (vec_load_pendulum): Rename...
27774 (vec_pairing): ...to this.
27775 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
27776 (rs6000_sched_init): Adjust for name change.
27777 (struct rs6000_sched_context): Likewise.
27778 (rs6000_init_sched_context): Likewise.
27779 (rs6000_set_sched_context): Likewise.
27781 2017-04-07 Jakub Jelinek <jakub@redhat.com>
27787 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
27789 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
27790 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
27791 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
27793 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
27795 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
27797 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27799 PR rtl-optimization/70703
27800 * ira-color.c (update_conflict_hard_regno_costs): Use
27801 int64_t instead of HOST_WIDE_INT.
27803 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
27805 PR rtl-optimization/70478
27806 * lra-constraints.c (process_alt_operands): Disfavor alternative
27807 insn memory operands.
27809 2017-04-07 Jeff Law <law@redhat.com>
27811 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
27812 CALL and NOTE_INSN_CALL_ARG_LOCATION.
27814 2017-04-07 Martin Liska <mliska@suse.cz>
27817 * config/aarch64/aarch64.c (aarch64_process_target_attr):
27818 Show error message instead of an ICE.
27820 2017-04-07 Martin Liska <mliska@suse.cz>
27823 * ipa-split.c (split_function): Add function part to a same comdat
27826 2017-04-07 Richard Biener <rguenther@suse.de>
27828 PR middle-end/80341
27829 * tree.c (get_unwidened): Also handle ! for_type case for
27831 * convert.c (do_narrow): Split out from ...
27832 (convert_to_integer_1): ... here. Do not pass final truncation
27833 type to get_unwidened for TRUNC_DIV_EXPR.
27835 2017-04-07 Richard Biener <rguenther@suse.de>
27837 * tree-affine.c (wide_int_ext_for_comb): Take type rather
27839 (aff_combination_const): Adjust.
27840 (aff_combination_scale): Likewise.
27841 (aff_combination_add_elt): Likewise.
27842 (aff_combination_add_cst): Likewise.
27843 (aff_combination_convert): Likewise.
27844 (add_elt_to_tree): Likewise. Remove unused argument.
27845 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
27847 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
27849 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
27851 * config/arm/arm.c (arm_default_short_enums): Use
27852 ARM_DEFAULT_SHORT_ENUMS.
27853 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
27855 2017-04-06 Jakub Jelinek <jakub@redhat.com>
27858 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
27859 members with redundant out-of-class redeclaration.
27861 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27864 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
27865 * config/i386/i386.md (*zero_extendsidi2):
27866 Add (?*x,*x) and (?*v,*v) alternatives.
27868 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
27871 * config/i386/i386.c (ix86_expand_builtin)
27872 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
27873 mode from insn data. Convert operands to insn operand mode.
27874 Copy operands that don't satisfy insn predicate to a register.
27876 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
27878 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
27881 2017-04-06 Richard Biener <rguenther@suse.de>
27883 PR tree-optimization/80334
27884 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
27885 preserve alignment of accesses.
27887 2017-04-06 Richard Biener <rguenther@suse.de>
27889 PR tree-optimization/80262
27890 * tree-sra.c (build_ref_for_offset): Preserve address-space
27892 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
27893 Drop useless address-space information on MEM_REF offsets.
27895 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
27897 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
27899 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27901 PR rtl-optimization/70703
27902 * ira-color.c (update_conflict_hard_regno_costs): Use
27903 HOST_WIDE_INT instead of long.
27905 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
27908 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
27909 not defined for x86_64 target. Add -mmmx target option when __SSE2__
27911 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
27912 for x86_64 target. Handle -m3dnowa option.
27914 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
27916 PR rtl-optimization/70703
27917 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
27918 (update_conflict_hard_regno_costs): Use long instead of unsigned
27919 arithmetic for cost calculation.
27921 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27922 Bernd Edlinger <bernd.edlinger@hotmail.de>
27925 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
27928 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
27931 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
27932 ptr_mode with Pmode throughout.
27933 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
27934 into probe_stack_range and use DImode.
27936 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
27939 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
27940 call_eh_return is true.
27942 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27944 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
27945 Initialize last_match_fntype_index.
27947 2017-04-05 Jakub Jelinek <jakub@redhat.com>
27950 * tree-nvr.c: Include internal-fn.h.
27951 (pass_return_slot::execute): Ignore internal calls without
27954 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27955 Richard Biener <rguenther@suse.de>
27958 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
27959 captures used multiple times, except for the last use.
27960 * generic-match-head.c: Include gimplify.h.
27962 2017-04-04 Jakub Jelinek <jakub@redhat.com>
27964 PR tree-optimization/79390
27965 * target.h (struct noce_if_info): Declare.
27966 * targhooks.h (default_noce_conversion_profitable_p): Declare.
27967 * target.def (noce_conversion_profitable_p): New target hook.
27968 * ifcvt.h (struct noce_if_info): New type, moved from ...
27969 * ifcvt.c (struct noce_if_info): ... here.
27970 (noce_conversion_profitable_p): Renamed to ...
27971 (default_noce_conversion_profitable_p): ... this. No longer
27973 (noce_try_store_flag_constants, noce_try_addcc,
27974 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
27975 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
27976 instead of noce_conversion_profitable_p.
27977 * config/i386/i386.c: Include ifcvt.h.
27978 (ix86_option_override_internal): Don't override
27979 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
27980 (ix86_noce_conversion_profitable_p): New function.
27981 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
27982 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
27983 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
27984 * doc/tm.texi: Regenerated.
27986 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27988 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
27991 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
27994 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
27995 instructions for small multiply cores.
27997 2017-04-04 Jeff Law <law@redhat.com>
27999 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
28001 (mips_expand_vec_perm_const): Initialize elements in orig_perm
28002 that are not set by the loop over the elements.
28004 2017-04-04 Jakub Jelinek <jakub@redhat.com>
28007 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
28008 int mode, convert_modes it to mode as unsigned, otherwise use
28009 lowpart_subreg to mode rather than SImode.
28010 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
28011 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
28012 Use DImode instead of SImode for the shift count operand.
28013 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
28016 2017-04-04 Richard Biener <rguenther@suse.de>
28018 PR middle-end/80281
28019 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
28020 arithmetic done for the negate or the plus. Simplify.
28021 (A - (-B) -> A + B): Likewise.
28022 * fold-const.c (split_tree): Make sure to not negate pointers.
28024 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
28026 PR rtl-optimization/60818
28027 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
28028 a compare of comparisons with the thing compared if this results
28029 in a different machine mode.
28031 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
28033 * alias.c (base_alias_check): Fix typo in comment.
28034 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
28035 * cgraphunit.c (symbol_table::compile): Likewise.
28036 * collect2.c (maybe_run_lto_and_relink): Likewise.
28037 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
28038 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
28039 * config/avr/avr.c (avr_map_op_t): Likewise.
28040 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
28041 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
28042 * config/epiphany/epiphany.md (movcc): Likewise.
28043 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
28044 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
28046 * config/mips/mips.c (mips_save_restore_reg): Likewise.
28047 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
28048 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
28049 * config/sh/sh.c (sh_rtx_costs): Likewise.
28050 * fold-const.c (fold_truth_andor): Likewise.
28051 * genautomata.c (collapse_flag): Likewise.
28052 * gengtype.h (struct type::u::s): Likewise.
28053 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
28054 * input.c (FORMAT_AMOUNT): Likewise.
28055 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
28056 (known_aggs_to_agg_replacement_list): Likewise.
28057 * ipa-inline-analysis.c: Likewise.
28058 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
28059 * ipa-polymorphic-call.c
28060 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
28061 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
28062 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
28064 * modulo-sched.c (apply_reg_moves): Likewise.
28065 * omp-expand.c (build_omp_regions_1): Likewise.
28066 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
28067 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
28068 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
28069 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
28070 * value-prof.c: Likewise.
28071 * var-tracking.c (val_reset): Likewise.
28073 2017-04-03 Richard Biener <rguenther@suse.de>
28075 PR tree-optimization/80275
28076 * fold-const.c (split_address_to_core_and_offset): Handle
28079 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
28081 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
28082 descriptors is at least equal to that of functions.
28084 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28086 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
28088 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
28091 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
28092 (mov<IMOD4:mode>): New expander.
28093 (*mov<IMOD4:mode>_internal): New insn and split pattern.
28095 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
28097 PR rtl-optimization/79405
28098 * fwprop.c (propagations_left): New variable.
28099 (forward_propagate_into): Decrement it.
28100 (fwprop_init): Initialize it.
28101 (fw_prop): If the variable has reached zero, stop propagating.
28102 (fwprop_addr): Ditto.
28104 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28107 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
28108 a FUNCTION_DECL, pass it as decl instead of origin to
28111 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
28113 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
28116 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
28119 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
28120 TARGET_VSX_SMALL_INTEGER.
28122 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28124 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
28125 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
28127 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
28129 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
28130 extraction from odd-numbered MSA register.
28132 2017-03-31 Jakub Jelinek <jakub@redhat.com>
28134 PR middle-end/80173
28135 * expmed.c (store_bit_field_1): Don't attempt to create
28136 a word subreg out of hard registers wider than word if they
28137 have HARD_REGNO_NREGS of 1 for their mode.
28139 PR middle-end/80163
28140 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
28141 conversions to integer types wider than word and pointer.
28144 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
28145 (rtx_equal_for_cselib_p): Pass 0 to it.
28146 * cselib.c (cselib_hasher::equal): Likewise.
28147 (rtx_equal_for_cselib_1): Add depth argument. If depth
28148 is 128, don't look up VALUE locs and punt. Increment
28149 depth in recursive calls when walking VALUE locs.
28151 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
28153 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
28154 (make_gcov_file_name): Use the canonical path name for generating
28156 (read_line): Fix handling of files with ascii null bytes.
28158 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
28160 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
28161 to initialise a vector register instead
28162 of using a const_int.
28164 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28166 PR translation/80189
28167 * gimplify.c (omp_default_clause): Use %qs instead of %s in
28168 diagnostic messages.
28170 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
28173 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
28174 (dfp_diex_<mode>): Update mode of operand 1.
28175 * doc/extend.texi (dxex, dxexq): Document change to return type.
28176 (diex, diexq): Document change to argument type.
28178 2017-03-30 Martin Jambor <mjambor@suse.cz>
28181 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
28182 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
28183 it reflects the signature changes performed at the callee side.
28184 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
28185 to cgraph_build_function_type_skip_args.
28186 (build_function_decl_skip_args): Adjust call to the above function.
28188 2017-03-30 Jakub Jelinek <jakub@redhat.com>
28191 * config/i386/sse.md
28192 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
28193 register as dest whenever it is a MEM not rtx_equal_p to the
28194 corresponding dup operand, and when forcing into reg move the
28195 reg into the memory afterwards.
28196 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
28197 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
28198 for the force_reg mode.
28199 (avx512vl_vextractf128<mode>): Use register as dest either
28200 always when a MEM, or when it is a MEM not rtx_equal_p to the
28201 corresponding dup operand, or even not when it is a CONST_VECTOR
28202 depending on the mode and lo vs. hi.
28203 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
28205 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
28206 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
28207 Likewise. Require that operands[2] is even.
28208 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
28209 Remove extraneous parens. Require that operands[2] is a multiple
28211 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
28212 operands[0] is a MEM if <mask_applied>, the predicates/constraints
28213 disallow memory then.
28215 2017-03-30 Richard Biener <rguenther@suse.de>
28217 PR tree-optimization/77498
28218 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
28219 to non-constants over backedges.
28221 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
28223 PR rtl-optimization/80233
28224 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
28225 as last_combined_insn. Do not test for BARRIER_P separately.
28227 2017-03-29 Andreas Schwab <schwab@suse.de>
28230 * calls.c (prepare_call_address): Convert funexp to Pmode before
28231 copying to temp reg.
28233 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28235 PR tree-optimization/80158
28236 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
28237 Handle possible future case of more than one alternate
28239 (replace_rhs_if_not_dup): Likewise.
28240 (replace_one_candidate): Likewise.
28242 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
28244 PR rtl-optimization/80193
28245 * ira.c (ira): Do not check allocation for LRA.
28247 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
28249 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
28250 (nvptx_output_simt_exit): Declare.
28251 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
28252 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
28253 (init_softstack_frame): Move initialization of crtl->is_leaf to...
28254 (nvptx_declare_function_name): ...here. Emit declaration of local
28255 memory space buffer for omp_simt_enter insn.
28256 (nvptx_output_unisimt_switch): New.
28257 (nvptx_output_softstack_switch): New.
28258 (nvptx_output_simt_enter): New.
28259 (nvptx_output_simt_exit): New.
28260 * config/nvptx/nvptx.h (struct machine_function): New fields
28261 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
28262 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
28263 (UNSPECV_SIMT_EXIT): Ditto.
28264 (omp_simt_enter_insn): New insn.
28265 (omp_simt_enter): New expansion.
28266 (omp_simt_exit): New insn.
28267 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
28269 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
28270 (expand_GOMP_SIMT_ENTER_ALLOC): New.
28271 (expand_GOMP_SIMT_EXIT): New.
28272 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
28273 (GOMP_SIMT_ENTER_ALLOC): Ditto.
28274 (GOMP_SIMT_EXIT): Ditto.
28275 * target-insns.def (omp_simt_enter): New insn.
28276 (omp_simt_exit): Ditto.
28277 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
28279 (lower_rec_simd_input_clauses): Implement SIMT privatization.
28280 (lower_rec_input_clauses): Likewise.
28281 (lower_lastprivate_clauses): Handle SIMT privatization.
28283 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
28284 (ompdevlow_adjust_simt_enter): New.
28285 (find_simtpriv_var_op): New.
28286 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
28287 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
28289 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
28290 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
28291 (copy_decl_for_dup_finish): Ditto.
28293 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
28295 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
28298 * config/i386/i386.c (ix86_option_override_internal): Always
28299 allow -mpreferred-stack-boundary=3 for 64-bit targets.
28301 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28303 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
28305 2017-03-28 Bin Cheng <bin.cheng@arm.com>
28307 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
28308 mark new edge's irreducible flag accordign to it.
28309 (vect_do_peeling): Check loop preheader edge's irreducible flag
28310 and pass it to function slpeel_add_loop_guard.
28312 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
28314 PR tree-optimization/80218
28315 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
28316 Update block frequencies and counts.
28318 2017-03-28 Richard Biener <rguenther@suse.de>
28320 PR tree-optimization/78644
28321 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
28322 of a simplification result we may not use it at all.
28324 2017-03-28 Richard Biener <rguenther@suse.de>
28327 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
28328 without arguments, generate default definition of a SSA name.
28330 2017-03-28 Richard Biener <rguenther@suse.de>
28332 PR middle-end/80222
28333 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
28334 TYPE_REF_CAN_ALIAS_ALL references.
28335 * fold-const.c (fold_indirect_ref_1): Likewise.
28337 2017-03-28 Martin Liska <mliska@suse.cz>
28340 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
28341 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
28343 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28344 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
28346 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
28347 (EXTRA_SPECS): Define.
28348 (SUBTARGET_EXTRA_SPECS): Likewise.
28349 (SUBTARGET_CPP_SPEC): Likewise.
28350 * config/arc/elf.h (EXTRA_SPECS): Renamed to
28351 SUBTARGET_EXTRA_SPECS.
28352 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
28354 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
28356 * config/arc/simdext.md (vst64_insn): Update pattern.
28357 (vld32wh_insn): Likewise.
28358 (vld32wl_insn): Likewise.
28359 (vld64_insn): Likewise.
28360 (vld32_insn): Likewise.
28362 2017-03-28 Marek Polacek <polacek@redhat.com>
28365 * fold-const.c (fold_comparison): Use protected_set_expr_location
28366 instead of SET_EXPR_LOCATION.
28368 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
28370 * tree.c (add_expr): Avoid name lookup warning.
28372 2017-03-27 Jeff Law <law@redhat.com>
28374 PR tree-optimization/80216
28375 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
28376 function name. Limit recursion depth.
28377 (record_temporary_equivalences): Corresponding changes.
28379 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
28381 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
28384 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28387 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
28389 * cfgcleanup.c (reg_note_cfa_p): New array.
28390 (insns_have_identical_cfa_notes): New function.
28391 (old_insns_match_p): Don't cross-jump in between /f
28392 and non-/f instructions. If both i1 and i2 are frame related,
28393 verify all CFA notes, their order and content.
28395 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
28398 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
28399 HImode and SImode with zero extend to DImode to one insn.
28400 (bswap<mode>2_extenddi): Likewise.
28401 (bswapsi2_extenddi): Likewise.
28402 (bswaphi2_extendsi): Likewise.
28403 (bswaphi2): Combine bswap HImode and SImode into one insn.
28404 Separate memory insns from swapping register.
28405 (bswapsi2): Likewise.
28406 (bswap<mode>2): Likewise.
28407 (bswaphi2_internal): Delete, no longer used.
28408 (bswapsi2_internal): Likewise.
28409 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
28410 store, and gpr<-gpr swap insns.
28411 (bswap<mode>2_store): Likewise.
28412 (bswaphi2_reg): Register only splitter, combine with the splitter.
28413 (bswaphi2 splitter): Likewise.
28414 (bswapsi2_reg): Likewise.
28415 (bswapsi2 splitter): Likewise.
28416 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
28417 the insns into load, store, and register/register insns.
28418 (bswapdi2_ldbrx): Likewise.
28419 (bswapdi2_load): Likewise.
28420 (bswapdi2_store): Likewise.
28421 (bswapdi2_reg): Likewise.
28423 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
28425 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
28426 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
28428 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
28431 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
28433 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28434 special handling for target option conflicts between dform
28435 options (-mpower9-dform, -mpower9-dform-vector,
28436 -mpower9-dform-scalar) and -mno-direct-move.
28438 2017-03-27 Richard Biener <rguenther@suse.de>
28440 PR tree-optimization/80181
28441 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
28443 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28445 * config/arc/predicates.md (move_double_src_operand): Replace the
28446 call to move_double_src_operand with a call to address_operand.
28448 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28450 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
28451 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
28452 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
28454 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28456 * config/arc/predicates.md (long_immediate_loadstore_operand):
28457 Consider scaled addresses cases.
28459 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
28461 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
28462 restored when in interrupt.
28463 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
28464 doesn't have delay slot.
28466 2017-03-27 Richard Biener <rguenther@suse.de>
28469 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
28470 inlined thunk clones.
28472 2017-03-27 Jakub Jelinek <jakub@redhat.com>
28475 * asan.c (instrument_derefs): Copy over last operand from
28476 original COMPONENT_REF to the new COMPONENT_REF with
28477 DECL_BIT_FIELD_REPRESENTATIVE.
28478 * ubsan.c (instrument_object_size): Likewise.
28480 2017-03-27 Richard Biener <rguenther@suse.de>
28482 PR tree-optimization/80170
28483 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
28484 sure DR/SCEV didnt fold in constants we do not see when looking
28485 at the reference base alignment.
28487 2017-03-27 Richard Biener <rguenther@suse.de>
28489 PR middle-end/80171
28490 * gimple-fold.c (fold_ctor_reference): Properly guard against
28491 NULL return value from canonicalize_constructor_val.
28493 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
28496 * config/i386/i386.c (ix86_expand_builtin)
28497 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
28498 flags reg setting and flags reg using instructions.
28499 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
28500 clobbering instructions to zero extend op2.
28502 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
28504 * doc/install.texi (Configuration) <--with-aix-soname>:
28505 Update link to AIX ld.
28507 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
28509 PR rtl-optimization/80160
28510 PR rtl-optimization/80159
28511 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
28512 reg_alternate_class into account.
28514 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
28517 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
28518 to consider in curr_insn_transform.
28520 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28522 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
28523 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
28524 and emit_mode_inner.
28526 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28528 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
28529 argument to the overloaded builtin variants. Use the new flag to
28530 deprecate certain builtin variants.
28531 * config/s390/s390-builtin-types.def: Add new builtin types.
28532 * config/s390/s390-builtins.h: Support new flags field for
28533 overloaded builtins.
28534 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
28535 (s390_macro_to_expand): Enable vector float data type.
28536 (s390_cpu_cpp_builtins_internal): Indicate support of the new
28537 builtins by incrementing the __VEC__ version number.
28538 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
28540 (s390_resolve_overloaded_builtin): Emit error messages depending
28541 on the builtin flags.
28542 * config/s390/s390.c (s390_expand_builtin): Support additional
28543 flags argument. Change error message to match the messages
28544 emitted in s390-c.c.
28545 * config/s390/s390.md: New UNSPEC_* constants.
28546 (op_type): Add new instruction types.
28547 * config/s390/vecintrin.h: Add new builtins and test data class
28549 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
28550 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
28551 (VEC_INEXACT, VEC_NOINEXACT): New constants.
28552 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
28553 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
28554 ("vec_mergel<mode>"): V_HW -> VEC_HW.
28556 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
28557 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
28558 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
28559 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
28561 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
28562 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
28563 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
28564 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
28566 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
28567 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
28568 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
28569 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
28570 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
28571 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
28572 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
28574 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
28575 ("vec_scatter_element<V_HW_4:mode>_DI")
28576 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
28577 ("vec_fpint<mode>", "vflls")
28578 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
28579 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
28580 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
28581 ("*vec_cmphe<mode>_cc"): ... these.
28583 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
28584 mode constant instead of magic value.
28586 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28588 * config/s390/s390.c (s390_expand_vec_compare): Support other
28589 vector floating point modes than just V2DF.
28590 (s390_expand_vcond): Likewise.
28591 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
28592 (s390_cannot_change_mode_class): Prevent mode changes between TF
28593 and V1TF in vector registers.
28594 * config/s390/s390.md (DF, SF): New mode attributes.
28595 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
28596 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
28597 SFmode support for VRs.
28598 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
28600 (VFT, VF_HW): New mode iterators.
28601 (vw, sdx): New mode attributes.
28602 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
28603 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
28604 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
28605 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
28606 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
28607 also the new vector floating point modes. Renaming to ...
28609 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
28610 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
28611 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
28612 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
28613 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
28614 ("vec_unordered<mode>"): ... these.
28616 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
28617 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
28618 ("*vec_extendv2df"): New insn definitions.
28620 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28622 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
28623 ("mulditi3_2", "*muldi3_sign"): New patterns.
28624 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
28625 rename the pattern definition.
28627 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28629 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
28631 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
28633 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28635 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
28636 instruction if possible.
28637 * config/s390/vector.md (vec_halfnumelts): New mode
28639 ("*vec_vllezlf<mode>"): New pattern.
28641 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28643 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
28644 ("popcountv4si2", "popcountv2di2"): Rename to ...
28645 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
28646 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
28648 ("popcount<mode>2_vxe"): New pattern.
28650 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28652 * common/config/s390/s390-common.c (processor_flags_table): Add
28654 * config.gcc: Add arch12.
28655 * config/s390/driver-native.c (s390_host_detect_local_cpu):
28656 Default to arch12 for unknown CPU model numbers.
28657 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
28658 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
28659 PROCESSOR_max sanity check.
28660 * config/s390/s390-opts.h (enum processor_type): Add
28662 * config/s390/s390.c (processor_table): Add arch12.
28663 (s390_expand_builtin): Add check for B_VXE flag.
28664 (s390_issue_rate): Add PROCESSOR_ARCH12.
28665 (s390_get_sched_attrmask): Likewise.
28666 (s390_get_unit_mask): Likewise.
28667 (s390_sched_score): Enable z13 scheduling for arch12.
28668 (s390_sched_reorder): Likewise.
28669 (s390_sched_variable_issue): Likewise.
28670 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
28672 (s390_tune_attr): Use z13 scheduling also for arch12.
28673 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
28674 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
28675 (TARGET_VXE_P): New macros.
28676 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
28677 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
28678 * config/s390/s390.opt: Add arch12 as processor_type.
28680 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28682 * config/s390/s390.md
28683 ("fixuns_truncdddi2", "fixuns_trunctddi2")
28684 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
28685 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
28687 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
28688 Rename expanders to ...
28690 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
28691 ("fixuns_truncdddi2_emu"): ... these.
28693 ("fixuns_trunc<mode>si2_emu"): New expander.
28695 ("*fixuns_truncdfdi2_z13"): Rename to ...
28696 ("*fixuns_truncdfdi2_vx"): ... this.
28698 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28700 * config/s390/2964.md: Remove the single element vector compare
28701 instructions which are no longer used.
28702 * config/s390/s390.c (s390_select_ccmode): Remove handling of
28704 (s390_canonicalize_comparison): Remove handling of DFmode
28706 (s390_expand_vec_compare_scalar): Remove function.
28707 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
28708 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
28710 ("*cmp<mode>_ccs"): Add wfcdb instruction.
28712 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28714 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
28716 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
28717 will anyway by matched by mov<mode>_64dfp.
28719 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28721 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
28722 vlef/vstef. Add missing operand to vleif.
28724 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28726 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
28727 pair for all vector types with 64 bit elements.
28728 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
28729 * config/s390/vector.md (V_HW_64): ... here.
28730 (V_128_NOSINGLE): New mode iterator.
28731 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
28732 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
28733 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
28734 ("*vec_load_pairv2di"): Change to ...
28735 ("*vec_load_pair<mode>"): ... this one.
28737 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28739 * config/s390/constraints.md: Add comments.
28740 (jKK): Reject element sizes > 8 bytes.
28741 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
28743 * config/s390/s390.md: Add the s_operand checks formerly in
28744 s390_split_ok_p to various splitters where they are still
28746 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
28747 for 128 bit vectors. Plus two splitters.
28749 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28751 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
28754 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28757 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
28758 error if the boundary argument is not constant.
28760 2017-03-24 Jakub Jelinek <jakub@redhat.com>
28762 PR rtl-optimization/80112
28763 * loop-doloop.c (doloop_condition_get): Don't check condition
28764 if cmp isn't SET with IF_THEN_ELSE src.
28766 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28768 PR tree-optimization/80158
28769 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
28770 replacing a candidate statement, also replace it for the
28771 candidate's alternate interpretation.
28772 (replace_rhs_if_not_dup): Likewise.
28773 (replace_one_candidate): Likewise.
28775 2017-03-24 Richard Biener <rguenther@suse.de>
28777 PR tree-optimization/80167
28778 * graphite-isl-ast-to-gimple.c
28779 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
28781 (translate_isl_ast_to_gimple::get_rename): Likewise.
28783 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
28785 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
28786 handling of certain combinations of target options, including the
28787 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
28788 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
28790 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28793 * config/arm/arm.md (*load_multiple): Add reload_completed to
28794 matching condition.
28796 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28797 Richard Biener <rguenth@suse.de>
28799 PR tree-optimization/79908
28800 PR tree-optimization/80136
28801 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28802 been cast away, gimplify_and_add suffices.
28804 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
28806 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
28808 2017-03-23 Richard Biener <rguenther@suse.de>
28810 PR tree-optimization/80032
28811 * gimplify.c (gimple_push_cleanup): Forced unconditional
28812 cleanups still have to go to the conditional_cleanups
28815 2017-03-22 Jakub Jelinek <jakub@redhat.com>
28817 PR tree-optimization/80072
28818 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
28820 (next_operand_entry_id): Change type to unsigned int.
28821 (sort_by_operand_rank): Make sure to return the right return value
28822 even if unsigned fields are bigger than INT_MAX.
28823 (struct oecount): Change cnt and id type to unsigned int.
28824 (oecount_hasher::equal): Formatting fix.
28825 (oecount_cmp): Make sure to return the right return value
28826 even if unsigned fields are bigger than INT_MAX.
28827 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
28830 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
28831 TREE_READONLY on result if writing it more than once.
28834 * doc/invoke.texi (-fsanitize=thread): Document that with
28835 -fnon-call-exceptions atomics are not able to throw
28839 * tsan.c: Include tree-eh.h.
28840 (instrument_builtin_call): Call maybe_clean_eh_stmt or
28841 maybe_clean_or_replace_eh_stmt where needed.
28842 (instrument_memory_accesses): Add cfg_changed argument.
28843 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
28844 if it returned true.
28845 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
28847 PR rtl-optimization/63191
28848 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
28849 wrapper function, moved the whole old content into ...
28850 (ix86_delegitimize_address_1): ... this. New inline function.
28851 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
28852 true as last argument instead of ix86_delegitimize_address.
28854 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28856 * config/aarch64/aarch64.c (generic_branch_cost): Copy
28857 cortexa57_branch_cost.
28859 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
28861 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
28863 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28866 * doc/md.texi (Constraints): Document wA constraint.
28867 * config/rs6000/constraints.md (wA): New.
28868 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
28869 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
28870 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
28871 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
28873 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
28876 * gimplify.c (is_oacc_declared): New function.
28877 (oacc_default_clause): Use it to set default flags for acc declared
28878 variables inside parallel regions.
28879 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
28880 declared variables.
28881 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
28882 declare attribute to any decl as necessary.
28884 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
28887 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
28888 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
28889 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
28890 (arm_arch_lpae): This.
28891 * config/arm/arm.c (arm_arch7ve): Rename into ...
28892 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
28893 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
28896 2017-03-22 Martin Liska <mliska@suse.cz>
28899 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
28900 error message instead of an ICE.
28902 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28904 * doc/extend.texi (6.11 Additional Floating Types): Revise.
28906 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
28908 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
28910 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
28913 2017-03-21 Martin Sebor <msebor@redhat.com>
28915 * doc/extend.texi: Use "cannot" instead of "can't."
28916 * doc/hostconfig.texi: Same.
28917 * doc/install.texi: Same.
28918 * doc/invoke.texi: Same.
28919 * doc/loop.texi: Same.
28920 * doc/md.texi: Same.
28921 * doc/objc.texi: Same.
28922 * doc/rtl.texi: Same.
28923 * doc/tm.texi: Same.
28924 * doc/tm.texi.in: Same.
28925 * doc/trouble.texi: Same.
28927 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
28930 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
28931 (collect_checksum_attributes): Set it.
28932 (die_checksum_ordered): Use it.
28934 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28936 PR tree-optimization/79908
28937 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
28938 change: For a VA_ARG whose LHS has been cast away, use
28939 force_gimple_operand to construct the side effects.
28941 2017-03-21 David Malcolm <dmalcolm@redhat.com>
28943 PR translation/80001
28944 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
28945 more amenable to translation.
28946 (oacc_loop_auto_partitions): Likewise.
28948 2017-03-21 Marek Polacek <polacek@redhat.com>
28949 Martin Sebor <msebor@redhat.com>
28951 PR tree-optimization/80109
28952 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
28953 on INTEGRAL_TYPE_P.
28955 2017-03-21 Jakub Jelinek <jakub@redhat.com>
28956 Segher Boessenkool <segher@kernel.crashing.org>
28959 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
28960 check reg_used_between_p between insn and one of succ or succ2
28961 depending on if succ is artificial insn not inserted into insn
28964 2017-03-21 Martin Liska <mliska@suse.cz>
28966 PR gcov-profile/80081
28967 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
28968 * doc/gcc.texi: Include gcov-dump stuff.
28969 * doc/gcov-dump.texi: New file.
28971 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
28973 PR rtl-optimization/79150
28974 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
28975 conditional jump, if the jump is the last insn of the loop.
28977 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28978 Richard Biener <rguenth@suse.de>
28980 PR tree-optimization/79908
28981 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
28982 been cast away, use force_gimple_operand to construct the side
28985 2017-03-21 Martin Liska <mliska@suse.cz>
28987 PR libfortran/79956
28988 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
28991 2017-03-21 Brad Spengler <spender@grsecurity.net>
28994 * plugin.c (htab_hash_plugin): New function.
28995 (add_new_plugin): Use it and adjust.
28996 (parse_plugin_arg_opt): Adjust.
28997 (init_one_plugin): Likewise.
28999 2017-03-21 Richard Biener <rguenther@suse.de>
29001 PR tree-optimization/80032
29002 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
29003 if set force the cleanup to happen unconditionally.
29004 (gimplify_target_expr): Push inserted clobbers with force_uncond
29005 to avoid them being removed by control-dependent DCE.
29007 2017-03-21 Richard Biener <rguenther@suse.de>
29009 PR tree-optimization/80122
29010 * tree-inline.c (copy_bb): Do not expans va-arg packs or
29011 va_arg_pack_len when the inlined call stmt requires pack
29013 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
29015 2017-03-21 Jakub Jelinek <jakub@redhat.com>
29018 * tsan.c (instrument_builtin_call): If the memory model argument
29019 is not a constant, assume it is valid.
29022 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
29025 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29027 PR rtl-optimization/79910
29028 * combine.c (can_combine_p): Do not allow combining an I0 or I1
29029 if its dest is used by an insn before I2 (other than the combined
29030 insns themselves, which are properly handled already).
29032 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
29035 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29037 * combine.c (record_used_regs): New static function.
29038 (try_combine): Handle situations where there is an additional
29039 instruction between I2 and I3 which needs to have a LOG_LINK
29043 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29045 * combine.c (try_combine): Delete redundant i1 test. Call
29046 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29048 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29051 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
29052 alternatives 13/14.
29054 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29056 PR tree-optimization/80054
29057 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
29058 the optimization if a PHI or any of its arguments is not dominated
29059 by the candidate's basis. Use gphi* rather than gimple* as
29061 (replace_profitable_candidates): Clean up a gimple* variable that
29062 should be a gphi* variable.
29064 2017-03-20 Martin Sebor <msebor@redhat.com>
29067 * doc/extend.texi (attribute constructor): Document present limitation.
29069 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
29072 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
29073 __POWER9_VECTOR__ #ifdef control, change template definition to
29074 use Power9-specific built-in function.
29075 (vec_any_eq): Likewise.
29076 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
29077 to control outcomes from this test.
29078 (vector_ae_<mode>p): For VEC_F modes, likewise.
29080 2017-03-20 Ian Lance Taylor <iant@google.com>
29082 * config/i386/i386.c (ix86_function_regparm): Save an extra
29083 register for -fsplit-stack with DECL_STATIC_CHAIN.
29085 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29088 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
29089 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
29091 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
29093 * config/riscv/riscv.c (riscv_print_operand): Use "fence
29095 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
29098 2017-03-20 Marek Polacek <polacek@redhat.com>
29101 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
29103 2017-03-20 Richard Biener <rguenther@suse.de>
29105 PR tree-optimization/80113
29106 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
29107 allocate extra SSA name for PHI def.
29108 (add_close_phis_to_outer_loops): Likewise.
29109 (add_close_phis_to_merge_points): Likewise.
29110 (copy_loop_close_phi_args): Likewise.
29111 (copy_cond_phi_nodes): Likewise.
29113 2017-03-20 Martin Liska <mliska@suse.cz>
29115 PR middle-end/79753
29116 * tree-chkp.c (chkp_build_returned_bound): Do not build
29117 returned bounds for a LHS that's not a BOUNDED_P type.
29119 2017-03-20 Martin Liska <mliska@suse.cz>
29123 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
29124 COMPLEX_CST and VECTOR_CST.
29126 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29129 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
29130 target operand. A new splitter adds the clobber statement in case
29131 the target operand is dead anyway.
29133 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
29135 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
29136 to age-old versions of binutils and glibc.
29138 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
29140 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
29142 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29144 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
29146 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
29148 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
29149 requirement for binutils 2.13.
29151 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
29153 * combine.c (try_combine): Delete redundant i1 test. Call
29154 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
29156 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
29158 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
29159 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
29161 <riscv64-*-elf>: Re-arrange section
29162 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
29163 <riscv32-*-linux>: Likewise.
29164 <riscv64-*-elf>: Likewise
29165 <riscv64-*-linux>: Likewise.
29167 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
29170 * aarch64.opt(verbose-cost-dump): Fix typo.
29172 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
29175 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
29176 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
29178 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
29180 * reload.c (find_reloads): When reloading a nonoffsettable address,
29181 use RELOAD_OTHER for it and its address reloads.
29183 PR rtl-optimization/79910
29184 * combine.c (record_used_regs): New static function.
29185 (try_combine): Handle situations where there is an additional
29186 instruction between I2 and I3 which needs to have a LOG_LINK
29189 2017-03-17 Jeff Law <law@redhat.com>
29191 PR tree-optimization/71437
29192 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
29193 conditional in the hash table first.
29194 (vrp_dom_walker::before_dom_children): Extract condition from
29195 ASSERT_EXPR. Record condition, its inverion and any implied
29196 conditions as well.
29198 2017-03-17 Marek Polacek <polacek@redhat.com>
29199 Markus Trippelsdorf <markus@trippelsdorf.de>
29201 PR tree-optimization/80079
29202 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
29205 2017-03-17 Richard Biener <rguenther@suse.de>
29207 PR middle-end/80075
29208 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
29209 Properly verify the LHS before the RHS possibly claims to be
29211 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
29214 2017-03-17 Martin Jambor <mjambor@suse.cz>
29216 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
29217 (List of -O2 options): Likewise.
29218 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
29221 2017-03-17 Tom de Vries <tom@codesourcery.com>
29223 * gcov-dump.c (print_usage): Print bug_report_url.
29225 2017-03-17 Richard Biener <rguenther@suse.de>
29227 PR middle-end/80050
29228 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
29229 (parser::peek): Likewise.
29231 2017-03-17 Richard Biener <rguenther@suse.de>
29233 PR tree-optimization/80048
29234 * sese.c (free_sese_info): Properly release rename_map and
29235 copied_bb_map elements.
29237 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
29239 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
29240 Add linked-list forward and backlinks. Insert on
29241 construction, remove on destruction.
29242 (class pass_store_merging): Add m_stores_head field.
29243 (pass_store_merging::terminate_and_process_all_chains):
29244 Iterate over m_stores_head list.
29245 (pass_store_merging::terminate_all_aliasing_chains):
29247 (pass_store_merging::execute): Check for debug stmts first.
29248 Push new chains onto the m_stores_head stack.
29250 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29253 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
29254 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
29255 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
29257 2017-03-16 Jeff Law <law@redhat.com>
29259 PR tree-optimization/71437
29260 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
29261 member function. Implementation moved into after_dom_children
29262 member function and into the threader's thread_outgoing_edges
29264 (dom_opt_dom_walker::after_dom_children): Simplify by moving
29265 some code into new thread_outgoing_edges.
29266 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
29267 definition. Simplify marker handling (do it here). Assume we always
29268 have the available expression and the const/copies tables.
29269 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
29271 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
29272 * tree-vrp.c (equiv_stack): No longer file scoped.
29273 (vrp_dom_walker): New class.
29274 (vrp_dom_walker::before_dom_children): New member function.
29275 (vrp_dom_walker::after_dom_children): Likewise.
29276 (identify_jump_threads): Setup domwalker. Use it rather than
29277 walking edges in a random order by hand. Simplify setup/finalization.
29278 (finalize_jump_threads): Remove.
29279 (vrp_finalize): Do not call identify_jump_threads here.
29280 (execute_vrp): Do it here instead and call thread_through_all_blocks
29283 PR tree-optimization/71437
29284 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
29286 (simplify_stmt_for_jump_threading): Add basic_block argument. All
29288 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
29289 (dom_opt_dom_walker::thread_across_edge): Remove
29290 handle_dominating_asserts argument. All callers changed.
29291 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
29292 changes. Remove calls to lhs_of_dominating_assert. Other
29293 uses of handle_dominating_asserts turn into unconditional code
29294 (simplify_control_stmt_condition_1): Likewise.
29295 (simplify_control_stmt_condition): Likewise.
29296 (thread_through_normal_block, thread_across_edge): Likewise.
29297 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
29298 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
29299 object if it is not an SSA_NAME.
29300 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
29301 before calling into the VRP specific simplifiers.
29302 (identify_jump_threads): Remove handle_dominating_asserts
29305 2017-03-16 Jakub Jelinek <jakub@redhat.com>
29308 * tree-diagnostic.c (default_tree_printer): No longer static.
29309 * tree-diagnostic.h (default_tree_printer): New prototype.
29311 2017-03-16 Tamar Christina <tamar.christina@arm.com>
29313 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
29314 Change ins into fmov.
29316 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29318 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
29319 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
29320 Use h_con constraint for operand 1.
29321 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
29322 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
29324 2017-03-15 Jeff Law <law@redhat.com>
29326 PR tree-optimization/71437
29327 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
29328 (record_temporary_equivalences): Use it.
29330 PR tree-optimization/71437
29331 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
29332 tree-ssa-scopedtables.
29333 (lookup_avail_expr, build_and_record_new_cond): Likewise.
29334 (record_conditions, record_cond, vuse_eq): Likewise.
29335 (record_edge_info): Adjust to API tweak of record_conditions.
29336 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
29337 (record_temporary_equivalences, optimize_stmt): Likewise.
29338 (eliminate_redundant_computations): Likewise.
29339 (record_equivalences_from_stmt): Likewise.
29340 * tree-ssa-scopedtables.c: Include options.h and params.h.
29341 (vuse_eq): New function, moved from tree-ssa-dom.c
29342 (build_and_record_new_cond): Likewise.
29343 (record_conditions): Likewise. Accept vector of conditions rather
29344 than edge_equivalence structure for first argument.
29345 for the first argument.
29346 (avail_exprs_stack::lookup_avail_expr): New member function, moved
29347 from tree-ssa-dom.c.
29348 (avail_exprs_stack::record_cond): Likewise.
29349 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
29350 from tree-ssa-dom.c.
29351 (avail_exprs_stack): Add new member functions lookup_avail_expr
29353 (record_conditions): Declare.
29355 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
29358 * lra-constraints.c (process_alt_operands): Increase reject for
29359 reloading an input/output operand.
29361 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
29364 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
29365 insns to convert from signed/unsigned char/short to IEEE 128-bit
29367 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
29369 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
29372 * config/i386/i386.c (ix86_vector_duplicate_value): Create
29373 subreg of inner mode for values already in registers.
29375 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
29377 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
29378 iteration reg is used after the loop.
29380 2017-03-14 Martin Sebor <msebor@redhat.com>
29382 PR tree-optimization/79800
29383 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
29384 precision in negative-positive range.
29385 (format_floating): Call non-const overload with adjusted precision.
29387 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
29390 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
29393 2017-03-14 Martin Sebor <msebor@redhat.com>
29395 PR middle-end/80020
29396 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
29397 * builtins.def (aligned_alloc): Use it.
29400 * Makefile.in (GTFILES): Add calls.c.
29401 * calls.c: Include "gt-calls.h".
29403 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
29405 PR rtl-optimization/79728
29406 * regs.h (struct target_regs): New field
29407 x_contains_allocatable_regs_of_mode.
29408 (contains_allocatable_regs_of_mode): New macro.
29409 * reginfo.c (init_reg_sets_1): Initialize it, and change
29410 contains_reg_of_mode so it includes global regs as well.
29411 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
29412 rather than contains_regs_of_mode.
29414 2017-03-14 Martin Liska <mliska@suse.cz>
29416 * doc/invoke.texi: Document options that can't be combined with
29417 -fcheck-pointer-bounds.
29419 2017-03-14 Martin Liska <mliska@suse.cz>
29421 PR middle-end/79831
29422 * doc/invoke.texi (-Wchkp): Document the option.
29424 2017-03-14 Martin Liska <mliska@suse.cz>
29426 * Makefile.in: Install gcov-dump.
29428 2017-03-14 Martin Liska <mliska@suse.cz>
29430 * multiple_target.c (expand_target_clones): Bail out for
29431 an invalid attribute.
29433 2017-03-14 Richard Biener <rguenther@suse.de>
29435 * alias.c (struct alias_set_entry): Pack properly.
29436 * cfgloop.h (struct loop): Likewise.
29437 * cse.c (struct set): Likewise.
29438 * ipa-utils.c (struct searchc_env): Likewise.
29439 * loop-invariant.c (struct invariant): Likewise.
29440 * lra-remat.c (struct cand): Likewise.
29441 * recog.c (struct change_t): Likewise.
29442 * rtl.h (struct address_info): Likewise.
29443 * symbol-summary.h (function_summary): Likewise.
29444 * tree-loop-distribution.c (struct partition): Likewise.
29445 * tree-object-size.c (struct object_size_info): Likewise.
29446 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
29447 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
29448 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
29449 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
29450 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
29451 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
29452 (struct _stmt_vec_info): Likewise.
29454 2017-03-14 Martin Liska <mliska@suse.cz>
29457 * multiple_target.c (create_dispatcher_calls): Check that
29458 a target can create a function dispatcher.
29460 2017-03-14 Martin Liska <mliska@suse.cz>
29463 * multiple_target.c (expand_target_clones): Drop local.local
29464 flag for default implementation.
29466 2017-03-14 Richard Biener <rguenther@suse.de>
29468 PR tree-optimization/80030
29469 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
29471 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
29473 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
29474 gcc_fallthrough() instead of __attribute__((fallthrough));
29476 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29478 * doc/gcc.texi: Remove "up" link to (DIR).
29479 * doc/gccint.texi: Ditto.
29481 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
29483 * doc/install.texi (Specific) <avr>: Remove reference to
29486 2017-03-13 Jeff Law <law@redhat.com>
29488 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
29489 attribute rather than comments.
29491 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
29492 match_scratch operand is highest.
29494 2017-03-13 Martin Liska <mliska@suse.cz>
29496 PR middle-end/78339
29497 * ipa-pure-const.c (warn_function_noreturn): If the declarations
29498 is a CHKP clone, use original declaration.
29500 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29502 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
29503 (arc_conditional_register_usage): Use a different allocation order
29504 when optimizing for size.
29505 * common/config/arc/arc-common.c (arc_option_optimization_table):
29506 Section anchors default on when optimizing for size.
29508 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29510 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
29512 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29514 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
29515 * config/arc/arc.md (cpu_facility): Add cd variant.
29516 (*movqi_insn): Add code density variant.
29517 (*movhi_insn): Likewise.
29518 (*movqi_insn): Likewise.
29519 (*addsi3_mixed): Likewise.
29520 (subsi3_insn): Likewise.
29522 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29524 * config/arc/arc.md (movsi_cond_exec): Update constraint.
29526 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
29528 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
29529 expressions with MINUS and UNARY ops.
29531 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29534 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
29536 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
29537 between vec_select and vector argument.
29538 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
29539 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
29540 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
29541 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
29542 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
29543 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
29545 2017-03-13 Richard Biener <rguenther@suse.de>
29548 * params.def (vect-max-peeling-for-alignment): Fix typo.
29550 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29552 * doc/install.texi (Specific) <mips-*-*>: Remove description of
29553 issue that only occurred with binutils below 2.18.
29555 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29557 * doc/install.texi (Specific) <cris-axis-elf>: No longer
29558 refer to binutils 2.11/2.12 minimum.
29560 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
29562 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
29563 ftp.kernel.org and simplify binutils requirement.
29565 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
29567 * doc/invoke.texi (Warning Options): Fix spelling of link-time
29569 (Optimize Options): Ditto. Also remove redundancy.
29571 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29573 PR translation/79848
29574 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
29576 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
29577 to G_ to avoid double translation.
29579 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29581 PR translation/79923
29582 * auto-profile.c (get_combined_location): Convert leading
29583 character of diagnostics to lower case and remove trailing period.
29584 (read_profile): Likewise for various diagnostics.
29585 * config/arm/arm.c (arm_option_override): Remove trailing period
29586 from various diagnostics.
29587 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
29588 (msp430_expand_delay_cycles): Likewise.
29590 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29593 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
29594 full command-line argument, rather than just "str".
29595 (aarch64_validate_march): Likewise.
29596 (aarch64_validate_mtune): Likewise.
29598 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
29600 PR rtl-optimization/78911
29601 * lra-assigns.c (must_not_spill_p): New function.
29602 (spill_for): Use it.
29604 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29606 PR tree-optimization/79981
29607 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
29608 ATOMIC_COMPARE_EXCHANGE ifn result.
29609 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
29610 IFN_ATOMIC_COMPARE_EXCHANGE.
29612 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29615 * opts.c (parse_sanitizer_options): Add missing question mark to
29616 "did you mean" message.
29618 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29620 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
29622 (VMULEUH_UNS): Likewise.
29623 (VMULOUB_UNS): Likewise.
29624 (VMULOUH_UNS): Likewise.
29625 * config/rs6000/rs6000.c (builtin_function_type): Remove
29626 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
29628 2017-03-10 David Malcolm <dmalcolm@redhat.com>
29631 * read-rtl-function.c (function_reader::read_rtx_operand): Update
29632 x with result of extra_parsing_for_operand_code_0.
29633 (function_reader::extra_parsing_for_operand_code_0): Convert
29634 return type from void to rtx, returning x. When reading
29635 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
29636 larger size containing struct block_symbol.
29638 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
29640 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
29641 -mfloat128-hardware without -m64.
29643 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
29646 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
29647 entries to the case statement that marks unsigned arguments to
29648 overloaded functions.
29650 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29652 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
29653 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
29655 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
29658 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
29659 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
29661 2017-03-10 Martin Liska <mliska@suse.cz>
29665 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
29666 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
29669 2017-03-10 Olivier Hainque <hainque@adacore.com>
29671 * tree-switch-conversion (array_value_type): Start by resetting
29672 candidate type to it's main variant.
29674 2017-03-10 Jakub Jelinek <jakub@redhat.com>
29676 PR rtl-optimization/79909
29677 * combine.c (try_combine): Use simplify_replace_rtx on individual
29678 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
29679 of the whole CALL_INSN_FUNCTION_USAGE.
29681 PR tree-optimization/79972
29682 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
29683 get_range_info on SSA_NAMEs. Formatting fixes.
29685 2017-03-10 Richard Biener <rguenther@suse.de>
29686 Jakub Jelinek <jakub@redhat.com>
29688 PR tree-optimization/77975
29689 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
29690 edge to be constant.
29691 (get_val_for): For constant x return it. Formatting fix.
29692 (loop_niter_by_eval): Avoid pointless looping if the next iteration
29693 would use the same bases as the current one.
29695 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29697 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
29698 instead of vec_select for V1TImode.
29699 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
29701 (VSX_LE_128): Add V1TI to this mode iterator.
29702 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
29703 (*vsx_le_perm_store_<mode>): Likewise.
29704 (pre-reload splitter for VSX stores): Likewise.
29705 (post-reload splitter for VSX stores): Likewise.
29706 (*vsx_xxpermdi2_le_<mode>): Likewise.
29707 (*vsx_lxvd2x2_le_<mode>): Likewise.
29708 (*vsx_stxvd2x2_le_<mode>): Likewise.
29710 2017-03-09 Michael Eager <eager@eagercon.com>
29712 Correct failures with --enable-checking=yes,rtl.
29714 * config/microblaze/microblaze.c (microblaze_expand_shift):
29715 Replace GET_CODE test with CONST_INT_P and INTVAL test with
29716 test for const0_rtx.
29717 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
29718 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
29720 2017-03-09 Richard Biener <rguenther@suse.de>
29722 PR tree-optimization/79977
29723 * graphite-scop-detection.c (scop_detection::merge_sese):
29724 Handle the case of extra exits to blocks dominating the entry.
29726 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
29728 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
29731 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
29733 PR rtl-optimization/79949
29734 * lra-constraints.c (process_alt_operands): Check memory when
29735 trying to predict a cycle. Print about the overall increase.
29737 2017-03-09 Richard Biener <rguenther@suse.de>
29739 PR middle-end/79971
29740 * gimple-expr.c (useless_type_conversion_p): Preserve
29741 TYPE_SATURATING for fixed-point types.
29743 2017-03-09 Richard Biener <rguenther@suse.de>
29746 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
29747 alignment of BLKmode params.
29749 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29752 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
29753 (VALL_NO_V2Q): Likewise.
29755 * config/aarch64/aarch64-simd.md
29756 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
29758 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
29759 VALL_NO_V2Q mode iterator.
29760 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
29762 2017-03-09 Martin Liska <mliska@suse.cz>
29764 PR tree-optimization/79631
29765 * tree-chkp-opt.c (chkp_is_constant_addr): Call
29766 tree_int_cst_sign_bit just for INTEGER constants.
29768 2017-03-09 Martin Liska <mliska@suse.cz>
29772 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
29775 2017-03-09 Marek Polacek <polacek@redhat.com>
29778 * tree.c (inchash::add_expr): Handle TREE_VEC.
29780 2017-03-09 Martin Liska <mliska@suse.cz>
29783 (chkp_narrow_size_and_offset): New function.
29784 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
29785 (void chkp_parse_bit_field_ref): New function.
29786 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
29787 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
29789 2017-03-09 Martin Liska <mliska@suse.cz>
29792 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
29793 (chkp_find_bounds_1): Remove gcc_unreachable.
29795 2017-03-09 Jakub Jelinek <jakub@redhat.com>
29798 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
29799 BUILT_IN_SYNC*, determine the access type from the size suffix and
29800 always build a MEM_REF with that type. Handle forgotten
29801 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
29804 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
29805 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
29806 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
29807 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
29808 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
29809 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
29810 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
29811 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
29812 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
29813 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
29814 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
29815 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
29816 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
29817 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
29818 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
29819 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
29820 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
29821 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
29822 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
29823 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
29824 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
29825 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
29826 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
29827 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
29828 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
29829 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
29830 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
29831 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
29832 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
29833 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
29834 definitions outside of __OPTIMIZE__ guarded section.
29837 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
29838 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
29839 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
29840 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
29843 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29845 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
29846 ("vfenez<mode>"): Add missing constraints.
29848 2017-03-08 Martin Sebor <msebor@redhat.com>
29851 * config/nds32/nds32.c (nds32_option_override):
29852 Fix misspelled diagnostic.
29854 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29857 * gimplify.c (gimplify_omp_for): Replace index var in outer
29858 taskloop statement with an artificial variable and add
29859 OMP_CLAUSE_PRIVATE clause for it.
29861 2017-03-08 Richard Biener <rguenther@suse.de>
29863 PR tree-optimization/79955
29864 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
29865 for accesses that are completely outside of the variable.
29867 2017-03-08 Andrew Haley <aph@redhat.com>
29869 PR tree-optimization/79943
29870 * tree-ssa-loop-split.c (compute_new_first_bound): When
29871 calculating the new upper bound, (END-BEG) should be added, not
29874 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29876 * config/avr/avr.md (setmemhi): Make sure match_dup
29877 operand number comes before match_scratch.
29879 2017-03-08 Richard Biener <rguenther@suse.de>
29881 PR tree-optimization/79920
29882 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
29883 with ncopies == 1 to ...
29884 (vect_transform_slp_perm_load): ... here. Properly compute
29885 all element loads by iterating VF times over the group. Do
29886 not handle ncopies (computed in a broken way) in
29887 vect_create_mask_and_perm.
29889 2017-03-08 Jakub Jelinek <jakub@redhat.com>
29892 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
29893 is a uniform vector, use uniform_vector_p return value instead of
29894 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
29896 2017-03-07 Marek Polacek <polacek@redhat.com>
29898 PR middle-end/79809
29899 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
29900 (alloca_call_type): Likewise.
29902 2017-03-07 Martin Liska <mliska@suse.cz>
29904 * gcov.c (process_args): Put comment to correct location.
29906 2017-03-07 Martin Liska <mliska@suse.cz>
29908 PR middle-end/68270
29909 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
29910 Use array_at_struct_end_p instead of DECL_CHAIN (field).
29911 (chkp_narrow_bounds_for_field): Likewise.
29912 (chkp_parse_array_and_component_ref): Pass one more argument to
29915 2017-03-07 Richard Biener <rguenther@suse.de>
29917 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
29920 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
29922 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
29923 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
29925 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29928 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
29929 to end of description.
29930 (PARAM_MAX_STORES_TO_MERGE): Likewise.
29932 2017-03-07 Jakub Jelinek <jakub@redhat.com>
29934 PR rtl-optimization/79901
29935 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
29937 (*avx512f_<code><mode>3<mask_name>): ... this.
29938 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
29939 iterator instead of VI8_AVX2_AVX512BW.
29941 PR rtl-optimization/79901
29942 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
29943 min/max expander, expand it using expand_vec_cond_expr.
29946 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
29949 2017-03-06 Jakub Jelinek <jakub@redhat.com>
29952 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
29953 to void * for PCH reasons.
29954 * dwarf2out.c (output_loc_operands, output_die): Cast
29955 v.val_vec.array to unsigned char *.
29957 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
29960 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
29963 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
29965 PR rtl-optimization/79571
29966 * lra-constraints.c (process_alt_operands): Calculate static
29967 reject and subtract it from overall when only addresses will be
29970 2017-03-06 Julia Koval <julia.koval@intel.com>
29973 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
29974 incoming stack boundary to 128 for 64-bit targets.
29976 2017-03-06 Richard Biener <rguenther@suse.de>
29978 PR tree-optimization/79894
29979 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
29980 to NULL after folding it.
29982 2017-03-06 Richard Biener <rguenther@suse.de>
29984 PR tree-optimization/79824
29985 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
29986 check disabling peeling for gaps.
29988 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
29990 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
29991 attributes): Document gettimeofday.
29993 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
29995 * config/s390/s390.c (s390_option_override_internal): Set
29996 PARAM_MIN_VECT_LOOP_BOUND
29998 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
30000 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
30001 * config/s390/s390.md: Likewise.
30003 2017-03-06 Jakub Jelinek <jakub@redhat.com>
30006 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
30007 (<avx2_avx512>_perm<mode>): Rename to ...
30008 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
30010 (<avx512>_perm<mode>_mask): Rename to ...
30011 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
30013 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
30014 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
30015 instead of VI8F_256_512.
30016 (avx512f_perm<mode>): New define_expand.
30017 (avx512f_perm<mode>_mask): Likewise.
30018 (avx512f_perm<mode>_1<mask_name>): New define_insn.
30019 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
30021 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30023 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
30024 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
30026 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
30028 2017-03-06 Martin Liska <mliska@suse.cz>
30031 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
30032 when having a SSA NAME w/o VAR_DECL assigned to it.
30034 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30036 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
30037 msa_dpsub_<su>_d): Fix MODE for vec_select.
30039 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
30041 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
30043 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
30045 2017-03-06 Richard Biener <rguenther@suse.de>
30047 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
30048 * plugin.c (register_plugin_info): Likewise.
30049 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
30051 2017-03-05 Jakub Jelinek <jakub@redhat.com>
30053 * config/i386/sse.md (sse_storehps, sse_storelps,
30054 avx_<castmode><avxsizesuffix>_<castmode>,
30055 avx512f_<castmode><avxsizesuffix>_<castmode>,
30056 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
30057 in condition that at least one operand is not a MEM.
30059 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30061 PR middle-end/79805
30062 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
30063 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
30065 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
30066 gimple_call_nothrow_p flag based on whether original builtin can throw.
30067 If it can, emit following stmts on the fallthrough edge.
30068 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
30069 don't create new bb if inserting just debug stmts on the edge, try to
30070 insert them on the fallthru bb or just reset debug stmts.
30072 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
30075 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
30076 restore recog_data (including the operand rtxes inside it) around
30077 the call to get_insn_template.
30079 2017-03-03 Martin Sebor <msebor@redhat.com>
30081 PR tree-optimization/79699
30082 * context.c (context::~context): Free MPFR caches to avoid
30083 a memory leak on program exit.
30085 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30087 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
30088 Use wide_int::ulow () instead of .elt (0).
30090 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30092 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
30093 (*pushxf): Limit oF constraint to 32bit targets and add oC
30094 constraint for 64bit targets.
30095 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
30096 (*pushdf): Change rmF constraint to rmC.
30098 2017-03-03 Martin Liska <mliska@suse.cz>
30100 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
30101 Remove unused variable.
30103 2017-03-03 Jakub Jelinek <jakub@redhat.com>
30106 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
30107 is a memory operand, increase num_memory.
30108 (ix86_expand_args_builtin): Likewise.
30110 2017-03-03 Jan Hubicka <jh@suse.cz>
30113 * ipa-devirt.c (maybe_record_node): Properly handle
30114 __cxa_pure_virtual visibility.
30116 2017-03-03 Martin Liska <mliska@suse.cz>
30118 PR tree-optimization/79803
30119 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
30121 (pass_loop_prefetch::execute): Disabled optimization if an
30122 assumption about L1 cache size is not met.
30124 2017-03-03 Martin Liska <mliska@suse.cz>
30126 PR rtl-optimization/79574
30127 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
30128 (hash_scan_set): Likewise.
30129 (dump_hash_table): Likewise.
30130 (hoist_code): Likewise.
30132 2017-03-03 Richard Biener <rguenther@suse.de>
30134 * fixed-value.c (fixed_from_string): Restore use of elt (1)
30135 in place of uhigh ().
30136 (fixed_convert_from_real): Likewise.
30138 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
30141 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
30143 2017-03-03 Richard Biener <rguenther@suse.de>
30145 PR middle-end/79818
30146 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
30147 TYPE_OVERFLOW_UNDEFINED check.
30149 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30151 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
30153 (vector_ae_<mode>_p): Likewise.
30154 (vector_nez_<mode>_p): Likewise.
30155 (vector_ne_v2di_p): Likewise.
30156 (vector_ae_v2di_p): Likewise.
30157 (vector_ne_<mode>_p): Likewise.
30158 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
30160 (vsx_tsqrt<mode>2_fe): Likewise.
30162 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
30165 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
30167 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30169 PR rtl-optimization/79780
30170 * cprop.c (one_cprop_pass): When second and further conditional trap
30171 in a single basic block is turned into an unconditional trap, turn it
30172 into a deleted note to avoid RTL verification failures.
30174 2017-03-02 Richard Biener <rguenther@suse.de>
30176 * fold-const.c (const_binop): Use ulow () instead of elt (0).
30178 2017-03-02 Richard Biener <rguenther@suse.de>
30180 PR tree-optimization/79345
30182 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
30183 param and abort the walk, returning -1 if it is hit.
30184 (walk_aliased_vdefs): Take a limit param and pass it on.
30185 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
30186 defaulting to 0 and return a signed int.
30187 * tree-ssa-uninit.c (struct check_defs_data): New struct.
30188 (check_defs): New helper.
30189 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
30190 about uninitialized memory.
30191 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
30192 bogus uninitialized warning.
30193 (fixed_convert_from_real): Likewise.
30195 2017-03-02 Bin Cheng <bin.cheng@arm.com>
30197 PR tree-optimization/66768
30198 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
30199 iv_use if base object can't be determined.
30201 2017-03-02 Jakub Jelinek <jakub@redhat.com>
30203 PR tree-optimization/79345
30204 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
30205 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
30206 (get_pattern_stats): Initialize it.
30207 * genemit.c (gen_expand): Verify match_scratch numbers come after
30208 match_operand/match_dup numbers.
30209 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
30210 match_scratch numbers.
30211 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
30213 * config/s390/s390.md (trunctdsd2): Likewise.
30215 2017-03-02 Richard Biener <rguenther@suse.de>
30217 * wide-int.h (wide_int_storage::operator=): Implement in terms
30220 2017-03-02 Richard Biener <rguenther@suse.de>
30222 PR tree-optimization/79777
30223 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
30224 the to insert expression to sth existing.
30226 2017-03-01 Martin Sebor <msebor@redhat.com>
30228 PR middle-end/79692
30229 * gimple-ssa-sprintf.c
30230 (directive::known_width_and_precision): New function.
30231 (format_integer): Use it.
30232 (get_mpfr_format_length): Consider the full range of precision
30233 when computing %g output with the # flag. Set the likely byte
30234 count to 3 rather than 1 when precision is indeterminate.
30235 (format_floating): Correct the lower bound of precision.
30237 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30239 * doc/invoke.texi: Document default code model for 64-bit Linux.
30241 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30244 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
30245 udiv rather than div since input pattern is unsigned.
30247 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
30249 * config/i386/i386.c (print_reg): Warn for values of
30250 unsupported size in integer register.
30252 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
30255 * config/rs6000/predicates.md (current_file_function_operand): Do
30256 not allow self calls to be local if the function is replaceable.
30258 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
30261 * config/rs6000/altivec.h (vec_ctz and others): Change the
30262 preprocessor macro that controls conditional compilation from
30263 _ARCH_PWR9 to __POWER9_VECTOR__.
30264 (vec_all_ne): Change parameterization of __altivec_scalar_pred
30265 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
30266 control (instead of _ARCH_PWR9 control) so that template
30267 definition uses power9-specific function.
30268 (vec_any_eq): Likewise.
30269 (vec_all_ne): Change macro definition to use a power9-specific
30270 expansion under #ifdef __POWER9_VECTOR__ control (instead of
30271 _ARCH_PWR9 control).
30272 (vec_any_eq) Likewise.
30273 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
30274 expansion for CMPNEF to remove support for xvcmpnesp instruction.
30275 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
30276 support for xvcmpnedp instruction.
30277 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
30278 macro expansion so that Power9 implementation of vec_all_ne does
30279 not use the AltiVec predicate framework.
30280 (VCMPNEH_P): Likewise.
30281 (VCMPNEW_P): Likewise.
30282 (VCMPNED_P): Likewise.
30283 (VCMPNEFP_P): Likewise.
30284 (VCMPNEDP_P): Likewise.
30285 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
30286 implementation of vec_any_eq to not use AltiVec predicate
30288 (VCMPAEH_P): Likewise.
30289 (VCMPAEW_P): Likewise.
30290 (VCMPAED_P): Likewise.
30291 (VCMPAEFP_P): Likewise.
30292 (VCMPAEDP_P): Likewise.
30293 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
30294 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
30295 not use the AltiVec predicate framework.
30296 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
30297 of vec_any_eq to not use AltiVec predicate framework.
30298 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
30299 support for predefined __POWER9_VECTOR__ macro to indicate that
30300 Power9 instruction selection is enabled.
30301 (altivec_overloaded_builtins): Remove extraneous
30302 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
30303 function argument types RS6000_BTI_bool_V16QI and
30304 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
30305 entry for overloaded function argument types RS6000_BTI_bool_V4SI
30306 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
30307 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
30308 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
30309 Power9 for implementations of vec_cmpne. Change the signature for
30310 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
30311 (representing vec_all_ne) to remove the previously described first
30312 argument of type RS6000_BTI_INTSI, as this was an artifact of
30313 reliance on the AltiVec predicate framework, which is no longer
30314 used in the implementation of these functions. Add
30315 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
30316 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
30317 since, unlike the AltiVec predicate framework implementation, we
30318 do not share function descriptors between vec_alle and vec_anyeq.
30319 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
30320 set of modes that receive special treatment even when
30321 TARGET_P9_VECTOR is true. The special treatment emits code that
30322 does not depend on Power9 instructions.
30323 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
30324 define_expand to not rely on AltiVec predicate framework.
30325 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30327 (vector_ne_v2di_p): Change this define_expand to not rely on
30328 AltiVec predicate framework.
30329 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
30331 (vector_ne_<mode>_p): Change this define_expand to not rely on
30332 AltiVec predicate framework.
30333 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
30335 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
30336 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
30337 define_insn pattern.
30338 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
30339 define_insn pattern because the xvcmpne<VSs>. instruction is not
30341 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
30342 instruction is not supported.
30344 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30346 * config/nvptx/nvptx.c: Include intl.h.
30348 2017-03-01 Martin Jambor <mjambor@suse.cz>
30351 * ipa-prop.h (ipa_bits): Removed field known.
30352 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
30353 to pointers. Adjusted their comments to warn about their sharing.
30354 (ipcp_transformation_summary): Change bits to a vector of pointers.
30355 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
30356 (ipa_get_ipa_bits_for_value): Declare.
30357 * tree-vrp.h (value_range): Mark as GTY((for_user)).
30358 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
30359 (ipa_bits_hash_table): Likewise.
30360 (ipa_vr_ggc_hash_traits): Likewise.
30361 (ipa_vr_hash_table): Likewise.
30362 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
30363 being pointers and vr_known being removed.
30364 (ipa_set_jf_unknown): Likewise.
30365 (ipa_get_ipa_bits_for_value): New function.
30366 (ipa_set_jfunc_bits): Likewise.
30367 (ipa_get_value_range): New overloaded functions.
30368 (ipa_set_jfunc_vr): Likewise.
30369 (ipa_compute_jump_functions_for_edge): Use the above functions to
30370 construct bits and vr parts of jump functions.
30371 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
30372 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30374 (ipcp_grow_transformations_if_necessary): Also allocate
30375 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
30377 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
30378 them. Fix too long lines.
30379 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
30380 vr_known being removed.
30381 (ipa_read_jump_function): Use new setter functions to construct bits
30382 and vr parts of jump functions or set them to NULL.
30383 (write_ipcp_transformation_info): Adjust for bits being pointers.
30384 (read_ipcp_transformation_info): Likewise.
30385 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
30387 Include gt-ipa-prop.h.
30388 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
30390 (ipcp_store_bits_results): Likewise.
30391 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
30392 Do not write to existing jump functions but use a temporary instead.
30394 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30397 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
30398 attempt to use its first operand as BIT_FIELD_REF base.
30400 2017-03-01 Richard Biener <rguenther@suse.de>
30402 PR middle-end/79721
30403 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
30404 interpolating formula in wrapping arithmetic.
30405 (chrec_apply): Convert chrec_evaluate return value to wanted type.
30407 2017-03-01 Jakub Jelinek <jakub@redhat.com>
30409 PR tree-optimization/79734
30410 * tree-vect-generic.c (expand_vector_condition): Optimize
30411 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
30412 Handle VEC_COND_EXPR where comparison has different inner width from
30413 type's inner width.
30415 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
30417 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
30418 markup, and similar issues. Remove @opindex entries for things
30419 that aren't options. Add missing -mmpy-option entries.
30421 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30423 PR tree-optimization/79737
30424 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
30425 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
30426 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
30427 instead of byte_size. Formatting fix.
30428 (shift_bytes_in_array_right): Formatting fix.
30430 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
30433 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
30434 condition on optimize for the leaf function test.
30436 2017-02-28 Martin Liska <mliska@suse.cz>
30439 * read-rtl-function.c (function_reader::handle_unknown_directive):
30440 Bail out when one uses -flto.
30442 2017-02-28 Martin Liska <mliska@suse.cz>
30444 * common.opt: Replace space with tabular for options of <number>
30446 * config/i386/i386.opt: Show <number> value for
30447 -mlarge-data-threshold.
30448 * opts.c (print_filtered_help): Do not display number in hexadecimal
30451 2017-02-28 Martin Liska <mliska@suse.cz>
30453 * common.opt: Fix --help=option -Q for options which are of
30456 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
30458 * config/i386/i386.c (print_reg): Error out for values
30459 of 8-bit size in invalid integer register.
30461 2017-02-28 Martin Sebor <msebor@redhat.com>
30463 PR tree-optimization/79691
30464 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
30466 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30469 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
30470 gcc_unreachable with output_operand_lossage.
30472 2017-02-28 Richard Biener <rguenther@suse.de>
30474 PR tree-optimization/79740
30475 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
30477 (visit_nary_op): Insert the nary into the hashtable if we
30478 pattern-matched sth.
30479 * tree-ssa-pre.c (eliminate_insert): Robustify.
30481 2017-02-28 Richard Biener <rguenther@suse.de>
30483 PR middle-end/79731
30484 * fold-const.c (decode_field_reference): Reject out-of-bound
30487 2017-02-28 Jakub Jelinek <jakub@redhat.com>
30489 * config/i386/i386.c: Include intl.h.
30490 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
30491 instead of just cond ? "..." : "...".
30492 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
30493 * coverage.c (read_counts_file): Likewise.
30494 * omp-offload.c: Include intl.h.
30495 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
30496 of just cond ? "..." : "...".
30497 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
30498 of just cond ? "..." : "...".
30500 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
30503 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
30505 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
30506 'tune for' CPU name.
30507 * config/arm/arm-cpu-data.h: Regenerated.
30509 2017-02-28 Richard Biener <rguenther@suse.de>
30511 PR tree-optimization/79732
30512 * tree-inline.c (expand_call_inline): Do not shadow var.
30514 2017-02-28 Richard Biener <rguenther@suse.de>
30516 PR tree-optimization/79723
30517 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
30518 address-space properly.
30520 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
30522 * doc/optinfo.texi (Optimization groups): Fix option used for
30524 * doc/invoke.texi (-fopt-info): Document "omp".
30525 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
30526 (OPTGROUP_ALL): Add OPTGROUP_OMP.
30527 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
30528 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
30529 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
30531 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
30533 * dumpfile.c (optgroup_options): Instead of "openmp", associate
30534 OPTGROUP_OMP with "omp".
30536 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
30539 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
30540 for arithmetic shift of unsigned V2DI.
30542 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
30544 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
30545 arc/linux.h headers.
30546 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
30547 (LINK_SPEC): Likewise.
30548 (ARC_TLS_EXTRA_START_SPEC): Likewise.
30549 (EXTRA_SPECS): Likewise.
30550 (STARTFILE_SPEC): Likewise.
30551 (ENDFILE_SPEC): Likewise.
30552 (LIB_SPEC): Likewise.
30553 (TARGET_SDATA_DEFAULT): Likewise.
30554 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
30555 (MULTILIB_DEFAULTS): Likewise.
30556 (DWARF2_UNWIND_INFO): Likewise.
30557 * config/arc/big.h: New file.
30558 * config/arc/elf.h: Likewise.
30559 * config/arc/linux.h: Likewise.
30560 * config/arc/t-uClibc: Remove.
30562 2017-02-27 Bin Cheng <bin.cheng@arm.com>
30564 PR tree-optimization/77536
30565 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
30566 (tree_transform_and_unroll_loop): Use above function to compute the
30567 estimated niter of unrolled loop and use it when scaling profile.
30568 Also use count info rather than frequency if it's non-zero.
30569 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
30570 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
30571 (vect_transform_loop): Call above function.
30573 2017-02-27 Richard Biener <rguenther@suse.de>
30575 PR tree-optimization/45397
30576 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
30577 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
30578 (visit_nary_op): Add pattern matching for CSEing sign-changed
30579 or truncated operations with wider ones.
30581 2017-02-27 Richard Biener <rguenther@suse.de>
30583 PR tree-optimization/79690
30584 * tree-vect-stmts.c (vectorizable_store): Use vector type
30585 built from the DR with address-space.
30587 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
30589 * doc/invoke.texi (Optimize Options): Refine the description
30590 of asan-use-after-return.
30592 2017-02-25 Alan Modra <amodra@gmail.com>
30594 PR rtl-optimization/79584
30595 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
30596 base, not ad->base_term, the reg within base. Remove assertion
30597 that ad->base == ad->base_term. Replace gen_int_mode using
30598 bogus mode with const0_rtx.
30600 2017-02-25 Jakub Jelinek <jakub@redhat.com>
30602 PR middle-end/79396
30603 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
30604 FMA_EXPR like tcc_binary or tcc_unary.
30606 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
30609 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
30611 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
30612 (output_loc_operands): Handle DW_OP_call_ref and
30613 DW_OP_GNU_variable_value.
30614 (struct variable_value_struct): New type.
30615 (struct variable_value_hasher): Likewise.
30616 (variable_value_hash): New variable.
30617 (string_types): Remove.
30618 (copy_loc_descr): New function.
30619 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
30620 (prepend_loc_descr_to_each): New function.
30621 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
30622 instead of add_loc_descr_to_each if the first argument is single
30623 location list and the second has multiple.
30624 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
30625 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
30626 when looking for variable value which doesn't have other location info.
30627 (loc_list_from_tree): Formatting fix.
30628 (gen_array_type_die): Simplify DW_AT_string_length handling.
30629 (adjust_string_types): Remove.
30630 (gen_subprogram_die): Don't call adjust_string_types nor test/set
30631 string_types. Call resolve_variable_values.
30632 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
30633 (resolve_addr_in_expr): Likewise. Add A argument.
30634 (copy_deref_exprloc): Remove deref argument. Adjust for the
30635 original expression being DW_OP_GNU_variable_value with optionally
30636 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
30637 optionally after it.
30638 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
30639 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
30640 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
30641 (variable_value_hasher::hash, variable_value_hasher::equal): New
30643 (resolve_variable_value_in_expr, resolve_variable_value,
30644 resolve_variable_values, note_variable_value_in_expr,
30645 note_variable_value): New functions.
30646 (dwarf2out_early_finish): Call note_variable_value on all toplevel
30649 2017-02-24 Jakub Jelinek <jakub@redhat.com>
30652 * opts.h (handle_generated_option): Add GENERATED_P argument.
30653 * opts-common.c (handle_option): Adjust function comment.
30654 (handle_generated_option): Add GENERATED_P argument, pass it to
30656 (control_warning_option): Pass false to handle_generated_option
30658 * opts.c (maybe_default_option): Pass true to handle_generated_option
30660 * optc-gen.awk: Likewise.
30662 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30664 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
30665 a REG, look at the REG it is a SUBREG of.
30666 (splitter for cmpeqsi_t): Ditto.
30668 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30670 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
30671 the special USEs with the pattern of the insn, not the insn itself.
30673 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
30676 * doc/invoke.texi: Document -mload-store-pairs.
30678 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30679 Sandra Loosemore <sandra@codesourcery.com>
30681 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
30682 argument isn't a CONST_INT.
30683 (nios2_alternate_compare_const): Assert op is a CONST_INT.
30684 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
30685 (nios2_validate_compare): Bypass alternate compare logic if *op2
30686 is not a CONST_INT.
30687 (ldstwm_operation_p): Return false if first_base is not a REG or
30688 if first_offset is not a CONST_INT.
30690 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30692 * config/cris/cris.md: Use correct operand in a define_peephole2.
30694 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30696 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
30698 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
30700 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
30701 this_insn if it is an INSN or JUMP_INSN.
30702 (force_offsettable): Look at base, not at addr.
30703 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
30704 on things that aren't necessarily CONST_INTs.
30706 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
30708 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
30709 -mfpmath=sse is the default also for x86-32 targets with SSE2
30710 instruction set when @option{-ffast-math} is enabled
30712 2017-02-24 Jeff Law <law@redhat.com>
30714 PR rtl-optimizatoin/79286
30715 * ira.c (update_equiv_regs): Drop may_trap_p exception to
30718 2017-02-24 Richard Biener <rguenther@suse.de>
30720 PR tree-optimization/79389
30721 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
30724 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
30726 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
30727 function comment to reflect reality.
30728 (loop_exits_before_overflow): Fix typo in function description.
30730 2017-02-24 Richard Biener <rguenther@suse.de>
30732 PR tree-optimization/79389
30733 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
30734 properly that a threading opportunity exists. Detect conditional
30735 copy/constant propagation opportunities.
30737 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
30739 * config/visium/visium.md (type): Add trap.
30740 (b): New mode attribute.
30741 (*btst): Rename into...
30742 (*btst<mode>): ...this and adjust.
30743 (*cbranchsi4_btst_insn): Rename into...
30744 (*cbranch<mode>4_btst_insn): ...this and adjust.
30745 (trap): New define_insn.
30747 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30749 PR tree-optimization/79389
30750 * ifcvt.c (struct noce_if_info): Add rev_cond field.
30751 (noce_reversed_cond_code): New function.
30752 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
30753 reversed_comparison_code. Formatting fix.
30754 (noce_try_store_flag): Test rev_cond != NULL in addition to
30755 reversed_comparison_code.
30756 (noce_try_store_flag_constants): Likewise.
30757 (noce_try_store_flag_mask): Likewise.
30758 (noce_try_addcc): Use rev_cond if non-NULL instead of
30759 reversed_comparison_code.
30760 (noce_try_cmove_arith): Likewise. Formatting fixes.
30761 (noce_try_minmax, noce_try_abs): Clear rev_cond.
30762 (noce_find_if_block): Initialize rev_cond.
30763 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
30764 instead of false as last argument never attempt to reverse it
30767 2017-02-23 Bin Cheng <bin.cheng@arm.com>
30769 PR tree-optimization/79663
30770 * tree-predcom.c (combine_chains): Process refs in reverse order
30771 only for ZERO length chains, and add explaining comment.
30773 2017-02-23 Jeff Law <law@redhat.com>
30775 PR tree-optimization/79578
30776 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
30777 in call to operand_equal_p.
30779 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
30782 * config/i386/cpuid.h: Fix another undefined behavior.
30784 2017-02-23 Richard Biener <rguenther@suse.de>
30786 PR tree-optimization/79683
30787 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
30788 vector types for data-refs.
30790 2017-02-23 Martin Liska <mliska@suse.cz>
30792 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
30794 2017-02-23 Jakub Jelinek <jakub@redhat.com>
30796 PR middle-end/79665
30797 * internal-fn.c (get_range_pos_neg): Moved to ...
30798 * tree.c (get_range_pos_neg): ... here. No longer static.
30799 * tree.h (get_range_pos_neg): New prototype.
30800 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
30801 are known to be in between 0 and signed maximum inclusive, try to
30802 expand both unsigned and signed divmod and use the cheaper one from
30805 2017-02-22 Jeff Law <law@redhat.com>
30807 PR tree-optimization/79578
30808 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
30809 to compare base operands.
30811 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
30814 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
30815 gpc_reg_operand instead of fpr_reg_operand.
30817 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
30819 * config/mips/mips.c (mips_return_in_memory): Force FP
30820 vector types to be returned in memory for o32 ABI.
30822 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30824 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
30825 instead of DW_TAG_member for static data member declarations and don't
30826 set no_linkage_name for static inline data members.
30827 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
30830 2017-02-22 Martin Liska <mliska@suse.cz>
30832 * doc/invoke.texi: Replace inequality signs with square brackets
30835 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
30838 * lra-constraints.c (simplify_operand_subreg): Handle
30839 WORD_REGISTER_OPERATIONS targets.
30841 2017-02-22 Jakub Jelinek <jakub@redhat.com>
30844 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
30845 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
30846 elimination by swapping fld*.
30848 2017-02-22 Richard Biener <rguenther@suse.de>
30850 PR tree-optimization/79673
30851 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
30852 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
30853 irrelevant address-space qualifiers and avoiding a
30854 ADDR_SPACE_CONVERT_EXPR from fold_convert.
30856 2017-02-22 Richard Biener <rguenther@suse.de>
30858 PR tree-optimization/79666
30859 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
30860 to not symbolically negate if that may introduce undefined
30863 2017-02-22 Martin Liska <mliska@suse.cz>
30866 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
30867 * data-streamer-out.c (streamer_write_gcov_count_stream):
30869 * value-prof.c (stream_out_histogram_value): Make assert more
30870 precise based on type of counter.
30872 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
30875 * config/i386/i386.md (standard_x87sse_constant_load splitter):
30876 Use nonimmediate_operand instead of memory_operand for operand 1.
30877 (float-extend standard_x87sse_constant_load splitter): Ditto.
30879 2017-02-21 Jeff Law <law@redhat.com>
30881 PR tree-optimization/79621
30882 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
30883 blocks with edges to themselves.
30885 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30888 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
30889 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
30890 Use gimple_call_builtin_p.
30893 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
30894 on temporarily removed DEBUG_INSNs.
30896 PR tree-optimization/79649
30897 * tree-loop-distribution.c (classify_partition): Give up on
30898 non-generic address space loads/stores.
30900 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
30902 * doc/loop.texi (Loop manipulation): Remove nonexistent
30903 tree_ssa_loop_version from the documentation.
30904 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
30906 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30909 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
30910 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
30911 * config/rs6000/rs6000.c: Include except.h.
30912 (rs6000_expand_split_stack_prologue): Call
30913 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
30915 2017-02-21 Martin Jambor <mjambor@suse.cz>
30918 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
30919 have been analyzed.
30921 2017-02-21 Martin Jambor <mjambor@suse.cz>
30923 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
30924 for backward compatibility only.
30925 * doc/invoke.texi (Option Summary): Remove all references to
30926 -fipa-cp-alignment.
30928 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
30932 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30934 * lra-constraints.c (curr_insn_transform): Handle
30935 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
30937 2017-02-21 Martin Liska <mliska@suse.cz>
30939 * config/i386/i386.opt: Replace -masm-dialect with -masm.
30941 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
30943 PR translation/79638
30944 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
30946 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
30949 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
30950 (arm_function_ok_for_sibcall): Return false for an indirect call by
30951 descriptor if all the argument registers are used.
30952 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
30953 alignment of the function.
30955 2017-02-21 Jakub Jelinek <jakub@redhat.com>
30957 PR tree-optimization/61441
30958 * simplify-rtx.c (simplify_const_unary_operation): For
30959 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
30960 the sNaN unmodified.
30962 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
30964 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
30965 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
30966 instead of SYSTEM_HEADER_DIR.
30968 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
30969 Martin Liška <mliska@suse.cz>
30971 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
30972 Fix typos and grammar, use active voice, and clarify.
30974 2017-02-20 Marek Polacek <polacek@redhat.com>
30976 PR middle-end/79537
30977 * gimplify.c (gimplify_expr): Handle unused *&&L;.
30980 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
30982 2017-02-20 Jakub Jelinek <jakub@redhat.com>
30985 * config/i386/i386.c (ix86_expand_builtin): Handle
30986 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
30987 ix86_builtins_isa[fcode].isa as a requirement of those
30988 flags and any other flag in the bitmask.
30989 (ix86_init_mmx_sse_builtins): Use 0 instead of
30990 ~OPTION_MASK_ISA_64BIT as mask.
30991 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
30992 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
30993 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
30994 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
30996 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
30999 * lra-constraints.c (split_reg): Check requested split mode
31000 is supported by the register.
31002 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31004 * lra-constraints.c (simplify_operand_subreg): Remove early
31007 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31010 * lra-constraints.c (curr_insn_transform): Tighten condition
31011 for converting SUBREG reloads from OP_OUT to OP_INOUT.
31013 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
31016 * lra-constraints.c (curr_insn_transform): Handle
31017 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
31019 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
31022 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
31024 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
31026 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
31029 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
31032 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31034 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
31037 2017-02-18 Jakub Jelinek <jakub@redhat.com>
31039 * final.c (last_columnnum, override_columnnum): New variables.
31040 (final_start_function): Set last_columnnum, pass it to begin_prologue
31041 hook and pass 0 to dwarf2out_begin_prologue.
31042 (final_scan_insn): Update override_columnnum. Pass last_columnnum
31043 to source_line debug hook.
31044 (notice_source_line): Compute last_columnnum and for debug_column_info
31045 return true on column changes.
31046 * debug.h (struct gcc_debug_hooks): Add column argument to
31047 source_line and begin_prologue hooks.
31048 (debug_nothing_int_charstar_int_bool): Remove prototype.
31049 (debug_nothing_int_int_charstar,
31050 debug_nothing_int_int_charstar_int_bool): New prototypes.
31051 (dwarf2out_begin_prologue): Add column argument.
31052 * debug.c (do_nothing_debug_hooks): Adjust source_line and
31053 begin_prologue hooks.
31054 (debug_nothing_int_charstar_int_bool): Remove.
31055 (debug_nothing_int_int_charstar,
31056 debug_nothing_int_int_charstar_int_bool): New functions.
31057 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
31058 through to dwarf2out_source_line.
31059 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
31060 (dwarf2out_source_line): Add column argument, emit it if requested.
31061 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
31063 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31064 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
31065 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
31066 through to dwarf2out_begin_prologue.
31067 (vmsdbgout_source_line): Add column argument, pass it through to
31068 dwarf2out_source_line.
31069 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
31070 dbxout_source_line caller.
31071 (dbxout_source_line): Add column argument.
31073 * common.opt (gno-column-info, gcolumn-info): New options.
31074 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
31075 (check_die): Also test for multiple DW_AT_decl_column attributes.
31076 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
31077 DW_AT_decl_column if requested.
31078 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
31080 (gen_variable_die): Likewise.
31081 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
31082 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
31085 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
31086 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
31087 (ix86_handle_option): Handle OPT_m3dnowa.
31088 * doc/invoke.texi (-m3dnowa): Document.
31089 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
31090 -m3dnowa instead of -m3dnow -march=athlon.
31093 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
31094 instead of gcc_assert for K, r and R code checks. Formatting fixes.
31096 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31099 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
31100 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
31101 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
31102 generator for vsx_xxpermdi_<mode>_be.
31103 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
31104 force big-endian semantics.
31105 (vsx_xxpermdi_<mode>_be): New define_expand with same
31106 implementation as previous version of vsx_xxpermdi_<mode>.
31108 2017-02-17 Jakub Jelinek <jakub@redhat.com>
31110 PR tree-optimization/79327
31111 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
31112 variable, its initialization and use.
31114 2017-02-17 Julia Koval <julia.koval@intel.com>
31116 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
31117 (OPTION_MASK_ISA_PKU_UNSET): New.
31118 (ix86_handle_option): Handle -mrdpid.
31119 * config/i386/cpuid.h (bit_RDPID): New.
31120 * config/i386/driver-i386.c (host_detect_local_cpu):
31121 Detect RDPID feature.
31122 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
31123 * config/i386/i386-c.c (ix86_target_macros_internal):
31125 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
31126 (ix86_valid_target_attribute_inner_p): Add "rdpid".
31127 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
31128 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
31129 * config/i386/i386.md (define_insn "rdpid"): New.
31130 * config/i386/i386.opt Add -mrdpid.
31131 * config/i386/immintrin.h (_rdpid_u32): New.
31133 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
31135 PR rtl-optimization/79541
31136 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
31137 instead of transforming it into USE.
31139 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
31141 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
31142 If HONOR_SNANS (SFmode) force the input to a register.
31143 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
31144 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
31145 an frsp or similar insn.
31147 2017-02-17 Martin Liska <mliska@suse.cz>
31149 PR rtl-optimization/79577
31150 * params.def (selsched-max-sched-times): Increase minimum to 1.
31152 2017-02-17 Martin Liska <mliska@suse.cz>
31154 PR rtl-optimization/79574
31155 * gcse.c (want_to_gcse_p): Prevent integer overflow.
31157 2017-02-17 Martin Liska <mliska@suse.cz>
31159 PR tree-optimization/79529
31160 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
31161 ssa_defined_default_def_p to handle cases which are implicitly
31163 * tree-ssa.c (ssa_defined_default_def_p): New function.
31164 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
31165 which are implicitly defined.
31166 * tree-ssa.h (ssa_defined_default_def_p): Declare.
31168 2017-02-17 Richard Biener <rguenther@suse.de>
31170 PR middle-end/79576
31171 * params.def (max-ssa-name-query-depth): Limit to 10.
31173 2017-02-17 Richard Biener <rguenther@suse.de>
31175 PR tree-optimization/79552
31176 * tree-ssa-structalias.c (visit_loadstore): Properly verify
31179 2017-02-17 Richard Biener <rguenther@suse.de>
31182 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
31184 2017-02-17 Marek Polacek <polacek@redhat.com>
31186 PR middle-end/79536
31187 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
31188 (fold_negate_expr): New wrapper.
31190 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
31192 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
31193 Correct terminology and de-emphasize pre-standard behavior.
31195 2017-02-16 Alan Modra <amodra@gmail.com>
31197 PR rtl-optimization/79286
31198 * ira.c (def_dominates_uses): New function.
31199 (update_equiv_regs): Don't create an equivalence for insns that
31200 may trap where the register def does not dominate the use.
31202 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
31204 PR rtl-optimization/78127
31205 * lra.c (lra): Call lra_eliminate before finish the loop after
31208 2017-02-16 Richard Biener <rguenther@suse.de>
31210 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
31212 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
31213 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
31214 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
31215 (isl_val_int_from_wi): New function.
31216 (extract_affine_gmp): Rename to ...
31217 (extract_affine_wi): ... this, take a widest_int.
31218 (extract_affine_int): Just wrap extract_affine_wi.
31219 (add_param_constraints): Use isl_val_int_from_wi.
31220 (add_loop_constraints): Likewise, and extract_affine_wi.
31222 2017-02-15 Jeff Law <law@redhat.com>
31224 PR middle-end/79521
31225 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
31226 ira_init_register_move_cost_if_necessary.
31228 2017-02-15 Martin Sebor <msebor@redhat.com>
31230 PR middle-end/32003
31231 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
31232 removed in a prior commit.
31234 2017-02-15 Bin Cheng <bin.cheng@arm.com>
31236 PR tree-optimization/79347
31237 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
31238 counters during peeling.
31240 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
31242 * Makefile.in (site.exp): Remove "set ISLVER".
31244 2017-02-15 Jakub Jelinek <jakub@redhat.com>
31247 * real.c (real_from_integer): Call real_convert even for decimal.
31249 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
31252 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
31254 2017-02-14 Andrew Pinski <apinski@cavium.com>
31256 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
31257 cores and change the partno/implementer to be correct.
31258 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
31259 the 'B" as the implementer.
31260 * config/aarch64/aarch64-tune.md: Regenerate.
31262 2017-02-14 Carl Love <cel@us.ibm.com>
31264 * config/rs6000/rs6000.c: Add case statement entry to make the
31265 xvcvuxdsp built-in argument unsigned.
31266 * config/rs6000/vsx.md: Fix the source and return operand types so they
31267 match the instruction definitions from the ISA document. Fix typo
31268 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
31271 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
31274 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
31275 member early_clobber_alts.
31276 * lra-lives.c (reg_early_clobber_p): New.
31277 (process_bb_lives): Use it.
31278 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
31279 (debug_operand_data): Initialize early_clobber_alts.
31280 (setup_operand_alternative): Set up early_clobber_alts.
31281 (collect_non_operand_hard_regs): Ditto. Pass early clobber
31282 alternatives to new_insn_reg.
31283 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
31285 (lra_update_insn_regno_info): Pass the new arg.
31287 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31289 PR middle-end/79505
31290 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
31291 (new_oacc_loop_raw): Don't clear already cleared fields.
31294 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
31295 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
31296 _mm512_prefetch_i64gather_ps): New inline functions and macros.
31298 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
31301 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
31303 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
31306 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
31307 the extra instruction to the right place to store 128-bit constant
31310 2017-02-14 Martin Sebor <msebor@redhat.com>
31312 PR middle-end/79448
31313 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
31314 warning for strings of unknown length.
31316 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
31318 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
31320 2017-02-14 Jeff Law <law@redhat.com>
31323 * ira-costs.c (scan_one_insn): Initialize register move costs
31324 for pseudos seen in USE/CLOBBER insns.
31326 PR tree-optimization/79095
31327 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
31328 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
31329 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
31330 if the operands are known to be not equal, then the resulting range
31332 (intersect_ranges): If the new range is ~[0,0] and the old range is
31333 wide, then prefer ~[0,0].
31334 * tree-vrp.c (overflow_comparison_p_1): New function.
31335 (overflow_comparison_p): New function.
31336 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
31337 if NAME is used in an overflow test.
31338 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
31339 overflow check that can be expressed as an equality test, then adjust
31340 ops to be that equality test.
31342 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31344 * config/s390/s390-builtin-types.def: Remove flags argument.
31345 * config/s390/s390.c (s390_init_builtins): Likewise.
31347 2017-02-14 Martin Liska <mliska@suse.cz>
31349 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
31350 vector. Fix trailing white spaces.
31352 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
31354 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
31357 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31359 PR rtl-optimization/68664
31360 * config/arm/arm.c (arm_sched_can_speculate_insn):
31361 New function. Declare prototype.
31362 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31364 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31366 PR rtl-optimization/68664
31367 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
31369 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
31371 2017-02-14 Amit Pawar <amit.pawar@amd.com>
31373 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
31374 max skip bytes for function, loop and jump.
31376 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31378 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
31379 ABS_EXPR for gimple dump.
31381 2017-02-14 Jakub Jelinek <jakub@redhat.com>
31384 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
31386 PR tree-optimization/79408
31387 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
31388 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
31389 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
31390 also if rhs1 is INTEGER_CST.
31392 2017-02-14 Richard Biener <rguenther@suse.de>
31394 PR middle-end/79432
31395 * tree-into-ssa.c (insert_phi_nodes): When the function can
31396 have abnormal edges rewrite SSA names with broken use-def
31397 dominance out of SSA and register them for PHI insertion.
31399 2017-02-13 Martin Sebor <msebor@redhat.com>
31401 PR middle-end/79496
31402 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
31403 clearing info.nowrite flag when snprintf size argument is a range.
31405 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31407 * cprop.c (cprop_jump): Add missing space in string literal.
31408 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
31409 (get_constraint_for_component_ref): Likewise.
31410 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
31411 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
31412 * lra-constraints.c (process_alt_operands): Likewise.
31413 * ipa-inline.c (inline_small_functions): Likewise.
31414 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
31415 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
31416 * trans-mem.c (diagnose_tm_1_op): Likewise.
31417 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
31418 (grid_parallel_clauses_gridifiable): Likewise.
31420 * config/nvptx/mkoffload.c (process): Add space in between
31423 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
31424 "MOD4_SSE_REGS" and "ALL_REGS".
31426 * spellcheck.c (test_data): Add , in between "foo" and "food".
31428 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31431 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
31432 boundary crossing check and subsequent code generation agree.
31434 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31436 * config/aarch64/aarch64.c (has_memory_op): Delete.
31437 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
31440 2017-02-13 Jakub Jelinek <jakub@redhat.com>
31442 PR rtl-optimization/79388
31443 PR rtl-optimization/79450
31444 * combine.c (distribute_notes): When removing TEM_INSN for which
31445 corresponding dest has last value recorded, invalidate that last
31448 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31450 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
31451 of explicit '@'. Add missing assembly comment marker on branch costs
31454 2017-02-13 Nathan Sidwell <nathan@acm.org>
31456 * gengtype-lex.l (<in_struct>): Add '/'.
31458 2017-02-13 Martin Liska <mliska@suse.cz>
31461 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
31463 2017-02-13 Richard Biener <rguenther@suse.de>
31465 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
31467 * configure: Re-generate.
31468 * config.in: Likewise.
31469 * graphite-dependences.c: Simplify as if
31470 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
31471 * graphite-isl-ast-to-gimple.c: Likewise.
31472 * graphite-optimize-isl.c: Likewise.
31473 * graphite-poly.c: Likewise.
31474 * graphite-sese-to-poly.c: Likewise.
31475 * graphite.h: Likewise.
31476 * toplev.c: Include isl/version.h and use isl_version () for
31477 printing the ISL version.
31478 * doc/install.texi: Update ISL requirement.
31480 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31482 * doc/standards.texi (Standards): Update reference to
31485 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
31487 * doc/extend.texi (Named Address Spaces): sourceware.org now
31489 * doc/install.texi (Binaries): Ditto.
31492 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31494 * doc/cpp.texi: Replace "stringify"/"stringification" with C
31495 standard terminology "stringize"/"stringizing" throughout.
31496 * doc/cppinternals.texi: Likewise.
31498 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
31500 * doc/extend.texi: Fix some spelling mistakes and typos.
31501 * doc/invoke.texi: Likewise.
31503 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31506 * params.def (inline-min-speedup) Change from 10 to 8.
31508 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31510 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
31513 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
31516 * ipa-inline-analysis.c (get_minimal_bb): New function.
31517 (record_modified): Use it.
31518 (remap_edge_change_prob): Handle also ancestor functions.
31520 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
31522 * doc/contrib.texi (Contributors): Remove broken link into
31523 the Mauve CVS repository.
31525 2017-02-11 Jakub Jelinek <jakub@redhat.com>
31527 PR middle-end/79454
31528 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
31529 result computation whenever lhs doesn't have vector mode, not
31530 just when it has BLKmode.
31532 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31534 * doc/makefile.texi (profiledbootstrap): Refer to the
31535 installation instructions only in textual form.
31537 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31540 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
31542 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
31544 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
31545 (Specific): Update mingw-w64 reference.
31547 (Specific): Remove broken link to Renesas RX processor.
31549 2017-02-10 Richard Biener <rguenther@suse.de>
31551 * toplev.c (process_options): Do not mention obsolete graphite
31552 options when printing sorry message about missing graphite support.
31553 Mention -floop-nest-optimize.
31555 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
31557 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
31558 (vtst_p16): Likewise.
31559 (vtstq_p8): Likewise.
31560 (vtstq_p16): Likewise.
31562 (vtstq_p64): Likewise.
31563 * config/arm/arm_neon.h (vgetq_lane_p64): New.
31564 (vset_lane_p64): New.
31565 (vsetq_lane_p64): New.
31567 2017-02-10 Jakub Jelinek <jakub@redhat.com>
31569 PR tree-optimization/79411
31570 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
31571 stmt operands are SSA_NAMEs used in abnormal phis.
31572 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
31575 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31578 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
31581 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
31583 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
31585 2017-02-09 Jakub Jelinek <jakub@redhat.com>
31587 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
31591 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
31592 not arbitrary TREE_CONSTANT.
31595 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
31596 "omp declare target link" attribute unless is_global_var.
31597 * omp-offload.c (find_link_var_op): Likewise.
31599 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
31600 Chung-Lin Tang <cltang@codesourcery.com>
31602 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
31604 (gimplify_adjust_omp_clauses): Don't delete TILE.
31605 (gimplify_omp_for): Deal with TILE.
31606 * internal-fn.c (expand_GOACC_TILE): New function.
31607 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
31609 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
31610 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
31612 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
31613 avoid DIV for outermost collapse var.
31614 (expand_oacc_for): Insert tile element loop as needed. Adjust.
31615 Remove out of date comments, fix whitespace.
31616 * omp-general.c (omp_extract_for_data): Deal with tiling.
31617 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
31618 adjust OLF_DIM_BASE value.
31619 (struct omp_for_data): Add tiling field.
31620 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
31621 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
31622 for auto loops. Remove default auto determining, moved to
31623 oacc_loop_fixed_partitions.
31624 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
31625 stmts, add e_mask field.
31626 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
31627 (oacc_thread_numbers): Use oacc_dim_call.
31628 (oacc_xform_tile): New.
31629 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
31630 (finish_oacc_loop): Adjust for ifns vector.
31631 (oacc_loop_discover_walk): Append loop abstraction sites to list,
31632 add case for GOACC_TILE fns.
31633 (oacc_loop_xform_loop): Delete.
31634 (oacc_loop_process): Iterate over call list directly, and add
31635 handling for GOACC_TILE fns.
31636 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
31638 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
31639 vector partitioning to outer loops. Assign 2 partitions to loops
31640 when available. Add TILE handling.
31641 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
31642 (execite_oacc_device_lower): Process GOACC_TILE fns,
31643 ignore unknown specs.
31644 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
31645 * tree.c (omp_clause_num_ops): Adjust TILE ops.
31646 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
31648 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
31650 * configure.ac (ACX_BUGURL): Update.
31651 * configure: Regenerate.
31653 2017-02-09 Richard Biener <rguenther@suse.de>
31655 PR tree-optimization/69823
31656 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
31657 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
31659 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31661 * config/arc/arc-c.def: Add __NPS400__ definition.
31662 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
31663 (TARGET_NPS400): Define.
31665 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
31667 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
31669 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
31670 pointer, arch_info.
31671 (arc_cpu_types): Fill the arch_info field with a pointer into the
31672 arc_arch_types table.
31673 (arc_selected_cpu): Declare.
31674 * config/arc/arc.c (arc_selected_cpu): Make global.
31675 (arc_selected_arch): Delete.
31676 (arc_base_cpu): Delete.
31677 (arc_override_options): Remove references to deleted variables,
31678 update access to arch information.
31679 (ARC_OPT): Update access to arch information.
31680 (ARC_OPTX): Likewise.
31681 * config/arc/arc.h (arc_base_cpu): Remove declaration.
31682 (TARGET_ARC600): Update access to arch information.
31683 (TARGET_ARC601): Likewise.
31684 (TARGET_ARC700): Likewise.
31685 (TARGET_EM): Likewise.
31686 (TARGET_HS): Likewise.
31687 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
31690 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
31693 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
31694 condition/operands for integer GE/LE/GEU/LEU operations.
31696 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
31698 PR translation/79397
31699 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
31702 2017-02-08 Martin Jambor <mjambor@suse.cz>
31705 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
31706 whether allocation happened.
31707 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
31708 nothing was allocated.
31710 2017-02-08 Jakub Jelinek <jakub@redhat.com>
31712 PR tree-optimization/79408
31713 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
31714 constant, but SSA_NAME with a known integer range, use the minimum
31715 of that range instead of op1 to determine if modulo can be replaced
31716 with its first operand.
31718 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31720 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
31722 2017-02-08 Richard Biener <rguenther@suse.de>
31724 PR tree-optimization/71824
31725 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31726 Check all loops contained in the merged region.
31728 2017-02-07 Andrew Pinski <apinski@cavium.com>
31730 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
31732 2017-02-07 Andrew Pinski <apinski@cavium.com>
31734 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
31735 (thunderxt88): Likewise.
31736 (thunderxt81): Disable LSE and change v8.1 to v8.
31737 (thunderxt83): Likewise.
31739 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31740 Richard Biener <rguenther@suse.de>
31742 PR middle-end/79399
31743 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
31744 type from int to size_t.
31745 * ira-costs.c (struct_costs_size): Change type from int to size_t.
31747 2017-02-07 Jakub Jelinek <jakub@redhat.com>
31749 PR rtl-optimization/79386
31750 * cprop.c (bypass_conditional_jumps): Initialize
31751 bypass_last_basic_block already before splitting bbs after
31752 unconditional traps...
31753 (bypass_conditional_jumps): ... rather than here.
31756 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
31757 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
31758 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
31759 fix -masm=intel patterns.
31761 2017-02-07 Richard Biener <rguenther@suse.de>
31763 PR tree-optimization/79256
31764 PR middle-end/79278
31765 * builtins.c (get_object_alignment_2): Use min_align_of_type
31766 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
31767 and ADJUST_FIELD_ALIGN.
31769 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
31771 * doc/tm.texi: Regenerate.
31772 * stor-layout.c (layout_decl): Adjust.
31773 (update_alignment_for_field): Likewise.
31774 (place_field): Likewise.
31775 (min_align_of_type): Likewise.
31776 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
31777 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
31778 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
31779 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
31780 * config/frv/frv.c (frv_adjust_field_align): Likewise.
31781 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
31782 * config/i386/i386.c (x86_field_alignment): Likewise.
31783 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
31784 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
31785 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
31786 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
31787 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
31788 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
31792 2017-01-30 Richard Biener <rguenther@suse.de>
31794 PR tree-optimization/79256
31795 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31796 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31799 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
31801 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
31802 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
31803 builtins to SImode and emit a zero-extend, if necessary.
31805 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31807 * docs/invoke.texi (RISC-V Options): Alphabetize.
31809 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31811 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
31814 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
31816 * config/riscv/riscv.c: New file.
31817 * common/config/riscv/riscv-common.c: Likewise.
31818 * config.gcc: Likewise.
31819 * config/riscv/constraints.md: Likewise.
31820 * config/riscv/elf.h: Likewise.
31821 * config/riscv/generic.md: Likewise.
31822 * config/riscv/linux.h: Likewise.
31823 * config/riscv/multilib-generator: Likewise.
31824 * config/riscv/peephole.md: Likewise.
31825 * config/riscv/pic.md: Likewise.
31826 * config/riscv/predicates.md: Likewise.
31827 * config/riscv/riscv-builtins.c: Likewise.
31828 * config/riscv/riscv-c.c: Likewise.
31829 * config/riscv/riscv-ftypes.def: Likewise.
31830 * config/riscv/riscv-modes.def: Likewise.
31831 * config/riscv/riscv-opts.h: Likewise.
31832 * config/riscv/riscv-protos.h: Likewise.
31833 * config/riscv/riscv.h: Likewise.
31834 * config/riscv/riscv.md: Likewise.
31835 * config/riscv/riscv.opt: Likewise.
31836 * config/riscv/sync.md: Likewise.
31837 * config/riscv/t-elf-multilib: Likewise.
31838 * config/riscv/t-linux: Likewise.
31839 * config/riscv/t-linux-multilib: Likewise.
31840 * config/riscv/t-riscv: Likewise.
31841 * configure.ac: Likewise.
31842 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
31843 Waterman as RISC-V maintainers.
31844 * doc/install.texi: Add RISC-V entries.
31845 * doc/invoke.texi: Add RISC-V options section.
31846 * doc/md.texi: Add RISC-V constraints section.
31847 * configure: Regenerated.
31849 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
31852 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
31853 false values to be constant vectors with all 0 or all 1 bits set.
31854 (vcondu<mode><mode>): Likewise.
31855 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
31857 (fpmask_comparison_operator): Update comment.
31858 (vecint_comparison_operator): New predicate.
31859 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
31860 vector conditionals when the true and false values are constant
31861 vectors with all 0 bits or all 1 bits set.
31863 2017-02-06 Martin Sebor <msebor@redhat.com>
31865 PR tree-optimization/79376
31866 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
31868 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
31870 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
31871 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
31872 to simplify split condition.
31874 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31876 * omp-expand.c (oxpand_omp_atomic_fetch_op,
31877 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
31880 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
31882 PR rtl-optimization/68664
31883 * target.def (can_speculate_insn): New hook.
31884 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
31885 * doc/tm.texi: Regenerate.
31886 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
31887 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
31888 (rs6000_sched_can_speculate_insn): New function.
31890 2017-02-06 Jakub Jelinek <jakub@redhat.com>
31892 PR tree-optimization/79284
31893 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
31894 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
31895 vectorizable_mask_load_store, vectorizable_operation,
31896 vect_is_simple_cond, get_same_sized_vectype): Use it instead
31897 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
31898 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
31899 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
31900 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
31901 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
31902 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
31903 is_gimple_assign (stmt). Replace another such test with
31904 is_gimple_assign (stmt).
31906 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31909 * config/avr/avr.c (rtl-iter.h): Include it.
31910 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
31911 (avr_legitimate_combined_insn): ...and implementation.
31913 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
31915 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
31916 * config/s390/s390.c (s390_const_operand_ok)
31917 (s390_canonicalize_comparison, s390_extract_part)
31918 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
31919 (s390_contiguous_bitmask_p, s390_rtx_costs)
31920 (legitimize_pic_address): Likewise.
31921 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
31922 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
31923 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
31924 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
31925 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
31927 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
31929 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
31930 REGNO($0) == REGNO($1).
31932 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31934 * config/s390/linux.h(SIZE_TYPE): Add comment.
31936 2017-02-06 Julian Brown <julian@codesourcery.com>
31937 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
31938 Virendra Pathak <virendra.pathak@broadcom.com>
31940 * config/aarch64/aarch64-cores.def: Change the scheduler
31942 * config/aarch64/aarch64.md: Include thunderx2t99.md.
31943 * config/aarch64/thunderx2t99.md: New file.
31945 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31947 * doc/standards.texi (Go Language): Update link to language
31950 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31952 * tree-eh.c (lower_resx): Sanitize profile.
31953 (cleanup_empty_eh_move_lp): Likewise.
31955 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
31958 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
31960 * cfgloopmanip.h (loop_version): Update prototype.
31961 * modulo-sched.c (sms_schedule): Update call of loop_version.
31962 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
31963 * tree-parloops.c (gen_parallel_loop): Likewise.
31964 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
31965 * tree-ssa-loop-split.c (split_loop): Likewise.
31966 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
31967 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
31969 2017-02-05 Martin Liska <mliska@suse.cz>
31972 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
31974 (print_operand_address): Initialize a struct to zero.
31976 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31978 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
31979 garbage collector only in textual form.
31981 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
31983 * doc/extend.texi (x86 specific memory model extensions for
31984 transactional memory): Simplify a phrase.
31986 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
31989 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
31990 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
31991 (atomic_storedi_1): Likewise.
31993 2017-02-04 Jakub Jelinek <jakub@redhat.com>
31995 PR tree-optimization/79338
31996 * tree-parloops.c (gather_scalar_reductions): Don't call
31997 vect_analyze_loop_form for loop->inner before destroying loop's
32000 2017-02-03 Martin Sebor <msebor@redhat.com>
32002 PR tree-optimization/79327
32003 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
32004 when precision has resulted in leading zeros.
32005 (format_integer): Adjust the likely counter to assume an unknown
32006 argument that may be zero is non-zero.
32008 2017-02-03 Jason Merrill <jason@redhat.com>
32011 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
32012 avoid copying non-taken branch.
32014 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32016 PR tree-optimization/79340
32017 * tree-vect-loop.c (vectorizable_reduction): Release
32018 vec_defs elements after safe_splicing them into other vectors.
32021 PR tree-optimization/79327
32022 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
32023 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
32025 (format_integer): Use wide_int_to_tree instead of build_int_cst
32026 + to_?hwi. If argmin is NULL, just set argmin and argmax to
32027 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
32028 of shortest and longest sequence.
32030 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
32032 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
32033 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
32035 2017-02-03 Walter Lee <walt@tilera.com>
32038 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
32039 after initial stackframe link reg save.
32040 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
32042 2017-02-03 Jakub Jelinek <jakub@redhat.com>
32045 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
32046 wu for stxssp alternative.
32048 2017-02-03 Martin Sebor <msebor@redhat.com>
32050 PR tree-optimization/79352
32051 * gimple-fold.c (get_range_strlen): Add argument.
32052 (get_range_strlen): Change return type to bool.
32053 (get_maxval_strlen): Pass in a dummy argument.
32054 * gimple-fold.h (get_range_strlen): Change return type to bool.
32055 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
32056 * tree.h (array_at_struct_end_p): Add argument.
32057 * tree.c (array_at_struct_end_p): Handle it.
32059 2017-02-03 Martin Liska <mliska@suse.cz>
32062 * multiple_target.c (create_dispatcher_calls): Redirect edge
32063 from a caller of a dispatcher.
32064 (expand_target_clones): Make the clones local.
32065 (ipa_target_clone): Do both target clones and resolvers.
32066 (ipa_dispatcher_calls): Remove the pass.
32067 (pass_dispatcher_calls::gate): Likewise.
32068 (make_pass_dispatcher_calls): Likewise.
32069 * passes.def (pass_target_clone): Put as very first IPA early
32072 2017-02-03 Martin Liska <mliska@suse.cz>
32074 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
32075 in case of a function with ifunc attribute.
32077 2017-02-03 Martin Liska <mliska@suse.cz>
32079 * cgraph.c (cgraph_node::dump): Dump function version info.
32080 * symtab.c (symtab_node::dump_base): Add missing new line.
32082 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32084 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
32085 (ifcombine_ifandif): Use it.
32087 2017-02-03 Martin Liska <mliska@suse.cz>
32089 * doc/invoke.texi: Document default value for
32090 use-after-scope-direct-emission-threshold.
32092 2017-02-03 Martin Liska <mliska@suse.cz>
32094 PR tree-optimization/79339
32095 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
32096 (format_floating): Likewise.
32098 2017-02-03 Martin Liska <mliska@suse.cz>
32101 * ipa-prop.c (ipa_node_params_t::insert): Remove current
32103 (ipa_node_params_t::remove): Likewise.
32104 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
32105 initialization from removed ipa_node_params_t::insert.
32106 (ipa_node_params::~ipa_node_params): Move from removed
32107 ipa_node_params_t::release.
32108 * symbol-summary.h (symbol_summary::m_released): New member.
32109 Do not release a summary twice. Do not allow to call finalizer
32110 for types of a summary that live in GGC memory.
32112 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
32114 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
32117 2017-02-02 Martin Sebor <msebor@redhat.com>
32119 PR middle-end/79275
32120 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
32121 (format_string): Tighten up the range of output for non-constant
32122 strings and correct the expected range for wide non-constant strings.
32124 2017-02-02 Martin Sebor <msebor@redhat.com>
32126 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
32128 PR middle-end/32003
32129 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
32131 (-fdump-tree-@var): Add to index and document how to come up
32132 with pass-specific option and dump file names.
32133 (-fdump-passes): Clarify where to look for output.
32135 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
32137 PR middle-end/77445
32138 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
32139 statistics of the analyzed path; allow threading for speed when
32140 any of BBs along the path are optimized for speed.
32142 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
32144 PR middle-end/78468
32145 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
32146 settings of the virtual registers.
32149 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
32151 * explow.c (get_dynamic_stack_size): Take known alignment of stack
32152 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
32155 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32157 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
32158 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
32160 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32162 * config/s390/s390.md: Add missing comments with the expanded
32164 * config/s390/vector.md: Likewise.
32165 * config/s390/vx-builtins.md: Likewise.
32167 2017-02-02 Jakub Jelinek <jakub@redhat.com>
32170 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
32171 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
32172 conditions on a single line.
32174 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32176 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32177 __S390_VX__ to __VX__.
32179 2017-02-01 Andrew Pinski <apinski@cavium.com>
32181 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
32182 stmt_info to record_stmt_cost.
32183 (vect_get_known_peeling_cost): Pass stmt_info if known to
32185 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
32186 cpu_vector_cost field into
32187 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
32188 field into vec_int_stmt_cost and vec_fp_stmt_cost.
32189 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
32190 splitting of scalar_stmt_cost and vec_stmt_cost.
32191 (thunderx_vector_cost): Likewise.
32192 (cortexa57_vector_cost): LIkewise.
32193 (exynosm1_vector_cost): Likewise.
32194 (xgene1_vector_cost): Likewise.
32195 (thunderx2t99_vector_cost): Improve after the splitting of the two
32197 (aarch64_builtin_vectorization_cost): Update for the splitting of
32198 scalar_stmt_cost and vec_stmt_cost.
32200 2017-02-01 Torvald Riegel <triegel@redhat.com>
32201 Richard Henderson <rth@redhat.com>
32203 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
32204 conditional on existance of a fast atomic load.
32205 * optabs-query.c (can_atomic_load_p): New function.
32206 * optabs-query.h (can_atomic_load_p): Declare it.
32207 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
32208 no fast atomic load is available for the particular size of access.
32209 (expand_atomic_compare_and_swap): Likewise.
32210 (expand_atomic_load): Likewise.
32211 (expand_atomic_store): Likewise.
32212 (expand_atomic_fetch_op): Likewise.
32213 * testsuite/lib/target-supports.exp
32214 (check_effective_target_sync_int_128): Remove x86 because it provides
32215 no fast atomic load.
32216 (check_effective_target_sync_int_128_runtime): Likewise.
32218 2017-02-01 Richard Biener <rguenther@suse.de>
32220 * graphite.c: Include tree-vectorizer.h for find_loop_location.
32221 (graphite_transform_loops): Provide opt-info for optimized nests.
32222 * tree-parloop.c (parallelize_loops): Provide opt-info for
32223 parallelized loops.
32225 2017-02-01 Richard Biener <rguenther@suse.de>
32227 PR middle-end/79315
32228 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
32229 was not set before.
32231 2017-02-01 Richard Biener <rguenther@suse.de>
32233 PR tree-optimization/71824
32234 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
32235 Verify the loops are valid in the merged SESE region.
32236 (scop_detection::can_represent_loop_1): Check analyzing the
32237 evolution of the number of iterations in the region succeeds.
32239 2017-01-31 Ian Lance Taylor <iant@golang.org>
32241 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
32242 REG_ARGS_SIZE note to 32-bit push insns and call insn.
32244 2017-01-31 David Malcolm <dmalcolm@redhat.com>
32246 PR preprocessor/79210
32247 * input.c (get_substring_ranges_for_loc): Replace line_width
32248 assertion with error-handling.
32250 2017-01-31 Richard Biener <rguenther@suse.de>
32252 PR tree-optimization/77318
32253 * graphite-sese-to-poly.c (extract_affine): Fix assert.
32254 (create_pw_aff_from_tree): Take loop parameter.
32255 (add_condition_to_pbb): Pass loop of the condition to
32256 create_pw_aff_from_tree.
32258 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32260 * config/s390/s390.c (s390_asan_shadow_offset): New function.
32261 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
32263 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
32267 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
32269 (convert_int_to_float128): Likewise.
32270 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
32271 (convert_int_to_float128): Likewise.
32272 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
32273 (UNSPEC_IEEE128_CONVERT): Likewise.
32274 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
32275 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
32276 Use local variables for IBM extended format.
32277 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
32278 (fix_trunc<mode>si2_fprs): Likewise.
32279 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
32280 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
32281 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
32282 to know that we can now have integers of all sizes in vector
32284 (fix<uns>_<mode>di2_hw): Likewise.
32285 (float<uns>_<mode>si2_hw): Likewise.
32286 (fix_<mode>si2_hw): Likewise.
32287 (fixuns_<mode>si2_hw): Likewise.
32288 (float<uns>_<mode>di2_hw): Likewise.
32289 (float_<mode>di2_hw): Likewise.
32290 (float_<mode>si2_hw): Likewise.
32291 (floatuns_<mode>di2_hw): Likewise.
32292 (floatuns_<mode>si2_hw): Likewise.
32293 (xscvqp<su>wz_<mode>): Delete, no longer used.
32294 (xscvqp<su>dz_<mode>): Likewise.
32295 (xscv<su>dqp_<mode>): Likewise.
32296 (ieee128_mfvsrd_64bit): Likewise.
32297 (ieee128_mfvsrd_32bit): Likewise.
32298 (ieee128_mfvsrwz): Likewise.
32299 (ieee128_mtvsrw): Likewise.
32300 (ieee128_mtvsrd_64bit): Likewise.
32301 (ieee128_mtvsrd_32bit): Likewise.
32303 2017-01-31 Martin Liska <mliska@suse.cz>
32306 * ipa-prop.c (ipa_free_all_node_params): Call release method
32307 instead of ~sumbol_summary to not to trigger double times
32310 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
32312 PR tree-optimization/71691
32313 * bitmap.h (class auto_bitmap): New.
32314 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
32315 is_maybe_undefined instead of ssa_undefined_value_p.
32317 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32319 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
32320 __S390_ARCH_LEVEL__ to __ARCH__.
32322 2017-01-31 Jakub Jelinek <jakub@redhat.com>
32324 PR tree-optimization/79267
32325 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
32326 if should_remove_lhs_p is true.
32328 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
32331 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
32332 (add_alignment_attribute): New.
32333 (base_type_die): Add alignment attribute.
32334 (subrange_type_die): Likewise.
32335 (modified_type_die): Likewise.
32336 (gen_array_type_die): Likewise.
32337 (gen_descr_array_type_die: Likewise.
32338 (gen_enumeration_type_die): Likewise.
32339 (gen_subprogram_die): Likewise.
32340 (gen_variable_die): Likewise.
32341 (gen_field_die): Likewise.
32342 (gen_ptr_to_mbr_type_die): Likewise.
32343 (gen_struct_or_union_type_die): Likewise.
32344 (gen_subroutine_type_die): Likewise.
32345 (gen_typedef_die): Likewise.
32346 (base_type_cmp): Compare alignment attribute.
32348 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32351 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
32352 (setb_unsigned) New pattern for setb with CCUNS.
32353 * config/rs6000/rs6000.c (expand_block_compare): Use a different
32354 subfc./subfe sequence to avoid overflow problems. Generate a
32355 shorter sequence with cmpld/setb for power9.
32356 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
32357 for generating subfc. instruction.
32358 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
32359 now uses this instruction.
32361 2017-01-30 Ian Lance Taylor <iant@google.com>
32364 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
32365 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
32367 2017-01-30 Martin Sebor <msebor@redhat.com>
32369 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
32370 Move constant to the right of a relational operator.
32371 (get_mpfr_format_length, format_character, format_string): Ditto.
32372 (should_warn_p, maybe_warn): Same.
32374 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
32376 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
32379 * asan.c (get_translation_unit_decl): Remove function.
32380 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
32382 2017-01-30 Martin Liska <mliska@suse.cz>
32384 PR gcov-profile/79259
32385 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
32386 -fprofile-generate.
32388 2017-01-30 Martin Liska <mliska@suse.cz>
32391 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
32392 Initialize variables with NULL value.
32394 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
32397 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
32399 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
32401 2017-01-30 Richard Biener <rguenther@suse.de>
32403 PR tree-optimization/79276
32404 * tree-vrp.c (process_assert_insertions): Properly adjust common
32405 when removing a duplicate.
32407 2017-01-30 Richard Biener <rguenther@suse.de>
32409 PR tree-optimization/79256
32410 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
32411 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
32413 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
32415 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
32418 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
32419 ("*r<noxa>sbg_<mode>_sll_bitmask")
32420 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
32421 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
32422 Use contiguous_bitmask_nowrap_operand.
32424 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32427 * config/rs6000/altivec.h (vec_xl): Revise #define.
32428 (vec_xst): Likewise.
32430 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
32432 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
32434 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
32436 PR rtl-optimization/79194
32437 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
32438 traps before call to bypass_conditional_jumps.
32440 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32442 PR tree-optimization/71374
32443 * lra-constraints.c (check_conflict_input_operands): New.
32444 (match_reload): Use it.
32446 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
32449 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
32450 account to calculate conflict_set.
32452 2017-01-27 Bin Cheng <bin.cheng@arm.com>
32454 PR rtl-optimization/78559
32455 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
32456 other_insn in combine.
32458 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
32460 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
32461 uint16_type_node for BT_UINT16.
32463 2017-01-27 David Malcolm <dmalcolm@redhat.com>
32465 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
32466 "RTL Tests" to menu.
32467 (GIMPLE Tests): New node.
32468 (RTL Tests): New node.
32470 2017-01-27 Richard Biener <rguenther@suse.de>
32472 PR tree-optimization/79245
32473 * tree-loop-distribution.c (distribute_loop): Apply cost
32474 modeling also to detected patterns.
32476 2017-01-27 Richard Biener <rguenther@suse.de>
32478 PR tree-optimization/71433
32479 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
32480 (compare_assert_loc): New function.
32481 (process_assert_insertions): Sort and optimize assert locations
32482 to remove duplicates and push down identical assertions on
32483 edges to their destination block.
32485 2017-01-27 Richard Biener <rguenther@suse.de>
32487 PR tree-optimization/79244
32488 * tree-vrp.c (remove_range_assertions): Forcefully propagate
32489 out SSA names even if abnormal.
32491 2017-01-27 Jakub Jelinek <jakub@redhat.com>
32493 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
32494 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
32495 instead of MPFR_RNDN.
32497 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
32500 * arm.c (arm_option_override): Don't call build_target_option_node
32501 until after doing all option overrides.
32502 (arm_valid_target_attribute_tree): Likewise.
32504 2017-01-27 Martin Liska <mliska@suse.cz>
32506 * doc/invoke.texi (-fprofile-arcs): Document profiling support
32507 for {cd}tors and C++ {cd}tors.
32509 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
32511 * config/s390/s390.md ("*setmem_long_and")
32512 ("*setmem_long_and_31z"): Use zero_extend instead of and.
32514 2017-01-26 Martin Sebor <msebor@redhat.com>
32516 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
32519 2017-01-26 Martin Sebor <msebor@redhat.com>
32521 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
32522 HAVE_DFmode before using XFmode or DFmode.
32523 (parse_directive): Avoid using the z length modifier to avoid
32524 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
32526 PR middle-end/78703
32527 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
32528 to accept adjustment as an array.
32529 (get_int_range): New function.
32530 (struct directive): Make width and prec arrays.
32531 (directive::set_width, directive::set_precision): Call get_int_range.
32532 (format_integer, format_floating): Handle width and precision ranges.
32533 (format_string, parse_directive): Same.
32535 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32538 * dwarf2out.c (generate_skeleton_bottom_up): For children with
32539 comdat_type_p set, just clone them, but keep the children in the
32543 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
32544 which have direct callers with -fvar-tracking-assignments enabled
32546 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
32547 inside of type units.
32549 2017-01-26 Martin Sebor <msebor@redhat.com>
32551 PR middle-end/78703
32552 * gimple-ssa-sprintf.c (struct result_range): Add likely and
32554 (struct format_result): Replace number_chars, number_chars_min,
32555 and number_chars_max with a single member of struct result_range.
32557 (format_result::operator+=): Adjust.
32558 (struct fmtresult): Remove bounded. Handle likely and unlikely
32560 (fmtresult::adjust_for_width_or_precision): New function.
32561 (fmtresult:type_max_digits): New function.
32562 (bytes_remaining): Handle likely and unlikely counters.
32563 (min_bytes_remaining): Remove.
32564 (format_percent): Simplify.
32565 (format_integer, format_floating): Set likely and unlikely counters.
32566 (get_string_length, format_character, format_string): Same.
32567 (format_plain, should_warn_p): New function.
32568 (maybe_warn): Call should_warn_p. Update diagnostic messages
32569 and handle those for all directives, including plain strings.
32570 (format_directive): Handle likely and unlikely counters.
32571 Remove unnecessary quoting from diagnostics. Add an informational
32573 (add_bytes): Remove.
32574 (pass_sprintf_length::compute_format_length): Simplify.
32575 (try_substitute_return_value): Handle likely and unlikely counters.
32577 2017-01-26 Carl Love <cel@us.ibm.com>
32579 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
32580 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
32582 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
32585 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
32586 endianess for subregs into account.
32587 * lra-constraints.c (lra_constraints): Do risky transformations
32588 always on the first iteration.
32589 * lra-lives.c (check_pseudos_live_through_calls): Add arg
32590 last_call_used_reg_set.
32591 (process_bb_lives): Define and use last_call_used_reg_set.
32592 * lra.c (lra): Always continue after lra_constraints on the first
32595 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
32597 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
32599 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
32601 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32603 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
32604 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
32605 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
32606 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
32607 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
32608 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
32609 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
32610 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
32611 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
32613 2017-01-26 Marek Polacek <polacek@redhat.com>
32616 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
32617 for the third operand.
32619 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32621 PR middle-end/79236
32622 * omp-low.c (struct omp_context): Add simt_stmt field.
32623 (scan_omp_for): Return omp_context *.
32624 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
32625 context to the _simt_ SIMD stmt.
32626 (lower_omp_for): For combined SIMD with sibling _simt_
32627 SIMD, make sure to use the same decls in _looptemp_
32628 clauses as in the sibling.
32630 2017-01-26 David Sherwood <david.sherwood@arm.com>
32632 PR middle-end/79212
32633 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
32636 2017-01-26 Jakub Jelinek <jakub@redhat.com>
32639 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
32640 emit fld b; fld a; if possible.
32642 * brig-builtins.def: Update copyright years.
32643 * config/arm/arm_acle_builtins.def: Update copyright years.
32645 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
32648 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
32649 constraint instead of o for the stxsd instruction.
32651 2017-01-25 Carl Love <cel@us.ibm.com>
32653 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
32654 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
32656 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
32658 * doc/invoke.texi (C++ Dialect Options): Fix typo.
32660 2017-01-25 Richard Biener <rguenther@suse.de>
32662 PR tree-optimization/69264
32663 * target.def (vector_alignment_reachable): Improve documentation.
32664 * doc/tm.texi: Regenerate.
32665 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
32667 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
32668 earlier changes with respect to TYPE_USER_ALIGN.
32669 (vector_alignment_reachable_p): Likewise. Improve dumping.
32671 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32674 * config/arm/arm.md (xordi3): Force constant operand into a register
32677 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32679 * doc/invoke.texi (-fstore-merging): Correct default optimization
32680 levels at which it is enabled.
32681 (-O): Move -fstore-merging from list to...
32684 2017-01-25 Richard Biener <rguenther@suse.de>
32687 * omp-expand.c: Include debug.h.
32688 (expand_omp_taskreg): Make sure to generate early debug before
32689 outlining anything from a function.
32690 (expand_omp_target): Likewise.
32691 (grid_expand_target_grid_body): Likewise.
32693 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
32696 * asan.c (get_translation_unit_decl): New function.
32697 (asan_add_global): Extract modules file name from globals
32698 TRANSLATION_UNIT_DECL name.
32700 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
32703 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
32704 for long calls with APCS frame and VFP.
32706 2017-01-24 David Malcolm <dmalcolm@redhat.com>
32708 * cfg.c (original_copy_tables_initialized_p): New function.
32709 * cfg.h (original_copy_tables_initialized_p): New decl.
32710 * cfgrtl.c (relink_block_chain): Guard the call to
32711 free_original_copy_tables with a call to
32712 original_copy_tables_initialized_p.
32713 * cgraph.h (symtab_node::native_rtl_p): New decl.
32714 * cgraphunit.c (symtab_node::native_rtl_p): New function.
32715 (symtab_node::needed_p): Don't assert for early assembly output
32716 for __RTL functions.
32717 (cgraph_node::finalize_function): Set "force_output" for __RTL
32719 (cgraph_node::analyze): Bail out early for __RTL functions.
32720 (analyze_functions): Update assertion to support __RTL functions.
32721 (cgraph_node::expand): Bail out early for __RTL functions.
32722 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
32724 * function.h (struct function): Update comment for field
32726 * gimple-expr.c: Include "tree-pass.h".
32727 (gimple_has_body_p): Return false for __RTL functions.
32728 * Makefile.in (OBJS): Add run-rtl-passes.o.
32729 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
32731 (gcc::pass_manager::get_clean_slate): New accessor.
32732 * passes.c: Include "insn-addr.h".
32733 (should_skip_pass_p): Add logging. Update logic for running
32734 "expand" to be compatible with both __GIMPLE and __RTL. Guard
32735 property-provider override so it is only done for gimple passes.
32737 (skip_pass): New function.
32738 (execute_one_pass): Call skip_pass when skipping passes.
32739 * read-md.c (md_reader::read_char): Support filtering
32740 the input to a subset of line numbers.
32741 (md_reader::md_reader): Initialize fields
32742 m_first_line and m_last_line.
32743 (md_reader::read_file_fragment): New function.
32744 * read-md.h (md_reader::read_file_fragment): New decl.
32745 (md_reader::m_first_line): New field.
32746 (md_reader::m_last_line): New field.
32747 * read-rtl-function.c (function_reader::create_function): Only
32748 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
32749 curr_properties. Set DECL_INITIAL to a dummy block.
32750 (read_rtl_function_body_from_file_range): New function.
32751 * read-rtl-function.h (read_rtl_function_body_from_file_range):
32753 * run-rtl-passes.c: New file.
32754 * run-rtl-passes.h: New file.
32756 2017-01-24 Jeff Law <law@redhat.com>
32758 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
32761 2017-01-24 Bin Cheng <bin.cheng@arm.com>
32763 PR tree-optimization/79159
32764 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
32765 (record_nonwrapping_iv): Improve boundary using above function if no
32766 value range information.
32768 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
32769 Martin Jambor <mjambor@suse.cz>
32771 * brig-builtins.def: New file.
32772 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
32773 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
32774 (DEF_HSAIL_SAT_BUILTIN): Likewise.
32775 (DEF_HSAIL_INTR_BUILTIN): Likewise.
32776 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
32777 * builtin-types.def (BT_INT8): New.
32778 (BT_INT16): Likewise.
32779 (BT_UINT8): Likewise.
32780 (BT_UINT16): Likewise.
32781 (BT_FN_ULONG): Likewise.
32782 (BT_FN_UINT_INT): Likewise.
32783 (BT_FN_UINT_ULONG): Likewise.
32784 (BT_FN_UINT_LONG): Likewise.
32785 (BT_FN_UINT_PTR): Likewise.
32786 (BT_FN_ULONG_PTR): Likewise.
32787 (BT_FN_INT8_FLOAT): Likewise.
32788 (BT_FN_INT16_FLOAT): Likewise.
32789 (BT_FN_UINT32_FLOAT): Likewise.
32790 (BT_FN_UINT16_FLOAT): Likewise.
32791 (BT_FN_UINT8_FLOAT): Likewise.
32792 (BT_FN_UINT64_FLOAT): Likewise.
32793 (BT_FN_UINT16_UINT32): Likewise.
32794 (BT_FN_UINT32_UINT16): Likewise.
32795 (BT_FN_UINT16_UINT16_UINT16): Likewise.
32796 (BT_FN_INT_PTR_INT): Likewise.
32797 (BT_FN_UINT_PTR_UINT): Likewise.
32798 (BT_FN_LONG_PTR_LONG): Likewise.
32799 (BT_FN_ULONG_PTR_ULONG): Likewise.
32800 (BT_FN_VOID_UINT64_UINT64): Likewise.
32801 (BT_FN_UINT8_UINT8_UINT8): Likewise.
32802 (BT_FN_INT8_INT8_INT8): Likewise.
32803 (BT_FN_INT16_INT16_INT16): Likewise.
32804 (BT_FN_INT_INT_INT): Likewise.
32805 (BT_FN_UINT_FLOAT_UINT): Likewise.
32806 (BT_FN_FLOAT_UINT_UINT): Likewise.
32807 (BT_FN_ULONG_UINT_UINT): Likewise.
32808 (BT_FN_ULONG_UINT_PTR): Likewise.
32809 (BT_FN_ULONG_ULONG_ULONG): Likewise.
32810 (BT_FN_UINT_UINT_UINT): Likewise.
32811 (BT_FN_VOID_UINT_PTR): Likewise.
32812 (BT_FN_UINT_UINT_PTR: Likewise.
32813 (BT_FN_UINT32_UINT64_PTR): Likewise.
32814 (BT_FN_INT_INT_UINT_UINT): Likewise.
32815 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
32816 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
32817 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
32818 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
32819 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
32820 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
32821 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
32822 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
32823 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
32824 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
32825 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
32826 * doc/frontends.texi: List BRIG FE.
32827 * doc/install.texi (Testing): Add BRIG tesring requirements.
32828 * doc/invoke.texi (Overall Options): Mention BRIG.
32829 * doc/standards.texi (Standards): Doucment BRIG HSA version.
32831 2017-01-24 Richard Biener <rguenther@suse.de>
32833 PR translation/79208
32834 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
32836 2017-01-24 Martin Jambor <mjambor@suse.cz>
32839 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
32840 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
32841 and known_contexts.
32843 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
32845 PR middle-end/79123
32846 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
32847 casts from signed to unsigned really don't have a range.
32849 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
32851 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
32852 GMP_RNDx for compatiblity.
32854 2017-01-24 Martin Liska <mliska@suse.cz>
32857 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
32858 that would prevent us to call alloca with -1 as argument.
32860 2017-01-24 Jakub Jelinek <jakub@redhat.com>
32862 * dwarf2out.c (output_compilation_unit_header, output_file_names):
32863 Avoid -Wformat-security warning.
32865 2017-01-23 Andrew Pinski <apinski@cavium.com>
32867 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
32870 2017-01-23 Martin Sebor <msebor@redhat.com>
32872 PR middle-end/78703
32873 * gimple-ssa-sprintf.c (warn_level): New global.
32874 (format_integer): Use it here and throughout the rest of the file.
32875 Use the same switch to compute sign as base.
32876 (maybe_warn): New function.
32877 (format_directive): Factor out warnings into maybe_warn.
32878 Add debugging output. Use warn_level.
32879 (add_bytes): Use warn_level.
32880 (pass_sprintf_length::compute_format_length): Add debugging output.
32881 (try_substitute_return_value): Same.
32882 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
32884 PR middle-end/78703
32885 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
32886 (struct fmtresult, format_integer, format_floating): Adjust.
32887 (fmtresult::fmtresult): Set max correctly in two argument ctor.
32888 (get_string_length, format_string,format_directive): Same.
32889 (pass_sprintf_length::compute_format_length): Same.
32890 (try_substitute_return_value): Simplify slightly.
32892 PR middle-end/78703
32893 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
32894 (fmtresult::operator+=): Outlined.
32895 (struct fmtresult): Add ctors.
32896 (struct conversion_spec): Rename...
32897 (struct directive): ...to this. Add and remove data members.
32898 (directive::set_width, directive::set_precision): New functions.
32899 (format_percent): Use fmtresult ctor.
32900 (get_width_and_precision): Remove.
32901 (format_integer): Make naming changes. Avoid computing width and
32903 (format_floating): Same. Adjust indentation.
32904 (format_character, format_none): New functions.
32905 (format_string): Moved character handling to format_character.
32906 (format_directive): Remove arguments, change return type.
32907 (parse_directive): New function.
32908 (pass_sprintf_length::compute_format_length): Move directive
32909 parsing to parse_directive.
32911 2017-01-23 Jakub Jelinek <jakub@redhat.com>
32913 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
32914 (assign_assembler_name_if_needed): ... this.
32915 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
32916 (assign_assembler_name_if_needed): ... this.
32917 (free_lang_data_in_cgraph): Adjust callers.
32918 * cgraphunit.c (cgraph_node::analyze): Likewise.
32919 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
32922 2017-01-23 Richard Biener <rguenther@suse.de>
32924 PR tree-optimization/79088
32925 PR tree-optimization/79188
32926 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
32927 resetting loop bounds after last path deletion. Reset loop
32928 bounds of the target loop, make code match the comments.
32929 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
32930 Make sure loops need no fixups.
32932 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
32934 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
32935 exponent support with double type for first argument.
32936 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
32937 type returned by __builtin_vec_extract_sig,
32938 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
32939 functions from "vector int" to "vector unsigned int" or from
32940 "vector long long int" to "vector unsigned long long int".
32941 Changed type returned by __builtin_vec_extract_exp,
32942 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
32943 functions from "vector int" to "vector unsigned int" or from
32944 "vector long long int" to "vector unsigned long long int".
32945 Changed return type of __builtin_vec_test_data_class,
32946 __builtin_vec_test_data_class_sp, and
32947 __builtin_vec_test_data_class_dp from "vector int" to
32948 "vector bool int" or from "vector long long int" to "vector bool
32949 long long int" and changed second argument type from "unsigned
32950 int" to "int". Added new overloaded function forms "vector float
32951 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
32952 "vector float __builtin_vec_insert_exp_sp (vector float, vector
32953 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
32954 double, vector unsigned long long int)" and "vector double
32955 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
32956 long int)". Changed return type of
32957 __builtin_scalar_test_data_class and
32958 __builtin_scalar_test_data_class_sp and
32959 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
32960 int" and changed second argument from "unsigned int" to "int".
32961 Changed type returned by __builtin_scalar_test_neg,
32962 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
32963 from "int" to "bool int". Added new overloaded function form
32964 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
32965 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
32966 exponent double-precision with floating point first argument.
32967 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
32968 documentation of scalar_test_data_class, scalar_test_neg,
32969 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
32970 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
32971 vec_test_data_class built-in functions to reflect refinements in
32972 their type signatures.
32974 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
32976 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
32978 (aarch64_elf_asm_destructor): Likewise.
32980 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
32982 PR rtl-optimization/78634
32983 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
32984 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
32985 * ifcvt.c (noce_try_cmove): Add missing cost check.
32987 PR rtl-optimization/71724
32988 * combine.c (if_then_else_cond): Look for situations where it is
32989 beneficial to undo the work of one of the recursive calls.
32991 2017-01-23 Bin Cheng <bin.cheng@arm.com>
32993 PR tree-optimization/70754
32994 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
32995 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
32996 combined stmt before it if not NULL.
32997 (combine_chains): Process refs reversely and compute dominance point
33000 2017-01-23 Martin Liska <mliska@suse.cz>
33002 PR tree-optimization/79196
33003 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
33004 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
33006 (strlen_optimize_stmt): Call the renamed function.
33008 2017-01-23 Michael Matz <matz@suse.de>
33010 PR tree-optimization/78384
33011 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
33013 2017-01-23 Richard Biener <rguenther@suse.de>
33015 PR tree-optimization/79186
33016 * tree-vrp.c (register_new_assert_for): Make sure we've seen
33017 both incoming edges before moving an assert.
33019 2017-01-23 Martin Jambor <mjambor@suse.cz>
33021 * ipa-prop.c (load_from_param_1): Removed.
33022 (load_from_unmodified_param): Bits from load_from_param_1 put back
33024 (load_from_param): Removed.
33025 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
33026 with stmt. Reverted back to use of load_from_unmodified_param.
33028 2017-01-23 Martin Jambor <mjambor@suse.cz>
33031 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
33032 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
33033 field a pointer to garbage collected vector, mark lattices and
33034 ipcp_orig_node with GTY((skip)).
33035 (ipa_get_param_count): Adjust to descriptors being a pointer.
33036 (ipa_get_param): Likewise.
33037 (ipa_get_type): Likewise.
33038 (ipa_get_param_move_cost): Likewise.
33039 (ipa_set_param_used): Likewise.
33040 (ipa_get_controlled_uses): Likewise.
33041 (ipa_set_controlled_uses): Likewise.
33042 (ipa_is_param_used): Likewise.
33043 (ipa_node_params_t): Move into garbage collector. New methods insert
33045 (ipa_node_params_sum): Annotate wth GTY(()).
33046 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
33048 (ipa_load_from_parm_agg): Adjust declaration.
33049 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
33050 * ipa-profile.c (ipa_profile): Likewise.
33051 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
33052 (ipa_populate_param_decls): Make descriptors parameter garbage
33054 (ipa_dump_param): Adjust to descriptors being a pointer.
33055 (ipa_alloc_node_params): Likewise.
33056 (ipa_initialize_node_params): Likewise.
33057 (load_from_param_1): Make descriptors parameter garbage collected.
33058 (load_from_unmodified_param): Likewise.
33059 (load_from_param): Likewise.
33060 (ipa_load_from_parm_agg): Likewise.
33061 (ipa_node_params::~ipa_node_params): Removed.
33062 (ipa_free_all_node_params): Remove call to delete operator.
33063 (ipa_node_params_t::insert): New.
33064 (ipa_node_params_t::remove): Likewise.
33065 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
33066 copy known_csts and known_contexts vectors.
33067 (ipa_read_node_info): Adjust to descriptors being a pointer.
33068 (ipcp_modif_dom_walker): Make m_descriptors field garbage
33070 (ipcp_transform_function): Make descriptors variable garbage
33073 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
33075 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
33076 * config/i386/avx512dqintrin.h: Ditto.
33077 * config/i386/avx512fintrin.h: Ditto.
33078 * config/i386/i386.c: Handle new builtins.
33079 * config/i386/i386-builtin.def: Add new builtins.
33080 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
33081 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
33083 2017-01-23 Jakub Jelinek <jakub@redhat.com>
33084 Martin Liska <mliska@suse.cz>
33086 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
33087 * asan.c (asan_expand_poison_ifn): Support stores and use
33088 appropriate ASAN report function.
33089 * internal-fn.c (expand_ASAN_POISON_USE): New function.
33090 * internal-fn.def (ASAN_POISON_USE): Declare.
33091 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
33092 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
33093 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
33094 ASAN_POISON calls w/o LHS.
33095 * tree-ssa.c (execute_update_addresses_taken): Create clobber
33096 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
33097 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
33098 * gimplify.c (asan_poison_variables): Add attribute
33099 use_after_scope_memory to variables that really needs to live
33101 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
33102 having the attribute.
33104 2017-01-23 Martin Liska <mliska@suse.cz>
33106 * asan.c (create_asan_shadow_var): New function.
33107 (asan_expand_poison_ifn): Likewise.
33108 * asan.h (asan_expand_poison_ifn): New declaration.
33109 * internal-fn.c (expand_ASAN_POISON): Likewise.
33110 * internal-fn.def (ASAN_POISON): New builtin.
33111 * sanopt.c (pass_sanopt::execute): Expand
33112 asan_expand_poison_ifn.
33113 * tree-inline.c (copy_decl_for_dup_finish): Make function
33115 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
33116 * tree-ssa.c (is_asan_mark_p): New function.
33117 (execute_update_addresses_taken): Rewrite local variables
33118 (identified just by use-after-scope as addressable) into SSA.
33120 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33122 * doc/install.texi (Specific): opensource.apple.com uses https
33123 now. Remove trailing slash.
33125 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
33127 * README.Portability: Remove note on an Irix compatibility issue.
33129 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
33131 * gcov.c (INCLUDE_ALGORITHM): Define.
33132 (INCLUDE_VECTOR): Define.
33133 No longer include <vector> and <algorithm> directly.
33135 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
33137 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
33139 * doc/invoke.texi (Code Gen Options): Ditto.
33141 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
33144 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
33146 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
33148 rtl-optimization/79125
33149 * cprop.c (local_cprop_pass): Handle cases where we make an
33150 unconditional trap.
33152 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
33156 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
33157 read from, for big endian.
33159 2017-01-20 Jiong Wang <jiong.wang@arm.com>
33161 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
33162 register pauth builtins for LP64 only.
33164 2017-01-20 Marek Polacek <polacek@redhat.com>
33167 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
33170 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33172 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
33174 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
33175 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
33176 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
33178 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33181 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
33182 in big-endian mode when they are not a single duplicated value.
33184 2017-01-20 Richard Biener <rguenther@suse.de>
33186 * BASE-VER: Bump to 7.0.1.
33188 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
33190 * omp-low.c (omplow_simd_context): New struct. Use it...
33191 (lower_rec_simd_input_clauses): ...here and...
33192 (lower_rec_input_clauses): ...here to hold common data. Adjust all
33193 references to idx, lane, max_vf, is_simt.
33195 2017-01-20 Graham Markall <graham.markall@embecosm.com>
33197 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
33200 2017-01-20 Martin Jambor <mjambor@suse.cz>
33202 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
33203 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
33205 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
33206 (GTFILES): Rename hsa.c to hsa-common.c.
33207 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
33208 * hsa-dump.c: Likewise.
33209 * hsa-gen.c: Likewise.
33210 * hsa-regalloc.c: Likewise.
33211 * ipa-hsa.c: Likewise.
33212 * omp-expand.c: Likewise.
33213 * omp-low.c: Likewise.
33214 * toplev.c: Likewise.
33216 2017-01-20 Marek Polacek <polacek@redhat.com>
33219 * doc/invoke.texi: Document -Wduplicated-branches.
33220 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
33221 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
33222 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
33223 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
33224 return 0 only when not OEP_LEXICOGRAPHIC.
33225 (fold_build_cleanup_point_expr): Use the expression
33226 location when building CLEANUP_POINT_EXPR.
33227 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
33228 * tree.c (add_expr): Handle error_mark_node.
33230 2017-01-20 Martin Liska <mliska@suse.cz>
33233 * tree-profile.c (init_ic_make_global_vars): Do not call
33235 (gimple_init_gcov_profiler): Likewise.
33237 2017-01-20 Martin Liska <mliska@suse.cz>
33240 * cgraph.h (maybe_create_reference): Remove argument and
33242 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
33244 * ipa-cp.c (create_specialized_node): Likewise.
33245 * symtab.c (symtab_node::maybe_create_reference): Handle
33246 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
33248 2017-01-20 Martin Liska <mliska@suse.cz>
33250 * read-rtl-function.c (function_reader::create_function): Use
33251 build_decl instread of build_decl_stat.
33253 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
33255 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
33256 * config/i386/avx512dqintrin.h: Ditto.
33257 * config/i386/avx512fintrin.h: Ditto.
33258 * config/i386/i386-builtin-types.def: Add new types.
33259 * config/i386/i386.c: Handle new types.
33260 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
33261 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
33262 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
33263 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
33264 (__builtin_ia32_kshiftridi): New.
33265 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
33267 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
33271 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
33272 define to rs6000_init_stack_protect_guard.
33273 (rs6000_init_stack_protect_guard): New function.
33275 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33276 Yunqiang Su <yunqiang.su@imgtec.com>
33278 * config.gcc (supported_defaults): Add madd4.
33279 (with_madd4): Add validation.
33280 (all_defaults): Add madd4.
33281 * config/mips/mips.opt (mmadd4): New option.
33282 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33284 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33286 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
33287 (ISA_HAS_FUSED_MADD4): Likewise.
33288 * doc/invoke.texi (-mmadd4): Document the new option.
33289 * doc/install.texi (--with-madd4): Document the new option.
33291 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33293 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
33294 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
33295 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
33296 (aarch64_init_pauth_hint_builtins): New.
33297 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
33298 (aarch64_expand_builtin): Expand new builtins.
33300 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33302 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
33303 * combine-stack-adj.c (no_unhandled_cfa): Handle
33304 REG_CFA_TOGGLE_RA_MANGLE.
33305 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
33306 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
33307 info for return address signing.
33308 (aarch64_expand_epilogue): Likewise.
33310 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33312 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
33313 * config/aarch64/aarch64-protos.h
33314 (aarch64_return_address_signing_enabled): New declaration.
33315 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
33317 (aarch64_expand_prologue): Sign return address before it's pushed onto
33319 (aarch64_expand_epilogue): Authenticate return address fetched from
33321 (aarch64_override_options): Sanity check for ILP32 and ISA level.
33322 (aarch64_attributes): New function attributes for "sign-return-address".
33323 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
33324 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
33325 ("*do_return"): Generate combined instructions according to key index.
33326 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
33327 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
33329 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
33330 * config/aarch64/aarch64.opt (msign-return-address=): New.
33331 * doc/extend.texi (AArch64 Function Attributes): Documents
33332 "sign-return-address=".
33333 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
33335 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
33337 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
33338 overall option summary.
33340 2017-01-19 Jiong Wang <jiong.wang@arm.com>
33342 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
33343 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
33344 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
33345 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
33347 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
33349 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
33350 -mpower9-minmax by default for -mcpu=power9.
33351 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
33352 128-bit floating point.
33354 2017-01-20 Alan Modra <amodra@gmail.com>
33356 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
33357 optimizing for size.
33359 2017-01-20 Alan Modra <amodra@gmail.com>
33362 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
33363 for strcmp and strncmp from corresponding builtin decl.
33365 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33367 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
33368 instead of i386/rtems-64.h.
33369 * config/i386/rtems-64.h: Remove.
33371 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
33375 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
33377 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
33379 2017-01-19 Tamar Christina <tamar.christina@arm.com>
33381 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
33382 Change int to HOST_WIDE_INT.
33383 * config/aarch64/aarch64-protos.h
33384 (aarch64_simd_gen_const_vector_dup): Likewise.
33385 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
33387 2017-01-19 David Malcolm <dmalcolm@redhat.com>
33389 * langhooks-def.h (lhd_type_for_size): New decl.
33390 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
33391 * langhooks.c (lhd_type_for_size): New function, taken from
33394 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
33396 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
33397 define_bypass for CR latency.
33398 (power9-cracked-alu): Update bypass latency and remove power9-branch.
33399 (power9-alu2): Add define_bypass for CR latency.
33401 (power9-mul): Update insn latency.
33402 (power9-mul-compare): Update insn latency, bypass latency and remove
33405 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33407 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
33409 * config/aarch64/aarch64.md
33410 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
33411 aarch64_nopcrelative_literal_loads.
33412 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
33414 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
33416 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
33417 TARGET_LOONGSON_3A.
33418 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
33420 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
33423 * config.gcc (supported_defaults): Add lxc1-sxc1.
33424 (with_lxc1_sxc1): Add validation.
33425 (all_defaults): Add lxc1-sxc1.
33426 * config/mips/mips.opt (mlxc1-sxc1): New option.
33427 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
33429 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
33430 __mips_no_lxc1_sxc1.
33431 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
33432 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
33433 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
33435 2017-01-19 Richard Biener <rguenther@suse.de>
33437 PR tree-optimization/72488
33438 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
33439 sure to restore SSA info.
33440 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
33442 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
33444 PR rtl-optimization/79121
33445 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
33446 of the inner type when shifting an extended value.
33448 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33451 * symtab.c (symtab_node::equal_address_to): Fix comparing of
33452 interposable aliases.
33454 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
33457 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
33458 Use the evmergelohi instruction.
33459 (mov_si<mode>_e500_subreg4_2_le): Likewise.
33460 (mov_sitf_e500_subreg8_2_be): Likewise.
33461 (mov_sitf_e500_subreg12_2_le): Likewise.
33462 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
33463 (mov_si<mode>_e500_subreg4_2_be): Likewise.
33464 (mov_sitf_e500_subreg8_2_le): Likewise.
33465 (mov_sitf_e500_subreg12_2_be): Likewise.
33467 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33469 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
33470 attribute from vecsimple to vecperm.
33471 (altivec_vbpermq2): Likewise.
33473 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33476 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
33478 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33479 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
33480 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
33481 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
33482 case where N arg is SIZE_MAX.
33483 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
33484 (cmpstrsi): Add pattern.
33486 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
33488 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
33489 __builtin_vec_revb builtins.
33490 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
33491 built-in functions to support generation of the ISA 3.0 XXBR<x>
33492 vector byte reverse instructions.
33493 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
33494 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
33495 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
33496 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
33497 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
33498 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
33499 (P9V_BUILTIN_VEC_REVB): Likewise.
33500 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
33501 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
33502 (p9_xxbrq_v16qi): Likewise.
33503 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
33504 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
33505 (p9_xxbrh_v8hi): Likewise.
33506 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
33507 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
33508 vec_revb built-in functions.
33510 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
33512 PR rtl-optimization/78952
33513 * config/i386/i386.md (any_extract): New code iterator.
33514 (*insvqi_2): Use any_extract for source operand.
33515 (*insvqi_3): Use any_shiftrt for source operand.
33517 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
33519 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
33521 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
33523 2017-01-18 Matthias Klose <doko@ubuntu.com>
33525 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
33527 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33529 * config/rs6000/altivec.h (vec_bperm): Change #define.
33530 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
33531 (altivec_vbpermq2): New define_insn.
33532 (altivec_vbpermd): Likewise.
33533 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
33534 function interface.
33535 (VBPERMD): Likewise.
33536 (VBPERM): New polymorphic function interface.
33537 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
33538 Add entries for P9V_BUILTIN_VEC_VBPERM.
33539 * doc/extend.texi: Add interfaces for vec_bperm.
33541 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33543 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
33544 first letter of error messages.
33545 (s390_resolve_overloaded_builtin): Likewise.
33546 * config/s390/s390.c (s390_expand_builtin): Likewise.
33547 (s390_invalid_arg_for_unprototyped_fn): Likewise.
33548 (s390_valid_target_attribute_inner_p): Likewise.
33549 * config/s390/s390.md ("tabort"): Likewise.
33551 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
33553 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
33554 (ISA_AVOID_DIV_HILO): New macro.
33555 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
33556 (ISA_HAS_DDIV): Likewise.
33558 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33560 * doc/invoke.texi (fabi-version): Correct number of occurrences.
33562 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33564 * doc/invoke.texi (fabi-version): Spelling fix.
33566 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33569 * doc/invoke.texi (fabi-version): Mention mangling fix for
33572 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
33575 * doc/invoke.texi (fabi-version): Document discriminator mangling.
33577 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
33580 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
33581 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
33583 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
33585 (stack_protect_test): Ditto.
33586 * config/rs6000/rs6000.opt (mstack-protector-guard=,
33587 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
33589 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
33590 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
33591 -mstack-protector-guard-offset=.
33592 (RS/6000 and PowerPC Options): Ditto.
33594 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33596 * config/i386/i386.h (MASK_CLASS_P): New define.
33597 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
33598 there are no registers from different register sets also when
33599 mask registers are used. Update function comment.
33600 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
33601 to (*k/*r) and (*k/*km) alternatives.
33603 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
33605 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
33606 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
33607 (EH_RETURN_HANDLER_RTX): New define.
33608 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
33609 Force frame pointer in EH return functions.
33610 (aarch64_expand_epilogue): Add barrier for eh_return.
33611 (aarch64_final_eh_return_addr): Remove.
33612 (aarch64_eh_return_handler_rtx): New function.
33613 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
33615 (aarch64_eh_return_handler_rtx): New prototype.
33617 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33619 * config/rs6000/altivec.h (vec_rlmi): New #define.
33620 (vec_vrlnm): Likewise.
33621 (vec_rlnm): Likewise.
33622 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
33623 (UNSPEC_VRLNM): Likewise.
33624 (VIlong): New mode iterator.
33625 (altivec_vrl<VI_char>mi): New define_insn.
33626 (altivec_vrl<VI_char>nm): Likewise.
33627 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
33629 (VRLDNM): Likewise.
33630 (RLNM): New polymorphic function entry.
33631 (VRLWMI): New monomorphic function entry.
33632 (VRLDMI): Likewise.
33633 (RLMI): New polymorphic function entry.
33634 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
33635 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
33636 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
33639 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33642 * dwarf2out.c (field_byte_offset): Restore the
33643 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
33644 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
33645 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
33648 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33651 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
33653 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33656 * dwarf2out.c (add_data_member_location_attribute): For constant
33657 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
33658 instead of DW_AT_data_member_location, DW_AT_bit_offset and
33659 DW_AT_byte_size attributes.
33661 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
33663 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
33664 after forcing to constant memory when the code model is medium.
33666 2017-01-17 Julia Koval <julia.koval@intel.com>
33669 * config/i386/avx512fintrin.h
33670 (_mm512_i32gather_ps): Change __addr type to void const*.
33671 (_mm512_mask_i32gather_ps): Ditto.
33672 (_mm512_i32gather_pd): Ditto.
33673 (_mm512_mask_i32gather_pd): Ditto.
33674 (_mm512_i64gather_ps): Ditto.
33675 (_mm512_mask_i64gather_ps): Ditto.
33676 (_mm512_i64gather_pd): Ditto.
33677 (_mm512_mask_i64gather_pd): Ditto.
33678 (_mm512_i32gather_epi32): Ditto.
33679 (_mm512_mask_i32gather_epi32): Ditto.
33680 (_mm512_i32gather_epi64): Ditto.
33681 (_mm512_mask_i32gather_epi64): Ditto.
33682 (_mm512_i64gather_epi32): Ditto.
33683 (_mm512_mask_i64gather_epi32): Ditto.
33684 (_mm512_i64gather_epi64): Ditto.
33685 (_mm512_mask_i64gather_epi64): Ditto.
33686 (_mm512_i32scatter_ps): Change __addr type to void*.
33687 (_mm512_mask_i32scatter_ps): Ditto.
33688 (_mm512_i32scatter_pd): Ditto.
33689 (_mm512_mask_i32scatter_pd): Ditto.
33690 (_mm512_i64scatter_ps): Ditto.
33691 (_mm512_mask_i64scatter_ps): Ditto.
33692 (_mm512_i64scatter_pd): Ditto.
33693 (_mm512_mask_i64scatter_pd): Ditto.
33694 (_mm512_i32scatter_epi32): Ditto.
33695 (_mm512_mask_i32scatter_epi32): Ditto.
33696 (_mm512_i32scatter_epi64): Ditto.
33697 (_mm512_mask_i32scatter_epi64): Ditto.
33698 (_mm512_i64scatter_epi32): Ditto.
33699 (_mm512_mask_i64scatter_epi32): Ditto.
33700 (_mm512_i64scatter_epi64): Ditto.
33701 (_mm512_mask_i64scatter_epi64): Ditto.
33702 * config/i386/avx512pfintrin.h
33703 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
33704 (_mm512_mask_prefetch_i32gather_ps): Ditto.
33705 (_mm512_mask_prefetch_i64gather_pd): Ditto.
33706 (_mm512_mask_prefetch_i64gather_ps): Ditto.
33707 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
33708 (_mm512_prefetch_i32scatter_ps): Ditto.
33709 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
33710 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
33711 (_mm512_prefetch_i64scatter_pd): Ditto.
33712 (_mm512_prefetch_i64scatter_ps): Ditto.
33713 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
33714 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
33715 * config/i386/avx512vlintrin.h
33716 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
33717 (_mm_mmask_i32gather_ps): Ditto.
33718 (_mm256_mmask_i32gather_pd): Ditto.
33719 (_mm_mmask_i32gather_pd): Ditto.
33720 (_mm256_mmask_i64gather_ps): Ditto.
33721 (_mm_mmask_i64gather_ps): Ditto.
33722 (_mm256_mmask_i64gather_pd): Ditto.
33723 (_mm_mmask_i64gather_pd): Ditto.
33724 (_mm256_mmask_i32gather_epi32): Ditto.
33725 (_mm_mmask_i32gather_epi32): Ditto.
33726 (_mm256_mmask_i32gather_epi64): Ditto.
33727 (_mm_mmask_i32gather_epi64): Ditto.
33728 (_mm256_mmask_i64gather_epi32): Ditto.
33729 (_mm_mmask_i64gather_epi32): Ditto.
33730 (_mm256_mmask_i64gather_epi64): Ditto.
33731 (_mm_mmask_i64gather_epi64): Ditto.
33732 (_mm256_i32scatter_ps): Change __addr type to void*.
33733 (_mm256_mask_i32scatter_ps): Ditto.
33734 (_mm_i32scatter_ps): Ditto.
33735 (_mm_mask_i32scatter_ps): Ditto.
33736 (_mm256_i32scatter_pd): Ditto.
33737 (_mm256_mask_i32scatter_pd): Ditto.
33738 (_mm_i32scatter_pd): Ditto.
33739 (_mm_mask_i32scatter_pd): Ditto.
33740 (_mm256_i64scatter_ps): Ditto.
33741 (_mm256_mask_i64scatter_ps): Ditto.
33742 (_mm_i64scatter_ps): Ditto.
33743 (_mm_mask_i64scatter_ps): Ditto.
33744 (_mm256_i64scatter_pd): Ditto.
33745 (_mm256_mask_i64scatter_pd): Ditto.
33746 (_mm_i64scatter_pd): Ditto.
33747 (_mm_mask_i64scatter_pd): Ditto.
33748 (_mm256_i32scatter_epi32): Ditto.
33749 (_mm256_mask_i32scatter_epi32): Ditto.
33750 (_mm_i32scatter_epi32): Ditto.
33751 (_mm_mask_i32scatter_epi32): Ditto.
33752 (_mm256_i32scatter_epi64): Ditto.
33753 (_mm256_mask_i32scatter_epi64): Ditto.
33754 (_mm_i32scatter_epi64): Ditto.
33755 (_mm_mask_i32scatter_epi64): Ditto.
33756 (_mm256_i64scatter_epi32): Ditto.
33757 (_mm256_mask_i64scatter_epi32): Ditto.
33758 (_mm_i64scatter_epi32): Ditto.
33759 (_mm_mask_i64scatter_epi32): Ditto.
33760 (_mm256_i64scatter_epi64): Ditto.
33761 (_mm256_mask_i64scatter_epi64): Ditto.
33762 (_mm_i64scatter_epi64): Ditto.
33763 (_mm_mask_i64scatter_epi64): Ditto.
33764 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
33765 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
33766 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
33767 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
33768 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
33769 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
33770 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
33771 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
33772 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
33773 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
33774 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
33775 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
33776 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
33777 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
33778 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
33779 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
33780 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
33781 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
33782 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
33783 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
33784 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
33785 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
33786 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
33787 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
33788 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
33789 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
33790 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
33791 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
33792 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
33793 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
33794 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
33795 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
33796 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
33797 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
33798 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
33799 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
33800 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
33801 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
33802 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
33803 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
33804 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
33805 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
33806 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
33807 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
33808 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
33809 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
33810 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
33811 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
33812 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
33813 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
33814 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
33815 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
33816 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
33817 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
33818 definitions accordingly.
33820 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
33821 Kuan-Lin Chen <kuanlinchentw@gmail.com>
33824 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
33827 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
33830 * ira-conflicts.c (ira_build_conflicts): Update total conflict
33831 hard regs for inner regno.
33833 2017-01-17 Martin Liska <mliska@suse.cz>
33836 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
33837 assumption and add comment.
33839 2017-01-17 Nathan Sidwell <nathan@acm.org>
33841 * ipa-visibility.c (localize_node): New function, broken out of ...
33842 (function_and_variable_visibility): ... here. Call it.
33844 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
33846 PR middle-end/77445
33847 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
33848 correctly set frequency of oudgoing edge.
33849 (duplicate_thread_path): Fix profile updating.
33851 2017-01-17 Jakub Jelinek <jakub@redhat.com>
33854 * configure.ac: Add GCC_BASE_VER.
33855 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
33856 version from BASE-VER file.
33857 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
33858 (gcc.o): Depend on $(BASEVER).
33859 * common.opt (dumpfullversion): New option.
33860 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
33861 * doc/invoke.texi: Document -dumpfullversion.
33862 * doc/install.texi: Document --with-gcc-major-version-only.
33863 * configure: Regenerated.
33865 2017-01-17 Richard Biener <rguenther@suse.de>
33867 PR tree-optimization/71433
33868 * tree-vrp.c (register_new_assert_for): Merge same asserts
33869 on all incoming edges.
33870 (process_assert_insertions_for): Handle insertions at the
33873 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
33875 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
33876 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
33878 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
33881 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
33884 2017-01-17 Alan Modra <amodra@gmail.com>
33887 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
33888 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
33889 symbolic stack limit when pic.
33891 2017-01-16 Martin Sebor <msebor@redhat.com>
33893 PR tree-optimization/78608
33894 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
33896 2017-01-16 Jeff Law <law@redhat.com>
33899 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
33900 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
33901 for several include directories that may be relative to sysroot.
33902 * config/i386/x-mingw32 (gplus_includedir): Define.
33903 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
33904 (native_system_includedir): Likewise.
33905 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
33906 override if TARGET_SYSTEM_ROOT is defined.
33907 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
33909 PR tree-optimization/79090
33910 PR tree-optimization/33562
33911 PR tree-optimization/61912
33912 PR tree-optimization/77485
33913 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
33914 and computed trims into the dump file.
33916 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
33918 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
33920 2017-01-16 Jakub Jelinek <jakub@redhat.com>
33923 * gimplify.c (gimplify_init_constructor): If want_value and
33924 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
33928 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
33929 sequence. Formatting fixes.
33930 (doloop_optimize): Formatting fixes.
33933 * gcc.c (debug_level_greater_than_spec_func): New function.
33934 (static_spec_functions): Add debug-level-gt spec function.
33935 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
33937 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33938 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
33939 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
33940 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
33941 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
33942 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
33944 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
33946 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
33947 QImode fixups to general and mask registers only.
33949 2017-01-16 Carl Love <cel@us.ibm.com>
33951 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
33952 for built-in functions
33953 vector signed char vec_nabs (vector signed char)
33954 vector signed short vec_nabs (vector signed short)
33955 vector signed int vec_nabs (vector signed int)
33956 vector signed long long vec_nabs (vector signed long long)
33957 vector float vec_nabs (vector float)
33958 vector double vec_nabs (vector double)
33959 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
33961 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
33962 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
33963 * doc/extend.texi: Update the documentation file for the new built-in
33966 2017-01-16 Martin Sebor <msebor@redhat.com>
33968 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
33971 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33973 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
33974 UNSPEC_VSX__XXSPLTD to require special splat handling.
33976 2017-01-16 David Malcolm <dmalcolm@redhat.com>
33979 * system.h: Poison strndup.
33981 2017-01-16 Alan Modra <amodra@gmail.com>
33984 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
33987 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
33989 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
33991 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
33993 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
33994 call recog here. Assert that INSN_CODE (insn) is non-negative.
33996 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
33999 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
34001 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
34002 in the currently scheduled RTL fragment.
34004 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
34006 PR rtl-optimization/78751
34007 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
34010 2017-01-14 Jeff Law <law@redhat.com>
34012 PR tree-optimization/79090
34013 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
34014 variable length stores.
34015 (compute_trims): Delete dead assignment to *trim_tail.
34016 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
34019 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
34021 PR rtl-optimization/78626
34022 PR rtl-optimization/78727
34023 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
34024 of a block, and split such blocks after everything else is finished.
34026 2017-01-14 Alan Modra <amodra@gmail.com>
34029 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
34030 target legitimate_combined_insn.
34031 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
34032 (rs6000_legitimate_combined_insn): New function.
34033 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
34035 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
34036 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
34037 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
34039 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
34041 * doc/frontends.texi (G++ and GCC): Remove references to Java.
34043 2017-01-13 Jeff Law <law@redhat.com>
34045 PR tree-optimization/33562
34046 PR tree-optimization/61912
34047 PR tree-optimization/77485
34048 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
34050 (delete_dead_assignment): Likewise.
34051 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
34052 statement to delete_dead_call and delete_dead_assignment.
34054 2017-01-13 David Malcolm <dmalcolm@redhat.com>
34057 * substring-locations.c (format_warning_va): Strengthen case 1 so
34058 that both endpoints of the substring must be within the format
34059 range for just the substring to be printed.
34061 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
34063 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
34064 * config/i386/i386.c (ix86_target_string): Add missing options
34065 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
34066 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
34067 flags_other and ix86_target_other to flags2_other. Display unknown
34069 (ix86_valid_target_attribute_inner_p): Add missing options and
34070 reorder options by implied ISAs, as in ix86_target_string.
34072 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34074 * hash-table.h (hash_table::too_empty_p): New function.
34075 (hash_table::expand): Use it.
34076 (hash_table::traverse): Likewise.
34077 (hash_table::empty_slot): Use sizeof (value_type) instead of
34078 sizeof (PTR) to convert bytes to elements. Shrink the table
34079 if the current size is excessive for the current number of
34082 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
34084 * ira-costs.c (record_reg_classes): Break from the inner loop
34085 early once alt_fail is known to be true. Update outer loop
34086 handling accordingly.
34088 2017-01-13 Jeff Law <law@redhat.com>
34090 * tree-ssa-dse.c (decrement_count): New function.
34091 (increment_start_addr, maybe_trim_memstar_call): Likewise.
34092 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
34093 when we know the partially dead statement is a mem* function.
34095 PR tree-optimization/61912
34096 PR tree-optimization/77485
34097 * tree-ssa-dse.c: Include expr.h.
34098 (maybe_trim_constructor_store): New function.
34099 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
34101 PR tree-optimization/33562
34102 PR tree-optimization/61912
34103 PR tree-optimization/77485
34104 * doc/invoke.texi: Document new dse-max-object-size param.
34105 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
34106 * tree-ssa-dse.c: Include params.h.
34107 (dse_store_status): New enum.
34108 (initialize_ao_ref_for_dse): New, partially extracted from
34110 (valid_ao_ref_for_dse, normalize_ref): New.
34111 (setup_live_bytes_from_ref, compute_trims): Likewise.
34112 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
34113 (maybe_trim_partially_dead_store): Likewise.
34114 (maybe_trim_complex_store): Likewise.
34115 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
34116 Track what bytes live from the original store. Return tri-state
34117 for dead, partially dead or live.
34118 (dse_dom_walker): Add constructor, destructor and new private members.
34119 (delete_dead_call, delete_dead_assignment): New extracted from
34121 (dse_optimize_stmt): Make a member of dse_dom_walker.
34122 Use initialize_ao_ref_for_dse.
34124 PR tree-optimization/33562
34125 PR tree-optimization/61912
34126 PR tree-optimization/77485
34127 * sbitmap.h (bitmap_count_bits): Prototype.
34128 (bitmap_clear_range, bitmap_set_range): Likewise.
34129 * sbitmap.c (bitmap_clear_range): New function.
34130 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
34132 2017-01-13 Martin Liska <mliska@suse.cz>
34135 * function.c (set_cfun): Add new argument force.
34136 * function.h (set_cfun): Likewise.
34137 * ipa-inline-transform.c (inline_call): Use the function when
34138 strict alising from is dropped for function we inline to.
34140 2017-01-13 Richard Biener <rguenther@suse.de>
34142 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
34143 for dumping GIMPLE INTEGER_CSTs.
34145 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
34147 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
34148 to 201112L since C++17.
34150 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
34153 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
34154 if -fsanitize=kernel-address is present.
34156 2017-01-13 Richard Biener <rguenther@suse.de>
34158 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
34159 as _Literal ( type ) number in case usual suffixes do not
34160 preserve all information.
34162 2017-01-13 Richard Biener <rguenther@suse.de>
34164 PR tree-optimization/77283
34165 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
34166 and ssa-iterators.h.
34167 (is_feasible_trace): Implement a cost model based on joiner
34170 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
34173 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
34174 char or short to __float128/_Float128 directly.
34176 2017-01-12 Martin Sebor <msebor@redhat.com>
34178 to -Wformat-overflow.
34179 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
34180 (min_bytes_remaining): Same.
34181 (get_string_length): Same.
34182 (format_string): Same.
34183 (format_directive): Same.
34185 (pass_sprintf_length::handle_gimple_call): Same.
34187 2017-01-12 Jakub Jelinek <jakub@redhat.com>
34189 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
34190 info.nowrite calls with no lhs that can't throw. Return bool
34191 whether gsi_remove has been called or not.
34192 (pass_sprintf_length::handle_gimple_call): Return bool whether
34193 try_substitute_return_value called gsi_remove. Formatting fix.
34194 (pass_sprintf_length::execute): Don't use gsi_remove if
34195 handle_gimple_call returned true.
34198 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
34199 be removed due to side-effects, don't remove following barrier nor
34200 turn the successor edge into fallthru edge.
34202 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
34205 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
34206 element-reversing loads and stores as not swappable.
34208 2017-01-12 Nathan Sidwell <nathan@acm.org>
34209 Nicolai Stange <nicstange@gmail.com>
34211 * combine.c (try_combine): Don't ignore result of overlap checking
34212 loop. Combine overlap & asm check into single loop.
34214 2017-01-12 Richard Biener <rguenther@suse.de>
34216 * tree-pretty-print.c (dump_generic_node): Provide -gimple
34217 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
34219 2017-01-12 Richard Biener <rguenther@suse.de>
34221 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
34222 and TS_TARGET_OPTION directly derive from TS_BASE.
34223 * tree-core.h (tree_optimization_option): Derive from tree_base.
34224 (tree_target_option): Likewise.
34226 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
34228 * config/i386/i386.c (memory_address_length): Increase len
34229 only when rip_relative_addr_p returns false.
34231 2017-01-11 Julia Koval <julia.koval@intel.com>
34233 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
34234 (OPTION_MASK_ISA_SGX_SET): New.
34235 (ix86_handle_option): Handle OPT_msgx.
34236 * config.gcc: Added sgxintrin.h.
34237 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
34238 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
34239 * config/i386/i386.c (ix86_target_string): Add -msgx.
34241 (ix86_option_override_internal): Handle new options.
34242 (ix86_valid_target_attribute_inner_p): Add sgx.
34243 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
34244 * config/i386/i386.opt: Add msgx.
34245 * config/i386/sgxintrin.h: New file.
34246 * config/i386/x86intrin.h: Add sgxintrin.h.
34248 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34251 * fold-const.c (maybe_nonzero_address): Return 1 for function
34253 (tree_single_nonzero_warnv_p): Don't handle function local objects
34257 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
34260 2017-01-11 David Malcolm <dmalcolm@redhat.com>
34263 * opts.c: Include "spellcheck.h"
34264 (struct string_fragment): New struct.
34265 (struct edit_distance_traits<const string_fragment &>): New
34267 (get_closest_sanitizer_option): New function.
34268 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
34270 2017-01-11 Jakub Jelinek <jakub@redhat.com>
34272 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
34274 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
34275 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
34276 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
34277 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
34278 for initial die_offset if dwarf_split_debug_info.
34279 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
34280 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
34282 (output_skeleton_debug_sections): Formatting fix. Use
34283 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
34284 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
34286 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
34288 * config/arm/cortex-a53.md: Add bypasses for
34289 cortex_a53_r2f_cvt.
34290 (cortex_a53_r2f): Only use for transfers.
34291 (cortex_a53_f2r): Likewise.
34292 (cortex_a53_r2f_cvt): Add reservation for conversions.
34293 (cortex_a53_f2r_cvt): Likewise.
34295 2017-01-11 Tamar Christina <tamar.christina@arm.com>
34297 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
34298 to all inlined functions, change static to extern.
34300 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
34303 * config/arm/arm.c (legitimize_pic_address): Handle reference to
34305 (arm_assemble_integer): Likewise.
34307 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
34309 * config.gcc: Use new awk script to check CPU, FPU and architecture
34310 parameters for --with-... options.
34311 * config/arm/parsecpu.awk: New file
34312 * config/arm/arm-cpus.in: New file.
34313 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
34315 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
34317 * config/arm/t-arm: Update dependency rules.
34318 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
34319 of processing .def files.
34320 * config/arm/genopt.sh: Deleted.
34321 * config/arm/gentune.sh: Deleted.
34322 * config/arm/arm-cores.def: Deleted.
34323 * config/arm/arm-arches.def: Deleted.
34324 * config/arm/arm-fpus.def: Deleted.
34325 * config/arm/arm-tune.md: Regenerated.
34326 * config/arm/arm-tables.opt: Regenerated.
34327 * config/arm/arm-cpu.h: New generated file.
34328 * config/arm/arm-cpu-data.h: New generated file.
34329 * config/arm/arm-cpu-cdata.h: New generated file.
34331 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
34334 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
34336 (input_varpool_node): Unpack dynamically_initialized bit.
34338 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
34340 PR rtl-optimization/79032
34341 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
34342 the alignment of the adjusted memory reference against that of MODE,
34343 instead of the alignment of the original memory reference.
34345 2017-01-11 Martin Jambor <mjambor@suse.cz>
34347 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
34349 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
34350 decorated functions.
34352 2017-01-11 Richard Biener <rguenther@suse.de>
34354 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
34355 set range/nonnull info for PHI results. Do not set it on
34356 stmts marked for removal.
34358 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
34360 * expr.c (store_field): In the bitfield case, fetch the return value
34361 from the registers before applying a single big-endian adjustment.
34362 Always do a final load for a BLKmode value not larger than a word.
34364 2017-01-10 David Malcolm <dmalcolm@redhat.com>
34367 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34368 that we correctly handle column numbers greater than
34369 LINE_MAP_MAX_COLUMN_NUMBER.
34371 2017-01-10 Martin Sebor <msebor@redhat.com>
34373 PR middle-end/78245
34374 * gimple-ssa-sprintf.c (get_destination_size): Call
34375 {init,fini}object_sizes.
34376 * tree-object-size.c (addr_object_size): Adjust.
34377 (pass_through_call): Adjust.
34378 (pass_object_sizes::execute): Adjust.
34379 * tree-object-size.h (fini_object_sizes): Declare.
34381 2017-01-10 Martin Sebor <msebor@redhat.com>
34383 PR tree-optimization/78775
34384 * builtins.c (get_size_range): Move...
34385 * calls.c: ...to here.
34386 (alloc_max_size): Accept zero argument.
34387 (operand_signed_p): Remove.
34388 (maybe_warn_alloc_args_overflow): Call get_size_range.
34389 * calls.h (get_size_range): Declare.
34391 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
34393 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
34394 from TI's devices.csv file as of September 2016.
34395 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
34397 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
34399 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
34400 * doc/invoke.texi: Likewise.
34401 * doc/md.texi: Likewise.
34402 * doc/objc.texi: Likewise.
34404 2017-01-10 Joshua Conner <joshconner@google.com>
34406 * config/arm/fuchsia-elf.h: New file.
34407 * config/fuchsia.h: New file.
34408 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
34409 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
34411 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
34413 2016-01-10 Richard Biener <rguenther@suse.de>
34415 PR tree-optimization/79034
34416 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
34417 Propagate out degenerate PHIs in the joiner.
34419 2017-01-10 Martin Liska <mliska@suse.cz>
34421 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
34422 (sort_congruence_classes_by_decl_uid): Likewise.
34423 (sort_congruence_class_groups_by_decl_uid): Likewise.
34424 (sem_item_optimizer::merge_classes): Sort class, groups in these
34425 classes and members in the groups by DECL_UID of declarations.
34426 This would make merge operations stable.
34428 2017-01-10 Martin Liska <mliska@suse.cz>
34430 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
34431 usage of m_classes_vec.
34432 (sem_item_optimizer::~sem_item_optimizer): Likewise.
34433 (sem_item_optimizer::get_group_by_hash): Likewise.
34434 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
34435 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
34436 (sem_item_optimizer::verify_classes): Likewise.
34437 (sem_item_optimizer::process_cong_reduction): Likewise.
34438 (sem_item_optimizer::dump_cong_classes): Likewise.
34439 (sem_item_optimizer::merge_classes): Likewise.
34440 * ipa-icf.h (congruence_class_hash): Rename from
34441 congruence_class_group_hash. Remove declaration of m_classes_vec.
34443 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
34445 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
34446 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
34447 * config.gcc: Add avx512vpopcntdqintrin.h.
34448 * config/i386/avx512vpopcntdqintrin.h: New.
34449 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
34450 * config/i386/i386-builtin-types.def: Add new types.
34451 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
34452 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
34453 __builtin_ia32_vpopcountq_v8di_mask): New.
34454 * config/i386/i386-c.c (ix86_target_macros_internal): Define
34455 __AVX512VPOPCNTDQ__.
34456 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
34457 (PTA_AVX512VPOPCNTDQ): Define.
34458 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
34459 TARGET_AVX512VPOPCNTDQ_P): Define.
34460 * config/i386/i386.opt: Add mavx512vpopcntdq.
34461 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
34462 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
34464 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34466 PR middle-end/77484
34467 * predict.def (PRED_CALL): Set to 67.
34469 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
34471 * expr.c (store_field): In the bitfield case, if the value comes from
34472 a function call and is of an aggregate type returned in registers, do
34473 not modify the field mode; extract the value in all cases if the mode
34474 is BLKmode and the size is not larger than a word.
34476 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
34479 * config/i386/cpuid.h: Fix undefined behavior.
34481 2017-01-04 Jeff Law <law@redhat.com>
34483 PR tree-optimization/79007
34484 PR tree-optimization/67955
34485 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
34486 conservative for pt.null when flag_non_call_exceptions is on.
34488 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34490 PR translation/79019
34491 PR translation/79020
34492 * params.def (PARAM_INLINE_MIN_SPEEDUP,
34493 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
34494 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
34496 * config/avr/avr.opt (maccumulate-args): Likewise.
34497 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
34498 * common.opt (freport-bug): Likewise.
34499 * cif-code.def (CIF_FINAL_ERROR): Likewise.
34500 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
34501 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
34502 translatable string.
34503 * config/i386/i386.c (function_value_32): Likewise.
34504 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
34505 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
34507 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
34508 * common/config/msp430/msp430-common.c (msp430_handle_option):
34510 * symtab.c (symtab_node::verify_base): Likewise.
34511 * opts.c (set_debug_level): Likewise.
34512 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
34513 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
34514 missing whitespace to translatable strings.
34515 * config/avr/avr.md (bswapsi2): Fix typo in comment.
34516 * config/sh/superh.h: Likewise.
34517 * config/i386/xopintrin.h: Likewise.
34518 * config/i386/znver1.md: Likewise.
34519 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
34520 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
34521 * double-int.h (struct double_int): Likewise.
34522 * double-int.c (div_and_round_double): Likewise.
34523 * wide-int.cc: Likewise.
34524 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
34525 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
34526 * cfgcleanup.c (crossjumps_occured): Renamed to ...
34527 (crossjumps_occurred): ... this.
34528 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
34531 PR tree-optimization/78899
34532 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
34533 returning bool return struct loop *, NULL for failure and the new
34535 (versionable_outer_loop_p): Don't version outer loop if it has
34536 dont_vectorized bit set.
34537 (tree_if_conversion): When versioning outer loop, ensure
34538 tree_if_conversion is performed also on the inner loop of the
34539 non-vectorizable outer loop copy.
34540 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
34541 LOOP_VECTORIZED in inner loop of the scalar outer loop and
34542 prevent vectorization of it.
34543 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
34544 the outer loop vectorization of the non-scalar version is attempted
34545 before vectorization of the inner loop in scalar version. If
34546 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
34547 vectorization of its inner loop.
34548 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
34549 has 2 inner loops, rename also on edges from bb whose single pred
34550 is outer_loop->header. Fix typo in function comment.
34552 2017-01-09 Martin Sebor <msebor@redhat.com>
34555 * asan.c (asan_emit_stack_protection): Increase local buffer size
34556 to avoid snprintf truncation warning.
34558 2017-01-09 Andrew Pinski <apinski@cavium.com>
34560 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
34561 to reference thunderx2t99 for the tuning structure
34562 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
34564 (thunderx2t99_extra_costs): This.
34565 * config/aarch64/aarch64-tune.md: Regenerate.
34566 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
34567 (vulcan_addrcost_table): This.
34568 (vulcan_regmove_cost): Rename to ...
34569 (thunderx2t99_regmove_cost): This.
34570 (vulcan_vector_cost): Rename to ...
34571 (thunderx2t99_vector_cost): this.
34572 (vulcan_branch_cost): Rename to ...
34573 (thunderx2t99_branch_cost): This.
34574 (vulcan_tunings): Rename to ...
34575 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
34576 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
34578 2017-01-09 Martin Jambor <mjambor@suse.cz>
34582 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
34583 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
34584 (propagate_vr_accross_jump_function): Use the above function for all
34585 value range computations for pass-through jump functions and type
34586 converasion from explicit value range values.
34587 (ipcp_propagate_stage): Do not attempt to deduce types of formal
34588 parameters from TYPE_ARG_TYPES.
34589 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
34590 (ipa_write_node_info): Stream type of the actual argument.
34591 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
34593 2017-01-09 Martin Liska <mliska@suse.cz>
34596 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
34597 (lookup_compiler): Do not show error message with have_E.
34599 2017-01-09 Jakub Jelinek <jakub@redhat.com>
34601 PR tree-optimization/78938
34602 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
34603 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
34604 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
34605 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
34608 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
34610 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
34613 2017-01-09 Richard Biener <rguenther@suse.de>
34615 PR tree-optimization/78997
34616 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
34617 name condition properly.
34619 2017-01-09 Richard Biener <rguenther@suse.de>
34622 * dwarf2out.c (is_cxx): New overload with context.
34623 (is_naming_typedef_decl): Use it.
34625 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34627 * invoke.texi (Option Summary): Correct spacing in option lists
34628 and add line breaks to fix over-long lines.
34630 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34632 PR middle-end/17660
34634 * extend.texi (Common Variable Attributes): Add xref to GCC
34635 Internals manual to explain mode attribute keywords.
34637 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
34640 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
34641 and Preprocessor Options.
34642 (Options for Linking): Document -pthread here....
34643 (RS/6000 and PowerPC Options): ...not here.
34644 (Solaris 2 Options): ...or here.
34645 * doc/cppopts.texi: Document -pthread.
34647 2017-01-08 Martin Sebor <msebor@redhat.com>
34649 PR middle-end/77708
34650 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
34651 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
34652 New member functions.
34653 (format_directive): Used them.
34655 (pass_sprintf_length::handle_gimple_call): Same.
34656 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
34657 to avoid truncation for any argument.
34658 (extract_affine_mul): Same.
34659 * tree.c (get_file_function_name): Same.
34661 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34663 PR middle-end/77484
34664 * predict.def (PRED_INDIR_CALL): Set to 86.
34666 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34668 PR preprocessor/54124
34669 * doc/cppopts.texi: Reformat -d subtable to list the full name
34670 of the options. Add cross-reference to the docs for the general
34671 compiler -d options.
34672 * doc/invoke.texi (Developer Options): Add cross-reference to the
34673 preprocessor-specific -d option documentation.
34675 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34677 PR preprocessor/13498
34678 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
34679 redudant material, and reflect new command-line options.
34680 (System Headers): Likewise.
34682 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
34684 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
34685 -isystem, and -idirafter. Copy-edit.
34686 * doc/cppopts.texi: Copy-edit. Remove contradiction about
34687 default for -ftrack-macro-expansion. Delete obsolete and
34688 badly-formatted implementation details about -fdebug-cpp output.
34689 * doc/cppwarnopts.texi: Copy-edit.
34691 2017-01-07 David Malcolm <dmalcolm@redhat.com>
34694 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
34695 that the transition from a max line width >= 1<<10 to narrower
34696 lines works correctly.
34698 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
34700 * doc/options.texi (PerFunction): New.
34701 * opt-functions.awk (switch_flags): Map both Optimization and
34702 PerFunction to CL_OPTIMIZATION.
34703 * opth-gen.awk: Test for PerFunction flag along with
34705 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
34706 it only when the latter is present. Skip those that don't in
34707 the hash function generator.
34708 * common.opt (fvar-tracking): Mark as PerFunction instead of
34710 (fvar-tracking-assignments): Likewise.
34711 (fvar-tracking-assignments-toggle): Likewise.
34712 (fvar-tracking-uninit): Likewise.
34714 2017-01-07 Jakub Jelinek <jakub@redhat.com>
34716 PR translation/79018
34717 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
34720 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
34723 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
34724 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
34725 TARGET_LONG_JUMP_TABLE_OFFSETS.
34726 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
34727 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
34728 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
34729 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
34730 * config/m68k/m68k.md (tablejump expander): Likewise.
34731 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
34732 TARGET_LONG_JUMP_TABLE_OFFSETS.
34733 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
34734 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
34736 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
34737 David Holsgrove <david.holsgrove@xilinx.com>
34739 * common/config/microblaze/microblaze-common.c
34740 (TARGET_EXCEPT_UNWIND_INFO): Remove.
34741 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
34743 * config/microblaze/microblaze.c (microblaze_must_save_register)
34744 (microblaze_expand_epilogue, microblaze_return_addr): Handle
34746 (microblaze_eh_return): New function.
34747 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
34748 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
34749 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
34750 * config/microblaze/microblaze.md (eh_return): New pattern.
34752 2017-01-06 Jakub Jelinek <jakub@redhat.com>
34754 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
34755 GCC_DIAGNOSTIC_STRINGIFY): Define.
34757 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
34759 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34761 * config/arm/arm.md (<mcrr>): New.
34763 * config/arm/arm.c (arm_arch5te): New.
34764 (arm_option_override): Set arm_arch5te.
34765 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
34767 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
34768 (arm_mcrr_qualifiers): ... this. New.
34769 (MRRC_QUALIFIERS): Define to...
34770 (arm_mrrc_qualifiers): ... this. New.
34771 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
34773 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
34774 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
34775 (MRRCI, mrrc, MRRC): New.
34776 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
34777 VUNSPEC_MRRC2): New.
34779 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34781 * config/arm/arm.md (<mcr>): New.
34783 * config/arm/arm.c (arm_coproc_builtin_available): Add
34784 support for mcr, mrc, mcr2 and mrc2.
34785 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
34786 (arm_mcr_qualifiers): ... this. New.
34787 (MRC_QUALIFIERS): Define to ...
34788 (arm_mrc_qualifiers): ... this. New.
34789 (MCR_QUALIFIERS): Define to ...
34790 (arm_mcr_qualifiers): ... this. New.
34791 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
34793 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
34794 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
34795 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
34796 VUNSPEC_MRC2): New.
34798 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34800 * config/arm/arm.md (*ldc): New.
34804 * config/arm/arm.c (arm_coproc_builtin_available): Add
34805 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
34806 (arm_coproc_ldc_stc_legitimate_address): New.
34807 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
34808 'qualifier_const_pointer'.
34809 (LDC_QUALIFIERS): Define to...
34810 (arm_ldc_qualifiers): ... this. New.
34811 (STC_QUALIFIERS): Define to...
34812 (arm_stc_qualifiers): ... this. New.
34813 * config/arm/arm-protos.h
34814 (arm_coproc_ldc_stc_legitimate_address): New.
34815 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
34816 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
34817 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
34818 stc2, stcl, stc2l): New.
34819 * config/arm/constraints.md (Uz): New.
34820 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
34821 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
34822 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
34823 VUNSPEC_STC2L): New.
34825 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34827 * config/arm/arm.md (<cdp>): New.
34828 * config/arm/arm.c (neon_const_bounds): Rename this ...
34829 (arm_const_bounds): ... this.
34830 (arm_coproc_builtin_available): New.
34831 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
34832 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
34833 (CDP_QUALIFIERS): Define to...
34834 (arm_cdp_qualifiers): ... this. New.
34836 (arm_expand_builtin_args): Add case for 6 arguments.
34837 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
34838 (arm_const_bounds): ... this.
34839 (arm_coproc_builtin_available): New.
34840 * config/arm/arm_acle.h (__arm_cdp): New.
34842 * config/arm/arm_acle_builtins.def (cdp): New.
34844 * config/arm/iterators.md (CDPI,CDP,cdp): New.
34845 * config/arm/neon.md: Rename all 'neon_const_bounds' to
34846 'arm_const_bounds'.
34847 * config/arm/types.md (coproc): New.
34848 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
34849 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
34850 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
34851 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
34853 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34855 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
34856 (UBINOP_QUALIFIERS): New.
34858 (acle_builtin_data): New. Change comment.
34859 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
34860 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
34861 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
34862 arm_acle_builtins.def.
34863 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
34864 (arm_init_acle_builtins): New.
34865 (CRC32_BUILTIN): Remove.
34866 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
34867 crc32cb, crc32ch and crc32cw.
34868 (arm_init_crc32_builtins): Remove.
34869 (arm_init_builtins): Use arm_init_acle_builtins rather
34870 than arm_init_crc32_builtins.
34871 (arm_expand_acle_builtin): New.
34872 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
34873 * config/arm/arm_acle_builtins.def: New.
34875 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
34877 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
34878 (arm_builtin_datum): ... this.
34879 (arm_init_neon_builtin): Rename to ...
34880 (arm_init_builtin): ... this. Add a new parameters PREFIX
34881 and USE_SIG_IN_NAME.
34882 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
34883 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
34884 'arm_builtin_datum'.
34885 (arm_init_vfp_builtins): Likewise.
34886 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
34887 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
34888 (arm_expand_neon_args): Rename to ...
34889 (arm_expand_builtin_args): ... this. Rename builtin_arg
34890 enum values and differentiate between ARG_BUILTIN_MEMORY
34891 and ARG_BUILTIN_NEON_MEMORY.
34892 (arm_expand_neon_builtin_1): Rename to ...
34893 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
34894 values, arm_expand_builtin_args and add bool parameter NEON.
34895 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
34896 (arm_expand_vfp_builtin): Likewise.
34897 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
34899 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34901 PR middle-end/77484
34902 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
34903 * predict.c (tree_estimate_probability_bb): Reverse direction of
34904 polymorphic call predictor.
34906 2017-01-06 David Malcolm <dmalcolm@redhat.com>
34908 * passes.c (execute_one_pass): Split out pass-skipping logic into...
34909 (determine_pass_name_match): ...this new function and...
34910 (should_skip_pass_p): ...this new function.
34912 2017-01-06 Nathan Sidwell <nathan@acm.org>
34914 * ipa-visibility.c (function_and_variable_visibility): Reformat
34915 comments and long lines. Remove extrneous if.
34916 * symtab.c (symtab_node::make_decl_local): Fix code format.
34917 (symtab_node::set_section_for_node): Fix comment typo.
34919 2017-01-06 Martin Liska <mliska@suse.cz>
34922 * lra-constraints.c: Rename invariant to lra_invariant.
34923 * predict.c (set_even_probabilities): Initialize e to NULL.
34925 2017-01-05 Martin Sebor <msebor@redhat.com>
34927 PR tree-optimization/78910
34928 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
34929 (format_integer): Correct off-by-one error in the handling
34930 of precision with negative numbers in signed conversions..
34932 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
34934 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
34936 2017-01-05 Jakub Jelinek <jakub@redhat.com>
34938 PR tree-optimization/71016
34939 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
34940 factor_out_conditional_conversion. Formatting fix.
34941 (factor_out_conditional_conversion): Add cond_stmt argument.
34942 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
34943 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
34946 2017-01-05 David Malcolm <dmalcolm@redhat.com>
34948 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
34949 read-rtl-function.o, and selftest-rtl.o.
34950 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
34951 (selftest::aarch64_test_loading_full_dump): New function.
34952 (selftest::aarch64_run_selftests): New function.
34953 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
34954 selftest::aarch64_run_selftests.
34955 * config/i386/i386.c
34956 (selftest::ix86_test_loading_dump_fragment_1): New function.
34957 (selftest::ix86_test_loading_call_insn): New function.
34958 (selftest::ix86_test_loading_full_dump): New function.
34959 (selftest::ix86_test_loading_unspec): New function.
34960 (selftest::ix86_run_selftests): Call the new functions.
34961 * emit-rtl.c (maybe_set_max_label_num): New function.
34962 * emit-rtl.h (maybe_set_max_label_num): New decl.
34963 * function.c (instantiate_decls): Guard call to
34964 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
34965 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
34967 * gensupport.c (gen_reader::gen_reader): Pass "false"
34968 for new "compact" param of rtx_reader.
34969 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
34970 rather than an empty string for NULL strings.
34971 * read-md.c: Potentially include config.h rather than bconfig.h.
34972 Wrap include of errors.h with #ifdef GENERATOR_FILE.
34973 (have_error): New global, copied from errors.c.
34974 (md_reader::read_name): Rename to...
34975 (md_reader::read_name_1): ...this, adding "out_loc" param,
34976 and converting "missing name or number" to returning false, rather
34978 (md_reader::read_name): Reimplement in terms of read_name_1.
34979 (md_reader::read_name_or_nil): New function.
34980 (md_reader::read_string): Handle "(nil)" by returning NULL.
34981 (md_reader::md_reader): Add new param "compact".
34982 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
34983 (md_reader::read_file): New method.
34984 * read-md.h (md_reader::md_reader): Add new param "compact".
34985 (md_reader::read_file): New method.
34986 (md_reader::is_compact): New accessor.
34987 (md_reader::read_name): Convert return type from void to file_location.
34988 (md_reader::read_name_or_nil): New decl.
34989 (md_reader::read_name_1): New decl.
34990 (md_reader::m_compact): New field.
34991 (noop_reader::noop_reader): Pass "false" for new "compact" param
34993 (rtx_reader::rtx_reader): Add new "compact" param.
34994 (rtx_reader::read_rtx_operand): Make virtual and convert return
34995 type from void to rtx.
34996 (rtx_reader::read_until): New decl.
34997 (rtx_reader::handle_any_trailing_information): New virtual function.
34998 (rtx_reader::postprocess): New virtual function.
34999 (rtx_reader::finalize_string): New virtual function.
35000 (rtx_reader::m_in_call_function_usage): New field.
35001 (rtx_reader::m_reuse_rtx_by_id): New field.
35002 * read-rtl-function.c: New file.
35003 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
35004 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
35005 (selftest::verify_three_block_rtl_cfg): New decl.
35006 * read-rtl-function.h: New file.
35007 * read-rtl.c: Potentially include config.h rather than bconfig.h.
35008 For host, include function.h, memmodel.h, and emit-rtl.h.
35009 (one_time_initialization): New function.
35010 (struct compact_insn_name): New struct.
35011 (compact_insn_names): New array.
35012 (find_code): Handle insn codes in compact dumps.
35013 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
35014 (bind_subst_iter_and_attr): Likewise.
35015 (add_condition_to_string): Likewise.
35016 (add_condition_to_rtx): Likewise.
35017 (apply_attribute_uses): Likewise.
35018 (add_current_iterators): Likewise.
35019 (apply_iterators): Likewise.
35020 (initialize_iterators): Guard usage of apply_subst_iterator with
35021 #ifdef GENERATOR_FILE.
35022 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
35023 (md_reader::read_mapping): Likewise.
35024 (add_define_attr_for_define_subst): Likewise.
35025 (add_define_subst_attr): Likewise.
35026 (read_subst_mapping): Likewise.
35027 (check_code_iterator): Likewise.
35028 (rtx_reader::read_rtx): Likewise. Move one-time initialization
35030 (one_time_initialization): New function.
35031 (rtx_reader::read_until): New method.
35032 (read_flags): New function.
35033 (parse_reg_note_name): New function.
35034 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
35035 Handle reuse_rtx ids.
35036 Wrap iterator lookup within #ifdef GENERATOR_FILE.
35037 Add parsing support for RTL dumps, mirroring the special-cases in
35038 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
35039 values, and calling handle_any_trailing_information.
35040 (rtx_reader::read_rtx_operand): Convert return type from void
35041 to rtx, returning return_rtx. Handle case 'e'. Call
35042 finalize_string on XSTR and XTMPL fields.
35043 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
35044 "(nil)" values were omitted. Call the postprocess vfunc on the
35046 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
35047 class ctor. Initialize m_in_call_function_usage. Call
35048 one_time_initialization.
35049 * rtl-tests.c (selftest::test_uncond_jump): Call
35050 set_new_first_and_last_insn.
35051 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
35052 * selftest-rtl.c: New file.
35053 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
35054 (selftest::get_insn_by_uid): New decl.
35055 * selftest-run-tests.c (selftest::run_tests): Call
35056 read_rtl_function_c_tests.
35057 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
35058 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
35061 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
35063 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
35064 operands in a special way. Assert that pos+len <= mode precision.
35066 2017-01-05 Jakub Jelinek <jakub@redhat.com>
35068 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
35069 3 argument Alias with unlimited for the negative form.
35070 (fno-vect-cost-model): Removed.
35072 2017-01-05 Martin Liska <mliska@suse.cz>
35074 * hsa-gen.c (gen_hsa_divmod): New function.
35075 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
35077 2017-01-05 Martin Liska <mliska@suse.cz>
35080 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
35083 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35085 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
35086 small constant length operands.
35088 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
35090 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
35091 between loop iterations.
35093 2017-01-05 Martin Liska <mliska@suse.cz>
35096 * gimplify.c (gimplify_decl_expr): Compare to
35097 asan_poisoned_variables instread of checking flags.
35098 (gimplify_target_expr): Likewise.
35099 (gimplify_expr): Likewise.
35100 (gimplify_function_tree): Conditionally initialize
35101 asan_poisoned_variables.
35103 2017-01-04 Jeff Law <law@redhat.com>
35105 PR tree-optimizatin/78812
35106 * rtl.h (contains_mem_rtx_p): Prototype.
35107 * ifcvt.c (containts_mem_rtx_p): Move from here to...
35108 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
35109 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
35110 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
35111 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
35113 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35115 * input.c (assert_char_at_range): Default-initialize actual_range.
35117 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35119 * df-scan.c (df_ref_create_structure): Make regno unsigned,
35120 to match the caller.
35122 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35124 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
35125 insns after final jump in test to emit dummy move.
35127 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35129 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
35130 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
35132 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
35134 * multiple_target.c (create_dispatcher_calls): Init e_next.
35135 * tree-ssa-loop-split.c (split_loop): Init border.
35136 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
35139 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
35144 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
35145 (altivec_register_operand): Do not return true if the operand
35146 contains a SUBREG mixing SImode and SFmode.
35147 (vsx_register_operand): Likewise.
35148 (vsx_reg_sfsubreg_ok): New predicate.
35149 (vfloat_operand): Do not return true if the operand contains a
35150 SUBREG mixing SImode and SFmode.
35151 (vint_operand): Likewise.
35152 (vlogical_operand): Likewise.
35153 (gpc_reg_operand): Likewise.
35154 (int_reg_operand): Likewise.
35155 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
35156 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
35157 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
35159 (rs6000_emit_move_si_sf_subreg): New helper function.
35160 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
35161 fixup SUBREGs involving SImode and SFmode.
35162 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
35163 numbers for the new peephole2 optimization.
35164 (peephole2 for SFmode unions): New peephole2 to optimize cases in
35165 the GLIBC math library that do AND/IOR/XOR operations on single
35166 precision floating point.
35167 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
35168 target macros to say whether we need to avoid SUBREGs mixing
35170 (TARGET_ALLOW_SF_SUBREG): Likewise.
35171 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
35172 (UNSPEC_SI_FROM_SF): Likewise.
35173 (iorxor): Change spacing.
35174 (and_ior_xor): New iterator for AND, IOR, and XOR.
35175 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
35176 (movdi_from_sf_zero_ext): Likewise.
35177 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
35178 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
35179 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
35180 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
35181 (fms<mode>4): Likewise.
35182 (fnma<mode>4): Likewise.
35183 (fnms<mode>4): Likewise.
35184 (nfma<mode>4): Likewise.
35185 (nfms<mode>4): Likewise.
35187 2017-01-04 Marek Polacek <polacek@redhat.com>
35190 * doc/invoke.texi: Document -Wpointer-compare.
35192 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35194 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
35197 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
35198 descriptions for -gdwarf-5 and emit them as uleb128 instead of
35201 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
35204 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
35205 documentation of the powerpc_popcntb_ok attribute.
35206 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
35207 code to issue warning messages if a requested CPU configuration is
35208 not supported by the binary (assembler and loader) toolchain.
35209 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
35210 made to define a built-in function that has been disabled.
35211 (paired_init_builtins): Add assertion to prevent ICE if attempt is
35212 made to define a built-in function that has been disabled.
35213 (altivec_init_builtins): Add comment explaining why definition
35214 of the DST built-in functions is not preceded by an assertion
35215 check. Add assertions to prevent ICE if attempts are made to
35216 define an altivec predicate or an abs* built-in function that has
35218 (htm_init_builtins): Add comment explaining why definition of the
35219 htm built-in functions is not preceded by an assertion check.
35221 2017-01-04 Jeff Law <law@redhat.com>
35223 PR tree-optimizatin/67955
35224 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
35225 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
35226 the points-to solution does not include pt_null. Use DECL_PT_UID
35229 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
35231 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
35232 Use gen_int_mode instead of gen_lopwart for const_int operands.
35234 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35236 PR tree-optimization/71563
35237 * match.pd: Simplify X << Y into X if Y is known to be 0 or
35238 out of range value - has low bits known to be zero.
35240 2017-01-04 Alan Modra <amodra@gmail.com>
35242 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
35243 * configure: Regenerate.
35244 * config.in: Regenerate.
35246 2017-01-04 Jakub Jelinek <jakub@redhat.com>
35249 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
35250 a substring of the message, but strcmp with the whole message. Ifdef
35251 ENABLE_NLS, translate the message first using dgettext.
35253 2017-01-03 Jeff Law <law@redhat.com>
35255 PR tree-optimizatin/78856
35256 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
35257 (mark_threaded_blocks): Remove code to truncate thread paths that
35258 cross multiple loop headers. Instead invalidate the cached loop
35259 iteration information and handle case of a thread path walking
35260 into an irreducible region.
35262 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
35265 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
35266 assertions. Add support for doing the signbit if the IEEE 128-bit
35267 floating point value is in a GPR.
35268 * config/rs6000/rs6000.md (Fsignbit): Delete.
35269 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
35270 Update the length attribute if the value is in a GPR.
35271 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
35272 the sign or zero extension instruction, since the value is always 0/1.
35273 (signbit<mode>2_dm2): Delete using <Fsignbit>.
35276 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
35277 extracting SImode to a GPR register so that we can generate a
35278 store, limit the vector to be in a traditional Altivec register
35279 for the vextuwrx instruction.
35281 2017-01-03 Ian Lance Taylor <iant@google.com>
35283 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
35285 2017-01-03 Martin Sebor <msebor@redhat.com>
35287 PR tree-optimization/78696
35288 * gimple-ssa-sprintf.c (format_floating): Correct handling of
35289 precision. Use MPFR for %f for greater fidelity. Correct handling
35291 (pass_sprintf_length::compute_format_length): Set width and precision
35292 specified by asrerisk to void_node for vararg functions.
35293 (try_substitute_return_value): Adjust dump output.
35295 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
35297 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
35299 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
35301 * doc/invoke.texi (SPARC options): Document -mlra as the default.
35302 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
35303 -mlra/-mno-lra was passed to the compiler.
35305 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
35307 PR rtl-optimization/65618
35308 * emit-rtl.c (try_split): Move initialization of "before" and
35309 "after" to just before the call to emit_insn_after_setloc.
35311 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
35313 * doc/md.texi (Standard Names): Remove reference to Java frontend.
35315 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
35317 * dwarf2out.c (gen_enumeration_type_die): When
35318 -gno-strict-dwarf, add a DW_AT_encoding attribute.
35320 2017-01-03 Jakub Jelinek <jakub@redhat.com>
35322 PR tree-optimization/78965
35323 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
35324 Change first argument from const call_info & to call_info &. For %n
35325 set info.nowrite to false.
35327 PR middle-end/78901
35328 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
35329 possibly throwing calls.
35331 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
35332 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
35333 and fns handling, rather than in a separate case SSA_NAME.
35335 2017-01-02 Jeff Law <law@redhat.com>
35337 * config/darwin-driver.c (darwin_driver_init): Const-correctness
35338 fixes for first_period and second_period variables.
35340 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
35343 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
35344 (*insvqi_1): New insn pattern.
35345 (*insvqi_1_mem_rex64): Ditto.
35346 (*insvqi_2): Ditto.
35347 (*insvqi_3): Rename from *insvqi.
35349 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
35351 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
35353 * doc/cfg.texi (Edges): Remove reference to Java.
35354 (Maintaining the CFG): Ditto.
35356 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35358 PR middle-end/77674
35359 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
35360 transparent aliases.
35362 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
35364 PR middle-end/77484
35365 * predict.def (PRED_CALL): Update hitrate.
35366 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
35367 * predict.c (tree_estimate_probability_bb): Split CALL predictor
35368 into direct/indirect/polymorphic variants.
35370 2017-01-01 Jakub Jelinek <jakub@redhat.com>
35372 Update copyright years.
35374 * gcc.c (process_command): Update copyright notice dates.
35375 * gcov-dump.c (print_version): Ditto.
35376 * gcov.c (print_version): Ditto.
35377 * gcov-tool.c (print_version): Ditto.
35378 * gengtype.c (create_file): Ditto.
35379 * doc/cpp.texi: Bump @copying's copyright year.
35380 * doc/cppinternals.texi: Ditto.
35381 * doc/gcc.texi: Ditto.
35382 * doc/gccint.texi: Ditto.
35383 * doc/gcov.texi: Ditto.
35384 * doc/install.texi: Ditto.
35385 * doc/invoke.texi: Ditto.
35387 Copyright (C) 2017 Free Software Foundation, Inc.
35389 Copying and distribution of this file, with or without modification,
35390 are permitted in any medium without royalty provided the copyright
35391 notice and this notice are preserved.